1
00:00:00,000 -->
00:00:03,440
Bret (2): were you sorta seeing where
this all could take us in terms of
2
00:00:03,530 -->
00:00:05,040
agents helping us in infrastructure
3
00:00:05,090 -->
00:00:09,540
Doron: we wanted to reduce the
cognitive load on engineers not
4
00:00:09,550 -->
00:00:13,629
having to stitch together Kubernetes,
Istio, Grafana, Prometheus, Vault,
5
00:00:13,630 -->
00:00:15,990
and all the CNCF, solutions out there.
6
00:00:16,090 -->
00:00:23,466
And we created an API that gives you
unified access to any cloud, any service,
7
00:00:23,566 -->
00:00:28,614
and we find that AI, if you just tell it,
"Create me the infrastructure," it's gonna
8
00:00:28,624 -->
00:00:34,104
duplicate a lot of resources: and it's
gonna be expensive, non-deterministic, and
9
00:00:34,104 -->
00:00:36,544
it's gonna be non-secure, non-compliant.
10
00:00:36,644 -->
00:00:40,394
And where is the logging, the metrics,
the tracing, the secrets management,
11
00:00:40,394 -->
00:00:42,484
the service discovery, the TLS, the DNS?
12
00:00:42,484 -->
00:00:44,584
And the list goes on and on.
13
00:00:44,684 -->
00:00:50,097
So what AI needs, just like an engineer
needs, is a reduced cognitive load.
14
00:00:50,117 -->
00:00:53,437
Give it a deterministic API
that says, "Hey, take my
15
00:00:53,437 -->
00:00:55,217
desired state and make it so.
16
00:00:55,317 -->
00:01:00,236
And by the way, make it so on any
cloud." It makes AI a lot more effective.
17
00:01:00,336 -->
00:01:05,026
we find AI in a perfect, intersection
because we can now make it do
18
00:01:05,076 -->
00:01:09,376
tricks that is otherwise very, very
expensive or even u- unattainable
19
00:01:16,105 -->
00:01:18,475
Bret (2): Welcome to another
episode of DevOps & Docker Talk.
20
00:01:18,515 -->
00:01:24,255
I'm your host, Bret, and in this one,
we go deep into the universal control
21
00:01:24,255 -->
00:01:29,595
plane for the cloud and the search for
the singular management system, which
22
00:01:29,695 -->
00:01:34,435
we have been doing my entire 30-year
career, been trying to find that perfect
23
00:01:34,435 -->
00:01:40,615
sweet spot of a tool that can manage
all of my infrastructure from one API.
24
00:01:40,895 -->
00:01:45,385
And what better person to have on
to talk about this than the CEO and
25
00:01:45,385 -->
00:01:50,605
co-founder of Control Plane, the company
building this unified API layer for
26
00:01:50,605 -->
00:01:52,165
controlling all your infrastructure.
27
00:01:52,465 -->
00:01:57,735
I'm having Doron Grinstein on the show,
and we go deep into what they're building,
28
00:01:57,745 -->
00:02:02,675
And he has a very similar story to me,
just trying to use the infrastructure
29
00:02:02,775 -->
00:02:07,275
that we had at the time and the tooling
we had at the time, and I have managed
30
00:02:07,285 -->
00:02:11,355
infrastructure for so long that I have
forgotten about all the tools that we've
31
00:02:11,365 -->
00:02:15,485
tried to use over the years to unify
everything, and we never quite get there.
32
00:02:15,485 -->
00:02:19,965
We get 60%, maybe if you're really
lucky, 80%, but then there's always
33
00:02:19,995 -->
00:02:22,075
that other thing that doesn't quite fit.
34
00:02:22,175 -->
00:02:25,195
And it's exciting to see a company
that's brave enough to take this
35
00:02:25,195 -->
00:02:26,725
on, because it's no small task.
36
00:02:26,735 -->
00:02:31,245
It's no small achievement to be able
to say you have a universal API control
37
00:02:31,245 -->
00:02:36,315
plane that you can access from the
browser or from Terraform or from Pulumi
38
00:02:36,325 -->
00:02:43,035
or from your agent harness, and it feels
like that's what they're achieving.
39
00:02:43,065 -->
00:02:45,005
And we walked through a bunch of demos.
40
00:02:45,005 -->
00:02:47,615
In fact, if you're interested in the
demos, go check out the link in the
41
00:02:47,665 -->
00:02:51,205
show notes to go to the live stream,
'cause we demoed for about an hour
42
00:02:51,215 -->
00:02:54,475
after the conversation, and, we
didn't put those into this podcast.
43
00:02:54,495 -->
00:02:56,865
So if you wanna check out all
those demos, go check the link.
44
00:02:57,215 -->
00:03:00,555
But we talk about everything from
how they onboard customers, what are
45
00:03:00,555 -->
00:03:03,955
their customers looking like in terms
of their infrastructure, and how are
46
00:03:03,955 -->
00:03:05,545
they managing it once they get there.
47
00:03:05,555 -->
00:03:07,155
Surprise, not a lot's changed.
48
00:03:07,165 -->
00:03:10,035
They still use Terraform or OpenTofu.
49
00:03:10,035 -->
00:03:13,865
All the things that they were using
before are compatible with their APIs.
50
00:03:13,875 -->
00:03:18,895
But they also have MCP and CLI tools
to manage this, and once you get into
51
00:03:18,895 -->
00:03:22,355
their interface, you realize that part
of what they can do, because they're
52
00:03:22,355 -->
00:03:26,785
agnostic and they want to operationalize
all of this across multi clouds, is
53
00:03:26,825 -->
00:03:30,265
they're setting up the Kubernetes
distributions across the clouds for you.
54
00:03:30,505 -->
00:03:35,585
They're making the decisions around how
the encrypted tunnels are gonna happen
55
00:03:35,595 -->
00:03:39,085
between the different providers, it
feels kind of like what Docker felt at
56
00:03:39,145 -->
00:03:42,765
first, where they had this old saying
at Docker of, "Batteries included, but
57
00:03:42,765 -->
00:03:46,455
swappable" and that is something that
I feel like I could call control plane.
58
00:03:46,455 -->
00:03:49,135
It feels like it's very easy out
of the box, but they hide a little
59
00:03:49,135 -->
00:03:52,155
bit of that complexity with their
opinionated defaults, which I like
60
00:03:52,155 -->
00:03:55,145
opinionated defaults, especially
when I agree with those opinions.
61
00:03:55,245 -->
00:03:59,665
And we dive into some of their extra
features that they're building on top
62
00:03:59,675 -->
00:04:04,175
of this universal control plane, like
a, a sandboxing approach that is so
63
00:04:04,175 -->
00:04:07,755
popular right now that we have agent
harnesses needing sandboxes, and we're
64
00:04:07,755 -->
00:04:12,665
now talking about the corporate vibe
coders, which I call app creators, the
65
00:04:12,665 -->
00:04:17,461
people that are outside of IT that maybe
need a safe space to experiment with
66
00:04:17,461 -->
00:04:22,541
ideas that they can then hand over to
either robots or humans to maybe develop
67
00:04:22,701 -->
00:04:24,151
further and get onto their production.
68
00:04:24,431 -->
00:04:28,901
Those people are coming, and us platform
engineers and DevOps engineers, we're
69
00:04:29,051 -->
00:04:32,901
having to deal with that problem, and it
feels like we need some solutions that
70
00:04:32,901 -->
00:04:37,671
help automate that middle layer out of the
way with a higher level of abstraction,
71
00:04:37,771 -->
00:04:38,891
and it looks like that's what they built.
72
00:04:38,991 -->
00:04:42,261
So let's get into my conversation
with Doron about Control Plane.
73
00:04:42,361 -->
00:04:43,251
welcome to the show.
74
00:04:43,261 -->
00:04:44,011
I'm glad to have you here.
75
00:04:44,111 -->
00:04:45,411
Doron: g-great to be here.
76
00:04:45,501 -->
00:04:46,141
looking forward to it
77
00:04:46,521 -->
00:04:50,721
Bret (2): Give us a little bit of your
background, Tell me the origin story,
78
00:04:50,756 -->
00:04:53,606
Doron: I ran a company, sold a
company, was chief architect at
79
00:04:53,606 -->
00:04:57,536
Dell, chief architect at, SAP, and
then ended up as, chief architect,
80
00:04:57,566 -->
00:04:59,936
for VMware for cloud services.
81
00:05:00,036 -->
00:05:04,899
And, at VMware, I missed my
interactions back at SAP with a guy
82
00:05:04,899 -->
00:05:10,569
named Dan Wilson, who was the best
DevOps/platform engineer I ever met.
83
00:05:10,669 -->
00:05:15,019
And what I wanted at VMware
was, Dan Wilson as a service.
84
00:05:15,119 -->
00:05:20,939
And, I really saw hundreds of,
outside customers and internal teams
85
00:05:20,979 -->
00:05:25,489
struggle with Kubernetes, Istio,
Grafana, Prometheus, and having to
86
00:05:25,489 -->
00:05:27,859
reinvent the wheel again and again.
87
00:05:27,939 -->
00:05:32,339
And so had the idea for Control
Plane, left VMware, started
88
00:05:32,349 -->
00:05:34,399
this company, called Dan Wilson.
89
00:05:34,469 -->
00:05:39,749
He was back at SAP and I said, "I want you
to join me as my CTO and really deliver to
90
00:05:39,759 -->
00:05:44,969
the world, what you do as a service." And
I thought it's gonna be a hard battle, but
91
00:05:44,979 -->
00:05:48,979
he jumped ship and he said, "Yeah, I'll
join you." And, that was the origin story.
92
00:05:49,079 -->
00:05:53,559
And then we brought on a few other
brilliant, engineers that have decades
93
00:05:53,569 -->
00:05:59,079
of, practice on, on, cloud native
architectures, and, we built this company
94
00:05:59,179 -->
00:06:00,639
Bret (2): this is pre-AI, right?
95
00:06:00,739 -->
00:06:03,979
So, th- we're gonna get into a
whole bunch of details later on,
96
00:06:03,999 -->
00:06:09,399
agent-driven infrastructure, like
MCP and CLI tooling, agent h- harness
97
00:06:09,399 -->
00:06:12,239
first, whatever we wanna call this
thing, th- this new future we're in.
98
00:06:12,339 -->
00:06:15,249
but what did, what was it like
to to be an infrastructure
99
00:06:15,249 -->
00:06:17,299
company and to see AI coming?
100
00:06:17,399 -->
00:06:20,859
Were, were you sorta like seeing
where this all could take us in
101
00:06:20,859 -->
00:06:23,879
terms of agents helping us in
infrastructure and stuff like that?
102
00:06:23,979 -->
00:06:24,829
Doron: 100%.
103
00:06:24,859 -->
00:06:27,129
we, I think, got more lucky.
104
00:06:27,289 -->
00:06:32,509
We have more luck than brain, because
it found us at a very interesting time.
105
00:06:32,509 -->
00:06:38,179
We, we wanted to reduce the cognitive
load on engineers not having to stitch
106
00:06:38,179 -->
00:06:42,299
together Kubernetes, Istio, Grafana,
Prometheus, Vault, and the things…
107
00:06:42,429 -->
00:06:44,509
all the CNCF, solutions out there.
108
00:06:44,609 -->
00:06:51,986
And we created an API that gives you
unified access to any cloud, any service,
109
00:06:52,086 -->
00:06:56,893
and we find that AI, if you just tell
it, "Create me the infrastructure," it's
110
00:06:56,903 -->
00:07:02,225
gonna duplicate a lot of resources: a
NAT gateway, a load balancer, brain, a
111
00:07:02,225 -->
00:07:07,075
Kubernetes set of nodes, and it's gonna
be expensive, non-deterministic, and
112
00:07:07,075 -->
00:07:09,515
it's gonna be non-secure, non-compliant.
113
00:07:09,615 -->
00:07:13,365
And where is the logging, the metrics,
the tracing, the secrets management,
114
00:07:13,365 -->
00:07:15,455
the service discovery, the TLS, the DNS?
115
00:07:15,455 -->
00:07:17,555
And the list goes on and on.
116
00:07:17,655 -->
00:07:23,067
So what AI needs, just like an engineer
needs, is a reduced cognitive load.
117
00:07:23,087 -->
00:07:27,077
In the era of AI is less
tokens, less context window.
118
00:07:27,177 -->
00:07:30,497
Give it a deterministic API
that says, "Hey, take my
119
00:07:30,497 -->
00:07:32,277
desired state and make it so.
120
00:07:32,377 -->
00:07:36,917
And by the way, make it so on
any cloud." It is a boon for AI.
121
00:07:37,652 -->
00:07:40,041
It makes AI a lot more effective.
122
00:07:40,141 -->
00:07:44,831
we find AI in a perfect, intersection
because we can now make it do
123
00:07:44,881 -->
00:07:49,181
tricks that is otherwise very, very
expensive or even u- unattainable
124
00:07:49,281 -->
00:07:53,371
Bret (2): Yeah, I always struggled
with, automation, pre-AI, in terms
125
00:07:53,371 -->
00:07:56,345
of, there's a point at which you're
trying to automate things and
126
00:07:56,345 -->
00:07:59,745
automate triggers and intelligence
with a deterministic language.
127
00:07:59,845 -->
00:08:03,985
And I always would get to a point where
I want this to be more automated, but
128
00:08:03,985 -->
00:08:08,255
the level of effort to automate that
small thing was way more than it was
129
00:08:08,255 -->
00:08:10,005
worth in terms of the automation.
130
00:08:10,005 -->
00:08:13,375
And, and in DevOps forever, we've been
talking about the toil, and I think
131
00:08:13,375 -->
00:08:16,365
that, I think the first time I heard
that was from Google's SRE paper, was
132
00:08:16,365 -->
00:08:20,325
talking about, the toil of ops, and
I always felt like there was a better
133
00:08:20,325 -->
00:08:23,765
way, a more complete way to automate
things, but every time I would try to
134
00:08:23,765 -->
00:08:27,665
take it to the next level, it, it was
like the management had no tolerance
135
00:08:27,675 -->
00:08:30,405
for me spending all of my time trying
to automate these little things.
136
00:08:30,405 -->
00:08:32,945
I'm like, "Yeah, but they're tiny
paper cuts and they add up over time,"
137
00:08:32,945 -->
00:08:35,695
and I was trying to, defend the,
the case for more automation, but
138
00:08:35,695 -->
00:08:37,045
we would just never quite get there.
139
00:08:37,045 -->
00:08:38,775
And now I feel like the
world is our oyster.
140
00:08:38,775 -->
00:08:42,935
I feel like we're, we're at a place where
you're only limited by your imagination
141
00:08:42,965 -->
00:08:46,415
and your ability to, manage the harness
and manage that agent inside it.
142
00:08:46,445 -->
00:08:48,515
Doron: we, we called it
the cloud is your oyster.
143
00:08:48,615 -->
00:08:55,255
We truly allow you to have a singular
API and leverage BigQuery on Google
144
00:08:55,335 -->
00:09:01,405
and Cosmos DB on Azure and S3 on Amazon
and-- or any one of the services.
145
00:09:01,505 -->
00:09:06,385
You can now write code and run it on your
Raspberry Pi or Dell machine or your EC2
146
00:09:06,405 -->
00:09:09,775
instances or EKS, AKS, GKE, whatever.
147
00:09:09,875 -->
00:09:14,759
And you can, not be
confined by any factor.
148
00:09:14,819 -->
00:09:20,109
So think of all the clouds as They have
merged, and now you can run the compute in
149
00:09:20,109 -->
00:09:26,319
a basement or on a cloud or on a secondary
or neo cloud, but leverage any service of
150
00:09:26,319 -->
00:09:28,773
any cloud, and, and let AI do the work.
151
00:09:28,773 -->
00:09:31,513
Like, I need a Postgres, I
need a Redis, I need a Kafka.
152
00:09:31,613 -->
00:09:36,123
You can conversationally get those
things up and running deterministically
153
00:09:36,223 -->
00:09:39,193
Bret (2): All right, so if we back up a
second, If someone hasn't heard of Control
154
00:09:39,193 -->
00:09:43,873
Plane, how is this different than a cloud
or different than a cloud management
155
00:09:43,873 -->
00:09:48,213
tool that we might all be using, like
Terraform or, any of the dozen, different
156
00:09:48,223 -->
00:09:52,623
opportunities I have to manage my cloud
with a CLI or some sort of service?
157
00:09:52,623 -->
00:09:52,973
Doron: Sure.
158
00:09:53,073 -->
00:09:56,653
So Terraform allows you
to create resources.
159
00:09:56,663 -->
00:10:00,203
We provide the Terraform
provider for control plane.
160
00:10:00,453 -->
00:10:04,923
We provide Pulumi, we provide other
interfaces Control Plane is a virtual
161
00:10:04,933 -->
00:10:09,253
cloud, allowing you to not have a
cloud and be cloudless and still
162
00:10:09,253 -->
00:10:14,243
run on Amazon, Google, Azure, Oracle
without a cloud account and pay by
163
00:10:14,253 -->
00:10:16,519
the millicore and megabyte of compute.
164
00:10:16,519 -->
00:10:20,039
So if you deploy the Bret app,
you can run essentially serverless
165
00:10:20,139 -->
00:10:23,899
on any cloud or multiple clouds
without having a cloud account.
166
00:10:23,909 -->
00:10:26,199
You can bring your own cloud account.
167
00:10:26,299 -->
00:10:30,229
If you're a SaaS and your customers
tell you, "Hey, you must run on Azure,"
168
00:10:30,229 -->
00:10:34,439
or, "You must run on Oracle," or,
"Must run on Hetzner," you now can.
169
00:10:34,459 -->
00:10:38,809
Your code, without any change,
can run anywhere, even on-prem.
170
00:10:38,909 -->
00:10:42,746
We let you have observability
built in We let you, as I
171
00:10:42,746 -->
00:10:44,446
mentioned, run by the millicore.
172
00:10:44,546 -->
00:10:47,566
We let you, communicate across clouds.
173
00:10:47,666 -->
00:10:52,036
If you have data in a VPC, let's say
RDS, Relational Database Service, and
174
00:10:52,036 -->
00:10:56,566
you want to run the workload on-prem,
yet communicate with RDS without
175
00:10:56,576 -->
00:10:59,036
poking holes in the firewall, you can.
176
00:10:59,136 -->
00:11:01,266
You can, run above the cloud.
177
00:11:01,266 -->
00:11:05,016
So today, when you run RDS as
an example, you're paying a
178
00:11:05,016 -->
00:11:07,266
high premium to run a database.
179
00:11:07,306 -->
00:11:08,616
You're not paying for the compute.
180
00:11:08,716 -->
00:11:12,506
Control Plane lets you run any
open source project very easily.
181
00:11:12,506 -->
00:11:15,276
We've certified, lots
and lots of, templates.
182
00:11:15,576 -->
00:11:21,396
You can run Kafka, Postgres, MySQL,
Cassandra, Mongo above the cloud.
183
00:11:21,396 -->
00:11:26,366
So if you wanted to move your
persistent store to Linode or Hetzner
184
00:11:26,526 -->
00:11:29,006
or Oracle or on-prem, you can.
185
00:11:29,106 -->
00:11:33,386
If you need multi-master replication,
we support that for many database types.
186
00:11:33,486 -->
00:11:39,246
we provide, sandboxes, so you
can deploy workloads or, or write
187
00:11:39,286 -->
00:11:42,296
code in a isolated environment.
188
00:11:42,396 -->
00:11:46,866
The difference about our, sandboxes
is that they allow you to communicate
189
00:11:46,876 -->
00:11:51,786
to your databases and other resources
in private data centers or VPCs.
190
00:11:51,886 -->
00:11:56,006
And you can traverse the network
boundary, easily connecting Amazon
191
00:11:56,006 -->
00:12:00,736
US East and Azure US West and Oracle
in, the Middle East, et cetera.
192
00:12:00,836 -->
00:12:06,766
So you can think of all the clouds
have, having merged and, with a hundred
193
00:12:06,816 -->
00:12:11,676
x simplification on how to consume
them, but without any limitation.
194
00:12:11,756 -->
00:12:16,166
So if you need to use the
native SDK for SQS, SNS,
195
00:12:18,549 -->
00:12:23,239
Dynamo, or any other native service,
you can we, we-- think of us as
196
00:12:23,249 -->
00:12:27,109
a virtual cloud, just like what
virtualization did to servers.
197
00:12:27,139 -->
00:12:30,459
You didn't care if it's
Dell or HP, we do to cloud.
198
00:12:30,469 -->
00:12:34,479
We make it so you don't care and the
agent doesn't care if you're running
199
00:12:34,479 -->
00:12:37,243
on Azure or any other, cloud provider.
200
00:12:37,443 -->
00:12:41,853
In fact, you can deliver traffic close
to where the customer is, and if there
201
00:12:41,853 -->
00:12:46,443
is a failure, let's say Amazon US East is
down, you automatically get routed to GCP
202
00:12:46,443 -->
00:12:48,463
or to Amazon US West or somewhere else.
203
00:12:48,473 -->
00:12:51,493
So you deliver low latency
and extreme high availability.
204
00:12:51,493 -->
00:12:55,653
We provide a five-nine guarantee,
translating to five minutes, fifteen
205
00:12:55,653 -->
00:12:58,343
seconds per year of cumulative downtime.
206
00:12:58,443 -->
00:13:04,553
So the reason customers come to us is
for unbreakable compute, for, extreme
207
00:13:04,613 -->
00:13:07,723
ease of use, for, cost optimization.
208
00:13:07,763 -->
00:13:12,203
Many of our customers save about fifty
percent compared to running natively
209
00:13:12,233 -->
00:13:17,083
on Amazon or Google because They can
pay for what they use and not pay for
210
00:13:17,103 -->
00:13:18,823
a cow when you need a glass of milk.
211
00:13:18,853 -->
00:13:22,563
'Cause when you buy a Kubernetes
cluster, you pay for the brain nodes,
212
00:13:22,663 -->
00:13:25,843
the worker nodes, you pay a hundred
percent of their utilization, even
213
00:13:25,843 -->
00:13:27,723
though they may consume twenty percent.
214
00:13:27,773 -->
00:13:30,273
And on control plane, you
pay twenty percent if you're
215
00:13:30,273 -->
00:13:31,403
consuming twenty percent.
216
00:13:31,503 -->
00:13:35,433
Bret (2): so you built this sort of
unified API layer over multiple clouds
217
00:13:35,533 -->
00:13:39,663
and, It sounds like you're adding almost
like your own custom services on top
218
00:13:39,663 -->
00:13:42,763
of that like you mentioned the, the
sandboxing, and obviously someone can,
219
00:13:42,763 -->
00:13:45,983
design their own little, EC2 instance
or whatever, or they can, they can try
220
00:13:45,983 -->
00:13:47,323
to design their own sandboxing solution.
221
00:13:47,323 -->
00:13:49,803
But it sounds like you're, you're
adding these as, specific services
222
00:13:49,803 -->
00:13:52,523
on top of that unified API, This
is actually a long question.
223
00:13:52,733 -->
00:13:53,763
I'm spitballing here.
224
00:13:53,933 -->
00:13:56,823
But, you could implement something
in one of those sandboxes or one of
225
00:13:56,823 -->
00:14:00,943
your other, ideas on top of any of the
clouds, whether or not that cloud has a
226
00:14:00,943 -->
00:14:03,613
particular offering labeled that, right?
227
00:14:03,613 -->
00:14:06,943
'Cause when I think of unified
cloud, I think of, reducing it down
228
00:14:06,943 -->
00:14:10,663
to its core elements and then, only
getting to use those core elements.
229
00:14:10,663 -->
00:14:13,463
But it sounds like you're
enriching on top of that API to
230
00:14:13,473 -->
00:14:14,683
give us higher level services.
231
00:14:14,743 -->
00:14:19,113
Doron: yes, people assume because
we unify the cloud, we give
232
00:14:19,113 -->
00:14:21,223
you, lowest common denominator.
233
00:14:21,323 -->
00:14:23,063
But in fact, we're the opposite.
234
00:14:23,093 -->
00:14:27,353
We give you the union of all clouds,
because now You can write the code
235
00:14:27,353 -->
00:14:33,223
with a native SDK for Google, for
Cloud Spanner, for Amazon, for, SQS,
236
00:14:33,233 -->
00:14:38,416
for, Azure, for cloud, for, let's
say, Entra ID or Cosmos DB or whatever
237
00:14:38,516 -->
00:14:42,246
machine learning service that is on
Azure, and we allow you to consume
238
00:14:42,246 -->
00:14:44,876
those services without credentials.
239
00:14:44,976 -->
00:14:47,326
Why is it important in the era of AI?
240
00:14:47,336 -->
00:14:50,856
You do not want to leak to
the LLM, the credentials.
241
00:14:50,956 -->
00:14:57,176
So neither in design time, meaning coding,
nor at runtime are keys exposed, and
242
00:14:57,176 -->
00:15:01,066
in fact, you don't even have to rotate
the keys because you get ephemeral,
243
00:15:01,236 -->
00:15:03,246
zero trust, least privileged tokens.
244
00:15:03,346 -->
00:15:07,486
and we patented the technology
called Universal Cloud Identity that
245
00:15:07,496 -->
00:15:13,656
basically gives you the identity, the
IAM of all the clouds in a much more,
246
00:15:13,756 -->
00:15:18,406
streamlined manner, and the code is
devoid of credentials completely.
247
00:15:18,706 -->
00:15:22,936
Yet the code can only do the
exact, operations on the exact
248
00:15:22,936 -->
00:15:25,426
resources you granted it access to.
249
00:15:25,526 -->
00:15:27,756
Bret (2): what are you
seeing with adoption?
250
00:15:27,816 -->
00:15:31,386
as a, as someone who's… I've managed
infrastructure for over 30 years
251
00:15:31,386 -->
00:15:35,076
at this point, and, w- we've been
chasing this dream in ops land of,
252
00:15:35,126 -->
00:15:36,706
the unified control plane, right?
253
00:15:36,706 -->
00:15:40,336
we've had for decades now, ever since
really the cloud or before, we've
254
00:15:40,336 -->
00:15:42,356
had all these attempts at unifying.
255
00:15:42,356 -->
00:15:43,776
You know, VMware wanted to do it.
256
00:15:43,786 -->
00:15:45,096
Microsoft tried to do it.
257
00:15:45,196 -->
00:15:49,866
we had just all these different pieces,
and the best we could do was, like, stick
258
00:15:49,866 -->
00:15:53,486
to one OS and then use their, I think
they, system management tools or whatever.
259
00:15:53,586 -->
00:15:56,996
And then I feel like ever since then
I've been chasing this dream, whether
260
00:15:56,996 -->
00:16:01,006
it was Terraform or Chef and Puppet
before that, and then we got Pulumi, and
261
00:16:01,006 -->
00:16:05,876
we've had all these different attempts
at, one control plane essentially that
262
00:16:05,876 -->
00:16:09,786
controls not necessarily my workloads,
but just the infrastructure behind it,
263
00:16:09,886 -->
00:16:12,266
and it's always felt like a compromise.
264
00:16:12,376 -->
00:16:17,166
It always felt like if I'm gonna do
Terraform, that's great, but it's gonna
265
00:16:17,166 -->
00:16:21,776
get much harder as I go multi-cloud, and
I'm gonna be spending my days in TOML
266
00:16:21,786 -->
00:16:25,136
hell just trying to make sure I don't
break things with a, with an update.
267
00:16:25,236 -->
00:16:28,396
So I'm curious of, like, some
customer stories around, onboarding.
268
00:16:28,416 -->
00:16:31,076
How do people get to control
plane from where they are,
269
00:16:31,076 -->
00:16:31,976
and what does that look like?
270
00:16:32,076 -->
00:16:32,376
Doron: Yeah.
271
00:16:32,376 -->
00:16:36,716
So we have, over, eighteen hundred,
now over two thousand customers.
272
00:16:36,786 -->
00:16:40,086
Some of them are very large,
some of them are tiny, doing
273
00:16:40,086 -->
00:16:41,746
vibe, coding type projects.
274
00:16:41,846 -->
00:16:46,576
And most of the larger customers,
the-- none of them are greenfield.
275
00:16:46,616 -->
00:16:51,683
They're all coming from Amazon,
from Google, from Heroku, from, uh,
276
00:16:52,263 -->
00:16:57,045
Azure, and they all have a sprawling,
degree of Helm charts and, Terraform
277
00:16:57,052 -->
00:16:59,852
or Pulumi or Crossplane, artifacts.
278
00:16:59,952 -->
00:17:06,695
And we have, mechanical translation in
the CLI allowing you to take a Kubernetes
279
00:17:06,695 -->
00:17:10,645
YAML and enrich it to the Control
Plane YAML because Kubernetes doesn't
280
00:17:10,655 -->
00:17:17,175
have the concept of the global virtual
cloud, a cloud made up of N clusters.
281
00:17:17,185 -->
00:17:21,345
So in Control Plane, I can take three
clusters or three hundred and turn
282
00:17:21,345 -->
00:17:25,125
them into one global virtual cloud
or actually subdivide them into many
283
00:17:25,165 -->
00:17:30,405
global virtual clouds and deploy an
application to the GVC, which is like
284
00:17:30,405 -->
00:17:36,965
a VPC, but spread across two regions or
a hundred in one cloud or many So our
285
00:17:36,965 -->
00:17:41,565
customers typically have a couple days
up to a couple weeks, depending on how
286
00:17:41,565 -->
00:17:45,795
large the customer is, of migration,
which is, again, that mechanical
287
00:17:45,895 -->
00:17:51,205
translation of their YAML or Helm chart
or other artifacts to Control Plane.
288
00:17:51,305 -->
00:17:54,475
But we integrate very nicely
into their CI/CD pipelines.
289
00:17:54,575 -->
00:17:56,935
We integrate with Flux or Argo CD.
290
00:17:57,035 -->
00:17:58,565
so it's very natural.
291
00:17:58,845 -->
00:18:00,675
The learning curve is very shallow.
292
00:18:00,775 -->
00:18:04,655
Once you learn Control Plane, you
feel, and what I hear from customers,
293
00:18:04,755 -->
00:18:07,155
we run circles around our peers.
294
00:18:07,255 -->
00:18:11,245
And one customer, I think the best
compliment I ever got was after
295
00:18:11,245 -->
00:18:15,085
using Control Plane, moving back
to not using Control Plane is like
296
00:18:15,085 -->
00:18:19,025
going from iPhone seventeen to
Nokia from nineteen ninety-seven.
297
00:18:19,125 -->
00:18:22,865
That's-- And, and now, that was
very rewarding for me to hear, and
298
00:18:22,945 -->
00:18:24,895
I would-- I'm eager to demonstrate.
299
00:18:24,995 -->
00:18:29,785
and my objective is after this,
call, y- are you gonna go back to
300
00:18:29,785 -->
00:18:33,355
the old ways or are you gonna adopt
Control Plane to deploy your code?
301
00:18:33,455 -->
00:18:37,405
I think it's the latter, because,
it's-- we, we truly give you,
302
00:18:37,505 -->
00:18:40,495
Dan Wilson as a service, the one
I, I mentioned, I started with.
303
00:18:40,505 -->
00:18:45,435
He's our CTO, and he really, rather than
you having to spend or the AI having
304
00:18:45,455 -->
00:18:51,005
to spend millions of tokens reinventing
the wheel, focus on what matters, which
305
00:18:51,005 -->
00:18:56,225
is your rocket shipping system or your,
whatever the thing that you're doing
306
00:18:56,325 -->
00:18:59,015
and not be, knee-deep in the toil.
307
00:18:59,295 -->
00:19:00,435
and there is a lot of toil.
308
00:19:00,465 -->
00:19:04,235
It's one thing to have toil,
but why should every company
309
00:19:04,335 -->
00:19:06,605
reinvent the same wheel?
310
00:19:06,645 -->
00:19:11,565
Logging, metrics, tracing, secrets,
m- mutual TLS between services because
311
00:19:11,575 -->
00:19:17,245
you want a service mesh, audit trail,
scaling, scaling strategies, failover
312
00:19:17,345 -->
00:19:19,475
Should every company reinvent that wheel?
313
00:19:19,475 -->
00:19:20,475
Does that make sense?
314
00:19:20,565 -->
00:19:24,715
Should every company have an Exchange
server engineer babysitting email?
315
00:19:24,995 -->
00:19:26,495
That was the case.
316
00:19:26,775 -->
00:19:29,915
It isn't anymore because it
doesn't make sense when you can,
317
00:19:30,015 -->
00:19:33,965
shift left or whatever, delegate
that, mundane responsibility to
318
00:19:33,965 -->
00:19:35,935
somebody who is excelling at it.
319
00:19:36,187 -->
00:19:36,987
And think about it.
320
00:19:36,997 -->
00:19:41,317
If every company reinvents the same
items I just mentioned, it represents
321
00:19:41,357 -->
00:19:46,484
technical debt because now you need to
maintain, fix the CVEs, upgrade Kubernetes
322
00:19:46,504 -->
00:19:50,434
every three months, upgrade Istio,
upgrade Prometheus, upgrade Grafana.
323
00:19:50,534 -->
00:19:54,594
When do you have time to do the
actual work when you're babysitting
324
00:19:54,594 -->
00:19:57,764
an infrastructure that your
peers are doing the same thing?
325
00:19:57,864 -->
00:20:01,214
We say stop with the babysitting
of infrastructure, start
326
00:20:01,254 -->
00:20:03,134
delivering value to customers,
327
00:20:03,234 -->
00:20:05,094
Bret (2): We talked about this
last time when we, we had a
328
00:20:05,114 -->
00:20:06,394
r- a call a couple weeks ago.
329
00:20:06,494 -->
00:20:10,890
The, the assumption that I have about
services like this, like if I'm looking
330
00:20:10,890 -->
00:20:13,590
at your homepage and I've not talked
to you and I don't know anything about
331
00:20:13,590 -->
00:20:17,640
the company, is that, to me, there's
always a compromise, I feel like.
332
00:20:17,650 -->
00:20:23,100
And so when I think about the complexity
of the cloud APIs, just for AWS, the,
333
00:20:23,100 -->
00:20:26,900
"This Week in AWS" podcast with Corey
Quinn, I've been, I've been on his
334
00:20:26,900 -->
00:20:30,950
show, he's been on mine, but I, I always
find his articles about, yet another
335
00:20:30,960 -->
00:20:33,960
AWS service that we don't understand
what it is, and it's redundant.
336
00:20:33,960 -->
00:20:37,370
And, and so he talks about the hundreds
and hundreds of services they have, and
337
00:20:37,630 -->
00:20:41,480
it's, th- there's always a customer for
the things, but it may not be you or me.
338
00:20:41,580 -->
00:20:46,373
And I always find it hilariously
complex that I am so behind on how
339
00:20:46,373 -->
00:20:50,053
many of those products I don't know
about and I've never used, and it just,
340
00:20:50,153 -->
00:20:52,223
Am I even an engineer if I
don't understand half the
341
00:20:52,223 -->
00:20:53,843
things that AWS deploys, right?
342
00:20:53,843 -->
00:20:54,763
Sometimes I question it.
343
00:20:54,763 -->
00:20:59,293
But, I'm curious about, how your API
relates to their product APIs, and how
344
00:20:59,293 -->
00:21:02,923
you guys have been able to innovate
around covering all the bases to make
345
00:21:02,923 -->
00:21:07,073
sure that we're not left with, this
reduced abstraction that's only the
346
00:21:07,173 -->
00:21:10,573
same core fundamentals across every
cloud, but you're allowed to, you're
347
00:21:10,573 -->
00:21:12,233
able to e- enrich it a little bit.
348
00:21:12,233 -->
00:21:13,103
Tell me a little bit about that.
349
00:21:13,123 -->
00:21:13,673
How's that work?
350
00:21:13,773 -->
00:21:15,423
Doron: Yeah, that's a great question.
351
00:21:15,470 -->
00:21:19,153
So I mentioned we're not the
lowest common denominator, but the
352
00:21:19,153 -->
00:21:20,673
union, and what does that mean?
353
00:21:20,773 -->
00:21:25,333
It means if Amazon came up with a
Bret service a microsecond from now,
354
00:21:25,603 -->
00:21:30,273
we would support it because we're
not trying to, abstract the API,
355
00:21:30,283 -->
00:21:32,583
the native API to the Bret service.
356
00:21:32,583 -->
00:21:35,543
We integrate at the identity
and access management layer.
357
00:21:35,617 -->
00:21:41,910
And so we would mechanically be able
to pull permissions from their IAM IAM
358
00:21:41,910 -->
00:21:44,170
system relating to the new Bret service.
359
00:21:44,210 -->
00:21:49,340
The coder, AI or human, would use
the native SDK if there is one to the
360
00:21:49,340 -->
00:21:51,670
Bret service or the C-- or the API.
361
00:21:51,743 -->
00:21:56,913
We just inject the token necessary
for the Bret service to do its work,
362
00:21:57,203 -->
00:22:01,433
and we're able to do it whether you--
the workload is running on-prem,
363
00:22:01,533 -->
00:22:06,587
on Hetzner, on DigitalOcean, , on
Google, on Amazon, on Azure.
364
00:22:06,687 -->
00:22:11,610
So because we integrate at the IAM, the
identity and access management layer
365
00:22:11,710 -->
00:22:12,100
Bret (2): Yeah
366
00:22:12,388 -->
00:22:16,388
Doron: We are supportive of any
new service that Google, Amazon or
367
00:22:16,388 -->
00:22:18,968
anybody else provides instantly.
368
00:22:18,968 -->
00:22:21,188
There is no work that we need to do.
369
00:22:21,288 -->
00:22:27,128
And the likelihood of Amazon or anybody
else changing their identity and access
370
00:22:27,128 -->
00:22:32,741
management APIs that we exploit are
zero, because if they change their APIs,
371
00:22:32,771 -->
00:22:34,631
the cl-- the world would break, right?
372
00:22:34,651 -->
00:22:38,521
Because everybody, millions of
people have, utilized their services.
373
00:22:38,531 -->
00:22:42,411
So if they were to change, they
would give you multi-year head start.
374
00:22:42,511 -->
00:22:48,151
but, because again, we integrate at the
IAM level, when you consume, let's say,
375
00:22:48,221 -->
00:22:54,125
RDS, but you're on, Azure, the, the, the
workload is running on Azure or on-prem,
376
00:22:54,225 -->
00:22:59,845
we obtain the correct token from their
STS, their security token service, and
377
00:22:59,845 -->
00:23:05,845
know how to inject it at runtime from
the SDK to RDS, in our case, And, so in
378
00:23:05,845 -->
00:23:11,418
effect, the workload is liberated and
no longer captive to the subject cloud.
379
00:23:11,418 -->
00:23:13,658
You can now move amongst the clouds.
380
00:23:13,758 -->
00:23:16,818
People spent millions of dollars
moving from Amazon to Google or
381
00:23:16,828 -->
00:23:20,888
Google to Oracle, but on Control
Plane, the cloud is your oyster.
382
00:23:20,888 -->
00:23:24,488
You can move from Amazon to Google,
from Google to on-prem, from on-prem
383
00:23:24,488 -->
00:23:29,958
to whatever with a click of a button,
and you can leverage all of them.
384
00:23:30,258 -->
00:23:34,608
So we flip the script where
the clouds want to-- their
385
00:23:34,638 -->
00:23:36,438
objective is to lock you in.
386
00:23:36,538 -->
00:23:41,988
We flip it on its head and, turn them
in-into just commoditized utilities.
387
00:23:42,088 -->
00:23:46,068
So when you consume electricity,
your TV doesn't care if it's
388
00:23:46,108 -->
00:23:48,061
Edison or somebody else.
389
00:23:48,161 -->
00:23:49,641
It cares about the interface.
390
00:23:49,741 -->
00:23:55,545
With Control Plane, you now have a virtual
cloud operating system that utilizes
391
00:23:55,585 -->
00:23:59,245
the power, the whole power of Amazon,
the whole power of Google, et cet- et
392
00:23:59,245 -->
00:24:01,415
cetera, but without locking you in.
393
00:24:01,435 -->
00:24:01,815
Bret (2): Yeah.
394
00:24:01,915 -->
00:24:05,095
I wanna change the conversation
a little bit to be more specific
395
00:24:05,115 -->
00:24:09,135
on some of the AI, uh, stuff that
we're all managing right now.
396
00:24:09,135 -->
00:24:14,025
So, this whole year for me has been really
focused in understanding other DevOps
397
00:24:14,045 -->
00:24:18,265
teams and, platform engineering teams,
like understanding the way that they're
398
00:24:18,265 -->
00:24:20,305
using agent harnesses to do their job.
399
00:24:20,305 -->
00:24:23,255
I think, if we f- if we rewound a year
ago, maybe even three or four years
400
00:24:23,255 -->
00:24:27,241
ago when Google Copilot was kinda like
the first IDE to throw AI completion
401
00:24:27,241 -->
00:24:30,221
in, and even in those early days,
I think I demoed it when it was in
402
00:24:30,221 -->
00:24:31,641
beta when we were on this channel.
403
00:24:31,641 -->
00:24:33,891
I remember one summer we were, I
think it was 2022 or something.
404
00:24:33,901 -->
00:24:37,131
It was before ChatGPT came out, and we
were, like, looking at the, the early
405
00:24:37,131 -->
00:24:41,851
GPT model that was embedded in, in
GitHub Copilot, and the, VS Code UI.
406
00:24:41,951 -->
00:24:44,891
And we were, we were trying to
see how far we could get with YAML
407
00:24:44,891 -->
00:24:47,391
and TOML by tab completing, right?
408
00:24:47,391 -->
00:24:50,561
Like, we were trying to figure out
how much time can it save me as an,
409
00:24:50,571 -->
00:24:54,501
as a ops engineer, but we were really
just talking about file completion,
410
00:24:54,551 -->
00:24:56,461
completion in a YAML statement.
411
00:24:56,471 -->
00:25:00,891
So, it was saving me keystrokes, and
it maybe, was saving me potentially
412
00:25:00,891 -->
00:25:03,481
some human error that would
hopefully be caught by a linter,
413
00:25:03,581 -->
00:25:05,041
but it wasn't a huge gain, right?
414
00:25:05,041 -->
00:25:07,611
And so you fast-forward a
little bit, and then we're all,
415
00:25:07,651 -->
00:25:08,751
we're doing that with ChatGPT.
416
00:25:08,751 -->
00:25:09,831
People are copying and pasting.
417
00:25:09,831 -->
00:25:13,371
And then last year when we got Claude
Code for the first time and we under- we
418
00:25:13,371 -->
00:25:17,281
started to understand what an agent loop
is, it, it started to shift and I started
419
00:25:17,281 -->
00:25:22,181
to notice that my, my production friends
were all dabbling with letting their, the
420
00:25:22,181 -->
00:25:25,531
first thing they would do is let their
agent have, a read access token to some
421
00:25:25,531 -->
00:25:29,088
part of their infrastructure so that it
could maybe run the kubectl commands for
422
00:25:29,088 -->
00:25:31,978
them or run the AWS CLI commands for them.
423
00:25:32,078 -->
00:25:36,695
And I was doing that same thing
for, like, GitHub CLIs for, for CI.
424
00:25:36,695 -->
00:25:36,705
CI.
425
00:25:36,715 -->
00:25:39,675
Like, I cared about GitHub Actions
and, my build status and stuff
426
00:25:39,675 -->
00:25:42,505
like that, so I was, I was managing
that middle portion, and I was
427
00:25:42,505 -->
00:25:43,875
using agents to, keep track of it.
428
00:25:43,875 -->
00:25:47,205
And now it's like these are all inherent
features that are really built in.
429
00:25:47,215 -->
00:25:49,675
They, they've got the background
workers, the sub-agents, and things
430
00:25:49,675 -->
00:25:51,165
that are making this really possible.
431
00:25:51,165 -->
00:25:53,745
But I'm curious what you're
seeing with your customers.
432
00:25:53,895 -->
00:25:57,205
how, how are they driving
their infrastructure with AI?
433
00:25:57,825 -->
00:26:00,305
Like, I guess maybe if we can get
to some, into some nitty-gritty
434
00:26:00,305 -->
00:26:03,915
details around, are they focused
on only using their agent harness?
435
00:26:03,915 -->
00:26:07,025
Are they using, MCP to, to Control Plane?
436
00:26:07,025 -->
00:26:09,058
Are they using, Control Plane CLIs?
437
00:26:09,098 -->
00:26:11,838
Like, how are they actually doing
this work, and what are some
438
00:26:11,838 -->
00:26:12,668
of the stories you're hearing?
439
00:26:12,768 -->
00:26:17,478
Doron: Yeah, our customers, pretty
much all use our MCP server,
440
00:26:17,578 -->
00:26:24,318
and we always pride ourselves in
providing all the accessibility.
441
00:26:24,418 -->
00:26:29,008
Of course, at the lower end is, is the
API, but everything you can do in the
442
00:26:29,028 -->
00:26:35,131
API, you can do in the CLI, Terraform,
Pulumi, Crossplane, UI, and MCP.
443
00:26:35,231 -->
00:26:37,810
To us, these are symmetric interfaces.
444
00:26:37,910 -->
00:26:42,640
Of course, with MCP, you can now bring
the intelligence to let it discover
445
00:26:42,640 -->
00:26:46,872
what can be done, and now you can
have conversational infrastructure.
446
00:26:46,872 -->
00:26:50,242
You now have the equivalent of
a enterprise architect/platform
447
00:26:50,362 -->
00:26:57,062
engineer/DevOps/SRE bundled into one
that you can have a conversation with.
448
00:26:57,202 -->
00:27:00,302
you can tell me the app you want
to build and the dependencies that
449
00:27:00,302 -->
00:27:04,698
you want to have, and the MCP will
go and negotiate and build those
450
00:27:04,698 -->
00:27:07,138
things that we want to build for us.
451
00:27:07,238 -->
00:27:11,521
And then if we need to troubleshoot
or need to look at observability, the
452
00:27:11,541 -->
00:27:15,981
MCP is able to look at the traces,
the logs, the metrics, a- and add the
453
00:27:15,981 -->
00:27:18,901
intelligence to give you actionable items.
454
00:27:19,201 -->
00:27:25,870
So you feel like you have the best SREs
and platform engineers at Google with you
455
00:27:25,870 -->
00:27:28,120
twenty-four seven, and you don't pay them.
456
00:27:28,220 -->
00:27:32,690
So you can vibe code a very complex
system made up of 20 microservices all
457
00:27:32,700 -->
00:27:37,570
intertwined with databases and queues
and, and caches and say, "Okay, give
458
00:27:37,570 -->
00:27:42,980
it to me codified." And our codified
version is always backwards compatible.
459
00:27:42,990 -->
00:27:46,060
So I'm sure Kubernetes deprecate
stuff every three months.
460
00:27:46,060 -->
00:27:48,170
You have some deprecations and new APIs.
461
00:27:48,270 -->
00:27:50,730
It's a fairly unstable substrate.
462
00:27:50,830 -->
00:27:53,920
So if you wanted to take YAML
from 10 years ago and just run
463
00:27:53,920 -->
00:27:56,040
it, you'll run into issues.
464
00:27:56,140 -->
00:27:57,830
So we made an early choice.
465
00:27:57,930 -->
00:27:59,790
Bret (2): At a minimum
warnings, possibly errors, yeah
466
00:27:59,890 -->
00:28:00,240
Doron: Right.
467
00:28:00,290 -->
00:28:01,800
And we made a decision.
468
00:28:01,880 -->
00:28:06,880
We, we noticed that six years ago and
made the very, I think, good decision of
469
00:28:06,910 -->
00:28:09,470
always having backwards compatibility.
470
00:28:09,570 -->
00:28:14,800
So if you take YAML from six years ago and
apply it today, you'll get the exact same
471
00:28:14,830 -->
00:28:19,880
artifacts on top of any compute substrate,
whether it's on-prem, Azure, etcetera.
472
00:28:19,980 -->
00:28:23,075
Bret (2): we-- Are-- When you s-
when you say that, are we talking
473
00:28:23,075 -->
00:28:27,495
about your customers can continue
to use Terraform or Pulumi?
474
00:28:27,605 -->
00:28:30,395
I'm assuming that that's how they're,
they're interacting with you, right?
475
00:28:30,395 -->
00:28:33,915
Like, they're, they're talking
to your API from these particular
476
00:28:34,015 -->
00:28:35,495
management tools, right?
477
00:28:35,495 -->
00:28:38,492
I'm ass- I'm assuming, um,
OpenTofu is s- is supported.
478
00:28:38,712 -->
00:28:40,432
What else, what are the other ways?
479
00:28:40,462 -->
00:28:43,332
'Cause, o- one of the things I've
started to see with agents is if,
480
00:28:43,342 -->
00:28:46,662
if you're a singular cloud, it might
just be better if your, if your
481
00:28:46,682 -->
00:28:50,952
agent is the complete interface to h-
whatever the description language is.
482
00:28:51,052 -->
00:28:54,612
I've seen some examples of people actually
switching away from Terraform back to
483
00:28:54,612 -->
00:28:59,072
the native, cloud formation or whatever
of their cloud, because they look at
484
00:28:59,072 -->
00:29:03,422
it as there's this agent abstraction
between me and the YAML or the JSON.
485
00:29:03,432 -->
00:29:08,222
I don't have to be the expert in every
single line of that anymore, so I
486
00:29:08,222 -->
00:29:12,012
actually care more about, some of the
inherent functionality of a different,
487
00:29:12,112 -->
00:29:16,442
of, of the language tooling itself rather
than, we always looked at Terraform
488
00:29:16,442 -->
00:29:19,152
as maybe a little bit more to, easier
to understand than cloud formation.
489
00:29:19,352 -->
00:29:22,552
A little bit, a little less comprehensive,
a little, little less verbose.
490
00:29:22,652 -->
00:29:25,082
But if you've got an agent in
the way, that matters less.
491
00:29:25,112 -->
00:29:26,912
Are you seeing anything like that?
492
00:29:26,962 -->
00:29:30,122
Are, are you seeing people,
switch, products now that they're
493
00:29:30,142 -->
00:29:32,192
using agents, as their interface?
494
00:29:32,292 -->
00:29:33,612
Doron: Somewhat, yes.
495
00:29:33,642 -->
00:29:37,152
I've seen people that, that used
certain… I don't want to name names
496
00:29:37,152 -->
00:29:42,132
'cause I don't want to dish on any one
platform, but there are many equivalents.
497
00:29:42,222 -->
00:29:46,122
Like when-- whether you use
Crossplane, Pulumi, CloudFormation,
498
00:29:46,262 -->
00:29:49,092
Terraform, and a few others,
they're functionally equivalent.
499
00:29:49,092 -->
00:29:50,872
They're like French, English, Spanish.
500
00:29:51,142 -->
00:29:53,622
You can express the same
ideas in these languages.
501
00:29:53,622 -->
00:29:57,332
They do things in slightly different
manners, like Pulumi is more,
502
00:29:57,432 -->
00:30:01,682
programmatic than, than, declarative,
and Terraform is more declarative.
503
00:30:01,782 -->
00:30:06,372
But at the end of the day, they
express a desired state, and there is a
504
00:30:06,382 -->
00:30:10,822
translation layer that takes the desired
state, like with DAG on, on the case
505
00:30:10,822 -->
00:30:16,222
of, Terraform, and applies in the right
order, calls the APIs that are affecting
506
00:30:16,222 -->
00:30:18,292
the, the actual resources in the cloud.
507
00:30:18,392 -->
00:30:20,032
It's just a translation, right?
508
00:30:20,032 -->
00:30:23,272
So like you said, if I don't
have to write to that language,
509
00:30:23,302 -->
00:30:27,722
why can't I talk in English and
let it make the calls directly?
510
00:30:27,822 -->
00:30:28,852
But here is the thing.
511
00:30:28,952 -->
00:30:35,472
Terraform, as an example, or OpenTofu,
invested heavily in doing the DAG and
512
00:30:35,472 -->
00:30:39,872
doing the validation and doing stuff that
the agent itself will have to reinvent
513
00:30:39,882 -->
00:30:44,232
what they had done to apply things in the
right order and give you the rigor that
514
00:30:44,232 -->
00:30:46,232
they have already baked into the platform.
515
00:30:46,332 -->
00:30:50,942
It is better, like in the Excel example
I've ga-- I've given, to let the
516
00:30:50,962 -->
00:30:56,462
agent leverage a tool like Terraform
than to learn, the underlying, API.
517
00:30:56,472 -->
00:30:57,542
It's, it's possible.
518
00:30:57,592 -->
00:31:00,422
here it's, it's, it's
not an exact science.
519
00:31:00,432 -->
00:31:03,222
Here is subject to, to, judgment.
520
00:31:03,322 -->
00:31:07,092
It can call the, the direct, it
can call the API, it can call
521
00:31:07,092 -->
00:31:09,372
the CLI of the AWS, let's say.
522
00:31:09,472 -->
00:31:14,542
It can call, a CDK, There are
many, again, equivalent, logically
523
00:31:14,542 -->
00:31:15,992
equivalent, options here.
524
00:31:16,242 -->
00:31:16,562
So
525
00:31:16,662 -->
00:31:16,792
Bret (2): So
526
00:31:16,896 -->
00:31:17,066
Doron: our
527
00:31:17,072 -->
00:31:17,382
Bret (2): you, I
528
00:31:17,482 -->
00:31:21,532
Doron: w- w- we, we just give you the,
the one API, and it's your choice.
529
00:31:21,562 -->
00:31:26,392
You tell the AI, " Use the API, use
Terraform, use YAML, use JSON, JSON,
530
00:31:26,402 -->
00:31:31,072
use, CRD." We give the choice to the
customer or the AI, and the AI would
531
00:31:31,072 -->
00:31:35,862
choose the, the least token-consuming
manner that it can operate in
532
00:31:35,962 -->
00:31:40,322
Bret (2): So when someone's migrating
to your platform, if they're using
533
00:31:40,322 -->
00:31:43,922
CloudFormation, d- I'm assuming you
don't support CloudFormation because
534
00:31:43,932 -->
00:31:45,702
you're sitting in front of the AWS.
535
00:31:45,792 -->
00:31:46,252
Like, you're, you're
536
00:31:46,388 -->
00:31:46,718
Doron: Correct.
537
00:31:46,822 -->
00:31:47,302
Bret (2): different
538
00:31:47,402 -->
00:31:50,232
Doron: we, we don't support cloud
formation for control plane.
539
00:31:50,242 -->
00:31:53,672
We have customers that use cloud
formation to create their S3 buckets
540
00:31:53,672 -->
00:31:57,152
and their RDS and their Dynamo,
then, then they tell control plane,
541
00:31:57,252 -->
00:32:02,242
run a workload that connects to
the S3 bucket to RDS to Dynamo.
542
00:32:02,342 -->
00:32:04,282
They're not, in conflict with each other.
543
00:32:04,322 -->
00:32:05,872
They're just not overlapping.
544
00:32:05,972 -->
00:32:10,502
There is no reason for us to introduce
a cloud formation concept 'cause
545
00:32:10,502 -->
00:32:13,412
it's a proprietary, lingo of AWS.
546
00:32:13,412 -->
00:32:17,052
So we are operating at
the cloud agnostic, layer.
547
00:32:17,112 -->
00:32:18,592
We fly above the cloud.
548
00:32:18,692 -->
00:32:20,772
We don't wanna fly in, in that cloud.
549
00:32:20,782 -->
00:32:21,712
We fly above it
550
00:32:21,812 -->
00:32:23,952
Bret (2): Do you see, do you
think that Terraform is still the
551
00:32:23,962 -->
00:32:26,622
primary way your customers are
managing their infrastructure?
552
00:32:26,642 -->
00:32:27,822
Do you see that as, the majority?
553
00:32:27,922 -->
00:32:29,569
Doron: I would call it 60%.
554
00:32:29,669 -->
00:32:30,059
Bret (2): Okay
555
00:32:30,169 -->
00:32:33,489
Doron: Just because of inertia,
Not because it has a, a, a
556
00:32:33,489 -->
00:32:35,219
specific advantage, like I said.
557
00:32:35,319 -->
00:32:39,899
Whether it's Terraform, Pulumi, or
Crossplane, or something else, All they
558
00:32:39,899 -->
00:32:45,009
are, are syntactic sugar on top of the
API with some DAG, that, that decides
559
00:32:45,029 -->
00:32:49,209
what to create first and what to destroy
first, and maybe save the state and so on.
560
00:32:49,309 -->
00:32:53,739
So to me, those are fairly
commoditized, languages.
561
00:32:53,839 -->
00:32:57,669
So for, for the audience to
understand, we don't compete with
562
00:32:57,669 -->
00:32:59,489
Terraform, not even a little bit.
563
00:32:59,589 -->
00:33:02,239
It's like a car doesn't
compete with, a sandwich.
564
00:33:02,439 -->
00:33:04,449
It's two different animals completely.
565
00:33:04,549 -->
00:33:04,809
Bret (2): yeah.
566
00:33:04,859 -->
00:33:07,144
so we talked about customer
base, and, like, the different
567
00:33:07,144 -->
00:33:08,384
size of customers, right?
568
00:33:08,384 -->
00:33:12,334
And it's hard for, any one cloud to
always, a-adopt all customers, right?
569
00:33:12,384 -->
00:33:16,024
We, we all think of, the digital oceans
and, the Vercels of the world, that
570
00:33:16,024 -->
00:33:19,374
they're all like a very niche cloud,
that they've got a very specific purpose,
571
00:33:19,384 -->
00:33:22,464
and then you have the three hyperscalers
that are just everything for everyone,
572
00:33:22,464 -->
00:33:26,294
but it's also impossible for anyone
to get on board really easily, right?
573
00:33:26,394 -->
00:33:31,324
Do you, do you see, are small teams
of, with n-no developers or very few
574
00:33:31,324 -->
00:33:34,784
operators or anything like that, do
they-- Is Control Plane being used by
575
00:33:34,784 -->
00:33:36,434
those kinds of small little companies?
576
00:33:36,464 -->
00:33:39,924
'Cause I, I have a scenario where
I'm thinking, I have, companies
577
00:33:39,924 -->
00:33:44,554
that are, they, th-they might have
one or two IT staff, but, they
578
00:33:44,554 -->
00:33:45,704
don't have internal developers.
579
00:33:46,004 -->
00:33:49,624
So usually when it, it's hard to figure
out, as a consultant, your heart, it's
580
00:33:49,624 -->
00:33:51,374
hard to figure out, what do I do for them?
581
00:33:51,374 -->
00:33:54,244
Because I either have to put them
on a hyperscaler, and then now they
582
00:33:54,244 -->
00:33:56,104
can't do anything without me, right?
583
00:33:56,104 -->
00:33:59,434
I become a bottleneck, which I, I, me as
a consultant, I don't ever wanna do that.
584
00:33:59,434 -->
00:34:01,394
I wanna be instantly
replaceable re-usually.
585
00:34:01,494 -->
00:34:06,124
Or I have to put them on some sort of
niche cloud that, if it's too niche, it
586
00:34:06,124 -->
00:34:10,084
may not be here in five years, or it might
be somehow limited i-in, in what it can
587
00:34:10,084 -->
00:34:14,394
do, and then I'm li- like a Lovable for
AI or whatever, like these, these little
588
00:34:14,424 -->
00:34:16,104
fly-by-night dashboard companies, right?
589
00:34:16,404 -->
00:34:19,784
Do you… W-what's it look like for the
really small teams on, on Control Plane?
590
00:34:19,884 -->
00:34:24,704
Doron: The good news, you will always
have a substrate to run upon 'cause it can
591
00:34:24,704 -->
00:34:29,394
be, like I said, your on-prem, it can be
Azure, Google, Amazon, Oracle, et cetera.
592
00:34:29,494 -->
00:34:34,194
And we have customers that have, non,
traditional developers, certainly no
593
00:34:34,194 -->
00:34:38,724
dev… We have DevOps engineers that
use Control Plane, platform engineers,
594
00:34:38,824 -->
00:34:43,584
but we have people that are just vibe
coding and not know anything about…
595
00:34:43,604 -->
00:34:46,284
They don't know the if
statement, and they still deploy
596
00:34:46,284 -->
00:34:48,064
successfully on Control Plane.
597
00:34:48,164 -->
00:34:51,174
case in point, my younger
son, wanted to build an app.
598
00:34:51,214 -->
00:34:53,834
He never wrote a code, any
piece of code in his life.
599
00:34:54,084 -->
00:34:57,824
He was able to write a full game,
which was pretty impressive, and
600
00:34:57,824 -->
00:35:00,704
run it on Control Plane, and he
didn't even ask me how to do it.
601
00:35:00,804 -->
00:35:05,244
So he deployed the whole game with
a back end, and, I was shocked.
602
00:35:05,284 -->
00:35:08,574
You know, I, I helped develop the
product, and I was shocked at,
603
00:35:08,624 -->
00:35:12,174
at, at what he was able to do, and
he's never written a line of code
604
00:35:12,194 -->
00:35:14,544
despite me wanting him to, to learn.
605
00:35:14,644 -->
00:35:15,664
But he was able to do it.
606
00:35:15,664 -->
00:35:19,294
So we have many people with
no coding experience that are
607
00:35:19,294 -->
00:35:20,674
able to be very successful.
608
00:35:20,774 -->
00:35:21,124
Bret (2): Yeah.
609
00:35:21,224 -->
00:35:23,844
Do you think that's a-- With AI, I mean,
I, I know my answer to this question,
610
00:35:23,844 -->
00:35:25,114
but, it's like a leading question.
611
00:35:25,114 -->
00:35:28,294
Do you think that with these agents
getting better and better and th-
612
00:35:28,304 -->
00:35:32,384
there's the… Now we got Cowork and,
and Claude Code built into people's
613
00:35:32,484 -->
00:35:35,674
Claude and ChatGPT subscriptions for
people that aren't technical, right?
614
00:35:35,674 -->
00:35:37,454
You don't have to use the TUI anymore.
615
00:35:37,554 -->
00:35:41,634
And so I feel like we're, we're heading to
a, pretty quickly to a place where non-IT
616
00:35:42,040 -->
00:35:45,780
staff are getting more and more power
to be able to potentially do things like
617
00:35:45,780 -->
00:35:48,910
vibe code their own app and deploy it on
the internet, and If they don't have the
618
00:35:51,187 -->
00:35:51,787
IT staff to help them manage it,
they can, they can at least get
619
00:35:51,787 -->
00:35:53,637
by with some, some simpler stuff.
620
00:35:53,687 -->
00:35:55,367
do you see that happening
on Control Plane too?
621
00:35:55,377 -->
00:35:56,807
It sounds like it is, since your, since
622
00:35:56,809 -->
00:35:57,439
Doron: 100%.
623
00:35:57,449 -->
00:36:01,619
We are, we are releasing a,
more consumer-oriented…
624
00:36:01,719 -->
00:36:02,679
I'll come out and say it.
625
00:36:02,699 -->
00:36:05,199
We are, the, shipping very soon.
626
00:36:05,299 -->
00:36:09,009
We're in the tail end of,
Vers- not Vercel, but, Lovable
627
00:36:09,189 -->
00:36:11,379
and Replit, Base44 killer.
628
00:36:11,479 -->
00:36:15,249
And why we're able to do what they
do, but they're not able to do
629
00:36:15,249 -->
00:36:17,989
what we do, is the coding agents.
630
00:36:18,089 -->
00:36:23,309
when you look at Replit, most of the
effort has been put into, Guardrails
631
00:36:23,529 -->
00:36:29,469
and skills that are just, wrappers,
assisting agents for, for the LLM.
632
00:36:29,569 -->
00:36:33,489
and, and those are now in the LLM
itself most of the, of the time.
633
00:36:33,589 -->
00:36:38,019
The heavy lifting is allowing you to
run on any cloud, letting you scale
634
00:36:38,029 -->
00:36:41,129
using a hundred different scaling
strategies, letting you fail over
635
00:36:41,129 -->
00:36:44,749
automatically, leveraging hundreds
of diff- the operational stuff.
636
00:36:44,849 -->
00:36:48,319
So we've been hard at work building
this operating system for six and a
637
00:36:48,319 -->
00:36:51,879
half years, and it's not something
you can vibe code, absolutely not.
638
00:36:51,979 -->
00:36:58,879
And so we built this operating system,
and for us to layer the, consumer-facing
639
00:36:58,979 -->
00:37:04,409
app builder is a lot easier than
for an, an app builder to create an
640
00:37:04,419 -->
00:37:06,499
operating system for running, anywhere.
641
00:37:06,599 -->
00:37:08,919
Replit is I, I played
with it a little bit.
642
00:37:09,019 -->
00:37:10,739
It's a, it's a good tool.
643
00:37:10,859 -->
00:37:14,489
I don't know that it will let
you scale to Netflix grade.
644
00:37:14,589 -->
00:37:18,629
I, I think what happens is, and I wrote
an article that I'm publishing, but many
645
00:37:18,629 -->
00:37:23,219
people, when they start scaling or need
security or need compliance, they get
646
00:37:23,249 -->
00:37:27,219
off of that vibe coding platform, where
Control Plane takes you all the way to
647
00:37:27,239 -->
00:37:29,839
Netflix grade cloud native maturity.
648
00:37:29,939 -->
00:37:35,186
You get the compliance, the security, the
scalability, the, unbreakable compute.
649
00:37:35,286 -->
00:37:37,696
Bret (2): I would just assume that
a lot of those platforms are not
650
00:37:37,706 -->
00:37:39,876
like, SOC 2, PCI, you know, HIPAA,
651
00:37:41,369 -->
00:37:42,419
Doron: HIPAA, right
652
00:37:42,527 -->
00:37:46,040
Bret (2): all the things that like a dec-
any decent sized company is gonna care
653
00:37:46,040 -->
00:37:49,480
about, and there's probably someone in
marketing vibe coding something on Lovable
654
00:37:49,480 -->
00:37:50,960
that's completely not up to standards.
655
00:37:51,040 -->
00:37:55,210
For me, my experience so far has been one
of the biggest challenges in like the vibe
656
00:37:55,220 -->
00:37:59,680
codings products out there, is they don't
really answer the data problem, right?
657
00:37:59,690 -->
00:38:02,870
Like, 'cause usually m- everyone that
I've talked to so far, like creating
658
00:38:02,870 -->
00:38:05,570
a dashboard is cute, and these are,
these are important things for teams.
659
00:38:05,640 -->
00:38:08,720
but th- th- there needs to
be data on that dashboard.
660
00:38:08,900 -->
00:38:12,680
There needs to be data in that app, and
it's usually existing cust- company data
661
00:38:12,890 -->
00:38:17,780
locked up somewhere, usually not behind
like a perfectly designed open portal
662
00:38:17,950 -->
00:38:23,480
with a proper API across every surface
with, with automatic, token delivery
663
00:38:23,490 -->
00:38:27,050
for people that request it to ins- Like,
there's just this whole infrastructure
664
00:38:27,050 -->
00:38:29,364
of cloud native that a lot of companies
665
00:38:29,364 -->
00:38:30,524
they haven't matured to yet.
666
00:38:30,814 -->
00:38:35,204
And that's to me, like building these vibe
coding platforms for companies, which I
667
00:38:35,214 -->
00:38:37,634
think, yeah, again, like we could probably
do a whole separate podcast on, but
668
00:38:37,634 -->
00:38:43,094
like that world is inherently necessary
that you have a very modern mature
669
00:38:43,094 -->
00:38:44,954
infrastructure before you can do that.
670
00:38:45,054 -->
00:38:50,629
' Cause a lot of it is about, automation
and security and encryption and, l-
671
00:38:50,629 -->
00:38:52,214
like you said, auditing and compliance.
672
00:38:52,214 -->
00:38:56,584
That's, that's where I live, and that's
where the vibe coded app, while it may
673
00:38:56,584 -->
00:39:01,154
not be stable or highly redundant, it has
to… It, it… Those things are almost
674
00:39:01,154 -->
00:39:03,254
like optional, maybe someday things.
675
00:39:03,254 -->
00:39:05,834
But what's not usually optional
is for this thing to be useful, I
676
00:39:05,834 -->
00:39:08,684
need to access that SQL database,
and how do I get that, right?
677
00:39:08,694 -->
00:39:10,564
How do I ensure that
that data doesn't leak?
678
00:39:10,564 -->
00:39:12,684
How do I make sure that
this is behind the firewall?
679
00:39:12,684 -->
00:39:15,744
Like, there's all those levels
of complexity that I don't see
680
00:39:15,744 -->
00:39:18,724
the Replitz and the Lovables of
the world ever truly tackling.
681
00:39:18,724 -->
00:39:21,274
So I think you're right that this is
probably… These, these solutions
682
00:39:21,274 -->
00:39:25,064
are probably gonna come more from
infrastructure companies like yourself
683
00:39:25,074 -->
00:39:31,057
rather than a, a startup that's focusing
on AI as a, as a sort of, magic button.
684
00:39:31,157 -->
00:39:33,037
So, I'm, I'm interested to get into that.
685
00:39:33,037 -->
00:39:36,197
So I know you brought some demos,
and we could talk about this product
686
00:39:36,217 -->
00:39:39,487
forever, but I, I feel like we've,
we've covered some of the questions
687
00:39:39,487 -->
00:39:41,077
that I wanted to, to know about.
688
00:39:41,177 -->
00:39:43,077
l- l- what are we gonna, what
are we gonna look at today?
689
00:39:43,177 -->
00:39:47,257
Doron: So I want to show you, I, I want
to do a tour around the platform, and
690
00:39:47,257 -->
00:39:52,577
I'll give a sneak peek at some of the
power you get right out of the box, and
691
00:39:52,677 -->
00:39:57,577
traverse the network boundaries, allow
you to run an app anywhere, even on-prem,
692
00:39:57,677 -->
00:40:01,777
that consumes any service of any cloud,
show you the observability, show you
693
00:40:01,777 -->
00:40:05,947
the audit trail, show you the Terraform,
capabilities and, and stuff like that.
694
00:40:05,947 -->
00:40:09,497
So just give you a general tour
so people know what exists.
695
00:40:09,527 -->
00:40:15,157
And, most people don't know that they
have at their fingertips so much power,
696
00:40:15,257 -->
00:40:17,227
and they keep struggling with Amazon.
697
00:40:17,327 -->
00:40:21,877
And, w-where you can get a hundred
x the productivity by leveraging,
698
00:40:21,917 -->
00:40:23,817
a, a platform like Control Plane.
699
00:40:23,917 -->
00:40:28,327
and not only that, save a bundle
because you're not running full EC2
700
00:40:28,367 -->
00:40:30,597
instances, you're running the millicores.
701
00:40:30,897 -->
00:40:35,077
So you can literally run your Bret
app for six-- five and a half cents a
702
00:40:35,077 -->
00:40:39,817
day, a day meaning twenty-four hours,
literally less than six cents a day.
703
00:40:39,917 -->
00:40:43,787
and yet people are spending money
on NAT gateways, load balancers and,
704
00:40:43,897 -->
00:40:47,247
Kubernetes nodes and pay hundreds
of dollars per month when they can
705
00:40:47,247 -->
00:40:48,857
be paying, cents on the dollar.
706
00:40:48,957 -->
00:40:50,007
And that's what I want to show.
707
00:40:50,107 -->
00:40:51,157
Boom questions?
708
00:40:51,257 -->
00:40:52,487
Bret (2): no, this is, this is a lot.
709
00:40:52,587 -->
00:40:56,077
That's, I'm trying to think of my
workflows and, like, how my workflows
710
00:40:56,087 -->
00:41:00,267
for creating infrastructure and managing
infrastructure need to adapt to this
711
00:41:00,267 -->
00:41:02,667
model, 'cause it's, it's flexible enough.
712
00:41:02,667 -->
00:41:04,057
you have click ops, right?
713
00:41:04,117 -->
00:41:09,827
but it seems so easy to hop back and forth
between the click ops, 'cause it feels
714
00:41:09,827 -->
00:41:14,777
like the click ops is backed because you
have that versioning and the audit trail.
715
00:41:14,877 -->
00:41:18,817
you don't, it doesn't, it looks like
you've removed the risk of the, sort of
716
00:41:18,817 -->
00:41:23,577
the rogue click that now is no longer
compatible or, or can- isn't tracked in
717
00:41:23,577 -->
00:41:25,827
a way that you can update your Terraform.
718
00:41:25,827 -->
00:41:28,707
this is the classic problem of, we're
git- we're trying to do GitOps, and
719
00:41:28,707 -->
00:41:33,017
we're putting all of our infrastructure
in, in Git, and we're storing changes
720
00:41:33,017 -->
00:41:35,767
there, and we've maybe got some
automation for Terraform plans and
721
00:41:35,767 -->
00:41:38,697
applies, or we've got, we've got some
stuff going on there, and that's the
722
00:41:38,697 -->
00:41:39,647
only way we're supposed to do it.
723
00:41:39,647 -->
00:41:43,597
And then we get some person that shows
up and in an emergency or because they
724
00:41:43,597 -->
00:41:48,177
didn't know or whatever, they go into
the AWS console and make a change,
725
00:41:48,457 -->
00:41:54,387
and now I'm bifurcated, and I have
no path to success because AWS isn't
726
00:41:54,387 -->
00:41:58,087
gonna really easily export, it's not
gonna to export Terraform for me.
727
00:41:58,357 -->
00:42:02,787
And so I have to figure out what they
changed before, I risk blowing their
728
00:42:02,787 -->
00:42:05,127
stuff away with my future Terraform apply.
729
00:42:05,227 -->
00:42:07,837
And so we, we've always had this
challenge in infrastructure of,
730
00:42:07,887 -->
00:42:09,347
sometimes we do need the click ops.
731
00:42:09,347 -->
00:42:12,997
Sometimes we ne- do need to be able to go
into a GUI really quick, troub- especially
732
00:42:12,997 -->
00:42:15,377
when you're, like, troubleshooting or
just trying to recover from outage.
733
00:42:15,637 -->
00:42:19,477
You don't wanna stare at a bunch
of l- code in TOML forever.
734
00:42:19,777 -->
00:42:23,357
So you're, you're usually in the, in
the dis- in this GUI app trying to
735
00:42:23,357 -->
00:42:26,447
figure things out, but you also know
that y- it's, you can't touch it.
736
00:42:26,647 -->
00:42:29,247
It has to be read-only because you
need to make your changes somewhere
737
00:42:29,247 -->
00:42:30,357
else, and you're so nervous.
738
00:42:30,357 -->
00:42:32,977
So it feels like you've kinda
thought about that a little bit.
739
00:42:33,177 -->
00:42:35,577
Doron: Yeah, and everything,
like I said, everything you
740
00:42:35,577 -->
00:42:38,587
can get as, YAML or, or JSON.
741
00:42:38,687 -->
00:42:42,547
But here is our app running, and
so we give you a more consumer-y
742
00:42:42,567 -->
00:42:44,057
interface to this thing.
743
00:42:44,157 -->
00:42:47,999
but the bottom line, it's really backed
by the control plane API, and the control
744
00:42:47,999 -->
00:42:54,809
plane API lets you run anywhere, and with,
with determinism, and, and it's the same.
745
00:42:54,809 -->
00:42:57,699
You c-- Again, you can run it on
your Raspberry Pi or Dell machine.
746
00:42:57,799 -->
00:43:01,619
And the cloud wormhole, which I
haven't demonstrated, allows you to
747
00:43:01,619 -->
00:43:07,149
very, very easily connect to an RDS
instance in a VPC or connect from
748
00:43:07,149 -->
00:43:11,439
a, the seven different locations
here to my own Redis on my machine.
749
00:43:11,539 -->
00:43:14,829
And we have a service catalog
where you can run above the cloud.
750
00:43:14,879 -->
00:43:19,379
If you want to run CockroachDB or
Cassandra or really any open source
751
00:43:19,389 -->
00:43:22,669
project, we certify these s-services.
752
00:43:22,769 -->
00:43:27,199
Like let's say I want to run Hermes
Agent or Langfuse or Keycloak or, you
753
00:43:27,249 -->
00:43:32,309
name it, Postgres, Mongo You can run
them at the cost of the infrastructure
754
00:43:32,409 -->
00:43:36,519
instead of paying a premium to RDS.
755
00:43:36,539 -->
00:43:39,719
You pay double what the compute
is, but Control Plane lets you
756
00:43:39,719 -->
00:43:43,689
do backups, lets you do leaders
election, lets you do multi-master
757
00:43:43,689 -->
00:43:45,669
replication with different add-ons.
758
00:43:45,769 -->
00:43:48,839
You can run Ollama, you can
run whatever you want to run.
759
00:43:48,939 -->
00:43:54,571
and you can conversationally say, "Hey,
get me an N8n, run me a Hermes agent,
760
00:43:54,581 -->
00:44:00,248
run me, know, OpenBaO or run me whatever,
Postgres very, very easily." you,
761
00:44:00,288 -->
00:44:02,038
you can do all of, all of that stuff.
762
00:44:02,038 -->
00:44:03,998
And again, you can run anywhere.
763
00:44:04,098 -->
00:44:04,458
Bret (2): Yeah
764
00:44:04,510 -->
00:44:06,700
Doron: w-we're not going to do
an exhaustive demonstration.
765
00:44:06,700 -->
00:44:10,340
one of the things that, that people
care about is security, and you
766
00:44:10,340 -->
00:44:14,940
can set very granular policies of
who is allowed to do exactly what.
767
00:44:14,950 -->
00:44:19,360
You can say, this agent is allowed only
to push image but not delete an image.
768
00:44:19,460 -->
00:44:23,600
This Lisa can only, read a
certain value but not write.
769
00:44:23,700 -->
00:44:26,090
Secrets are only available to X and Y.
770
00:44:26,190 -->
00:44:28,360
So everything is very, very granular.
771
00:44:28,460 -->
00:44:31,170
and th-there is a lot, a lot more.
772
00:44:31,270 -->
00:44:32,850
You can set custom domains.
773
00:44:32,950 -->
00:44:35,960
we have customers with twenty
thousand different domains, and
774
00:44:35,960 -->
00:44:38,300
they run perfectly on Control Plane.
775
00:44:38,400 -->
00:44:44,200
It gives you the confidence to operate
above the clouds without being bogged down
776
00:44:44,300 -->
00:44:46,770
by the complexity of any particular cloud.
777
00:44:46,770 -->
00:44:49,380
And the thing I wanted to show,
because we have this thing called
778
00:44:49,420 -->
00:44:52,273
Capacity AI That is by default on.
779
00:44:52,373 -->
00:44:57,463
If you look at the spend, you're spending
twenty-five millicores 'cause the system
780
00:44:57,463 -->
00:45:01,953
learned over time that Bret video game
only needs twenty-five one-thousandth
781
00:45:01,953 -->
00:45:04,183
of a CPU and thirty-two megs of RAM.
782
00:45:04,283 -->
00:45:08,553
If you go to the control plane pricing
and plug in-- And you can bring your
783
00:45:08,563 -->
00:45:12,523
own compute, or you can use control
plane compute, or you can do both.
784
00:45:12,623 -->
00:45:16,253
So twenty-five millicores and
thirty-two megs of RAM cost you a
785
00:45:16,253 -->
00:45:18,173
dollar seventy-eight per month, period.
786
00:45:18,183 -->
00:45:20,893
You don't pay for load balancers,
you don't pay for nodes,
787
00:45:20,893 -->
00:45:21,893
you don't pay for anything.
788
00:45:21,913 -->
00:45:22,933
That's all you pay.
789
00:45:23,033 -->
00:45:24,353
We charge you by the second.
790
00:45:24,453 -->
00:45:28,243
let's say a full hour would cost
you, No, sorry, a full day would
791
00:45:28,243 -->
00:45:30,503
cost you less than six cents.
792
00:45:30,603 -->
00:45:35,493
So if you ran the Bret app for a
full twenty-four hours in a location,
793
00:45:35,593 -->
00:45:37,363
it would cost you six cents.
794
00:45:37,363 -->
00:45:41,413
That's why I'm saying sometimes running
it in more locations is even cheaper
795
00:45:41,513 -->
00:45:45,003
than running it in a single location,
'cause if you were to stand up a EKS
796
00:45:45,033 -->
00:45:49,003
cluster, you would pay hundreds of
dollars in a month to heat up the planet
797
00:45:49,043 -->
00:45:50,843
Earth or to make Jeff Bezos richer.
798
00:45:50,943 -->
00:45:55,743
We let you pay for what you eat,
not for heating up the planet Earth.
799
00:45:55,843 -->
00:45:57,463
So extremely cost-effective
800
00:45:57,563 -->
00:45:59,713
Bret (2): In that, in the,
the workflows you were showing
801
00:45:59,905 -->
00:46:02,882
as you would, as you were doing this
stuff, I w- guess one of my questions is
802
00:46:02,932 -->
00:46:07,972
Deployment configurations and design can
often be, hard for teams, especially when
803
00:46:07,972 -->
00:46:09,542
they start to adopt Kubernetes, right?
804
00:46:09,542 -->
00:46:13,742
Because then they have to pick Flux
or Argo, and they have to decide on a
805
00:46:13,872 -->
00:46:17,682
particular way to deploy that, and then
they have to configure it all correctly,
806
00:46:17,682 -->
00:46:20,756
and then when they have b- a bunch of
different apps, yeah, so my premise was,
807
00:46:20,816 -->
00:46:25,286
like one of the, one of the challenges
that I, I train people on and is a lar-
808
00:46:25,336 -->
00:46:28,776
a d- decent amount of my consulting
over the last six years has been once
809
00:46:28,786 -->
00:46:32,706
teams adopt Kubernetes, there becomes
this deployment complexity because it's
810
00:46:32,716 -->
00:46:34,376
not enough to just have Kubernetes.
811
00:46:34,376 -->
00:46:38,266
Then, then you need, your code's stored
in a place, GitHub, GitLab, whatever, and
812
00:46:38,266 -->
00:46:39,716
then you gotta connect all that together.
813
00:46:40,006 -->
00:46:44,716
And, in the old days before containers,
a- and even when we first, first
814
00:46:44,716 -->
00:46:48,016
started using containers, it was very
much "I'm gonna write a CI job that
815
00:46:48,016 -->
00:46:51,866
runs kubectl apply," or, "I'm gonna
run a Docker run, and, and it's gonna
816
00:46:51,916 -->
00:46:53,136
connect to the server from my CI."
817
00:46:53,413 -->
00:46:55,623
And then we grew up, and we
realized we maybe need a more
818
00:46:55,693 -->
00:46:59,553
deterministic methodology, and then
we started adopting Argo and Flux.
819
00:46:59,653 -->
00:47:03,553
And, and so for the longest time,
I've been f- I've been advocating
820
00:47:03,553 -->
00:47:04,713
and teaching those technologies.
821
00:47:04,713 -->
00:47:07,823
We call it GitOps, but it doesn't
necessarily have to be labeled that.
822
00:47:07,903 -->
00:47:11,453
and I'm curious how you, how
your platform works with that.
823
00:47:11,463 -->
00:47:14,923
Does that m- It, it… You've got a
lot of, glue there that feels like
824
00:47:14,923 -->
00:47:18,563
it makes a lot of these rough edges
streamlined and it's almost like you're
825
00:47:18,563 -->
00:47:21,683
applying opinionated defaults, which I
love because I'm a, I'm a very much a,
826
00:47:21,953 -->
00:47:23,553
an opinionated defaults kind of guy.
827
00:47:23,653 -->
00:47:28,483
when it comes to like the code in
the repos, and is it as easy as
828
00:47:28,483 -->
00:47:31,973
someone on your platform pointing
to repos, and then it figures out
829
00:47:31,983 -->
00:47:33,543
how to get it into Kubernetes?
830
00:47:33,583 -->
00:47:36,193
do they need to even worry
about Argo at this point?
831
00:47:36,203 -->
00:47:39,293
'Cause you've been saying things like
deploy this app, and it's doing it.
832
00:47:39,323 -->
00:47:40,383
I don't know how it's doing it.
833
00:47:40,383 -->
00:47:41,823
I know it's, I know it's
running on Kubernetes.
834
00:47:42,063 -->
00:47:44,753
But what's that, that deployment
look like for Control Plane?
835
00:47:44,753 -->
00:47:45,613
Is it automating it?
836
00:47:45,633 -->
00:47:47,423
Is it, is it doing Argo for you?
837
00:47:47,523 -->
00:47:49,343
Doron: So it c-- you can use Argo.
838
00:47:49,913 -->
00:47:55,206
So we have, a CRD, a control plane
CRD, so you can have a cluster and,
839
00:47:55,306 -->
00:47:58,196
use Argo or Flux very naturally.
840
00:47:58,296 -->
00:48:02,226
the way it works, so you said, "I'm not
sure how it works," so let me explain.
841
00:48:02,326 -->
00:48:04,466
you get the YAML.
842
00:48:04,496 -->
00:48:05,956
Let me share real quick.
843
00:48:06,056 -->
00:48:07,786
and again, YAML is just one of many.
844
00:48:07,786 -->
00:48:10,636
So if… let's take this
guy for-- as an example.
845
00:48:10,736 -->
00:48:12,646
I can export, this YAML.
846
00:48:12,776 -->
00:48:14,496
So you can see the kind is workload.
847
00:48:14,506 -->
00:48:19,036
It's a new kind that doesn't exist on
Kubernetes, and it has the different,
848
00:48:19,056 -->
00:48:23,586
like what's the memory, what's the
scaling strategy, what GVC does it go on?
849
00:48:23,686 -->
00:48:28,176
And you j-just say CPLN apply
and give it, minus F and give
850
00:48:28,176 -->
00:48:30,013
it the YAML, and it deploys it.
851
00:48:30,113 -->
00:48:33,413
Like I said, we have other ways
to deploy, but at the end of
852
00:48:33,413 -->
00:48:36,133
the day, we have a data service.
853
00:48:36,173 -->
00:48:39,646
We have an-- we have the control
plane API, and you make a call to
854
00:48:39,646 -->
00:48:43,783
the plane itself is, is running
on Amazon, Google, and Azure, and
855
00:48:43,883 -->
00:48:45,643
the-- our data is distributed.
856
00:48:45,743 -->
00:48:51,786
It calls a component we call the which
is a control plane, Go-based application.
857
00:48:51,886 -->
00:48:55,476
It runs on the edge, we
call them edge clusters.
858
00:48:55,486 -->
00:48:59,056
So you can have one cluster, or in
our case, we have one, two, three,
859
00:48:59,056 -->
00:49:01,496
four, five, six, seven clusters.
860
00:49:01,596 -->
00:49:03,320
These are different clusters.
861
00:49:03,420 -->
00:49:05,090
They can have hundreds of nodes.
862
00:49:05,190 -->
00:49:09,850
we provide clusters that are, cloudless,
meaning we provide clusters, or you can
863
00:49:09,850 -->
00:49:13,510
bring your own, you can bring your own
cluster like I did in the Ron office.
864
00:49:13,610 -->
00:49:18,130
I created the Kubernetes brain in the
cloud, but I could have run OpenShift or
865
00:49:18,140 -->
00:49:22,100
EKS or whatever, in the local environment
The-- what you need is a Kubernetes
866
00:49:22,140 -->
00:49:27,070
cluster with the actuator and a few other
components like the, the log, collector,
867
00:49:27,080 -->
00:49:28,550
the metrics collector, and so on.
868
00:49:28,650 -->
00:49:32,450
But we control planify a cluster,
and then it becomes part of
869
00:49:32,450 -->
00:49:33,800
that global virtual cloud.
870
00:49:33,900 -->
00:49:38,500
We use NATs under the covers to have
our, control plane, if you will,
871
00:49:38,660 -->
00:49:44,070
communicate the desired state to all the
clusters, and they all take the desired
872
00:49:44,070 -->
00:49:46,550
state and actuate it in the cluster.
873
00:49:46,650 -->
00:49:51,040
So under the cover, you have, stateful
sets or, deployments or, pods and
874
00:49:51,050 -->
00:49:55,340
all the Kubernetes goodness that is
happening, but it has been abstracted.
875
00:49:55,440 -->
00:49:59,420
And if you want to see the Kubernetes
resources and, and it's a bring your
876
00:49:59,420 -->
00:50:04,650
own Kubernetes like I have here, you
can get the kubeconfig and go edit
877
00:50:04,920 -->
00:50:06,650
or use the dashboard and go edit.
878
00:50:06,750 -->
00:50:10,660
But we… You noticed we didn't
have to issue kubectl apply commands
879
00:50:10,660 -->
00:50:14,940
and all that because there is an
actuator component in this cluster
880
00:50:15,040 -->
00:50:18,680
that went and did the actuation
of the desired state Makes sense.
881
00:50:18,680 -->
00:50:24,390
And also, if we are down, the, the cluster
operates autonomously and serves traffic.
882
00:50:24,640 -->
00:50:29,340
Maybe you can't upgrade the version
of the workload, but A, we're very
883
00:50:29,340 -->
00:50:31,910
highly, redundant and highly available.
884
00:50:32,010 -->
00:50:34,170
so we're very, very unlikely to go down.
885
00:50:34,450 -->
00:50:37,820
But if we were to go down, we are the
control plane, not the data plane.
886
00:50:38,080 -->
00:50:41,780
The data plane is where the data,
flows, and you mentioned that earlier.
887
00:50:41,880 -->
00:50:43,360
so hopefully that gives you kind
888
00:50:43,514 -->
00:50:44,124
Bret (2): the goal, right?
889
00:50:44,124 -->
00:50:47,554
Is like the control plane never,
never takes down the data plane
890
00:50:47,654 -->
00:50:49,034
when it's, when it's unavailable.
891
00:50:49,034 -->
00:50:49,364
Yeah.
892
00:50:49,566 -->
00:50:50,196
Doron: Exactly.
893
00:50:50,246 -->
00:50:53,416
A, we're extremely… Yeah,
and we test everything, like,
894
00:50:53,436 -->
00:50:54,976
like I can't even tell you.
895
00:50:55,076 -->
00:50:59,816
From day one, we created what we call
the Observer Project, and it does
896
00:50:59,826 -->
00:51:04,146
twenty-four seven millions of tests
simulating node failure, multi-node
897
00:51:04,146 -->
00:51:08,726
failure, cluster failure, AZ failure,
region failure, whole cloud failure.
898
00:51:08,826 -->
00:51:10,386
And we do, fuzzing.
899
00:51:10,486 -->
00:51:15,436
We do random failure, introduction of
failure modes with a core objective
900
00:51:15,486 -->
00:51:17,336
of making sure you never go down.
901
00:51:17,436 -->
00:51:20,796
So one of the things you get with
Control Plane is mainframe-grade
902
00:51:20,936 -->
00:51:23,376
availability over commodity infrastructure
903
00:51:23,476 -->
00:51:26,561
Bret (2): All right Yeah, the
fractional CPU, stuff is, is
904
00:51:26,571 -->
00:51:27,931
pretty flexible and pretty awesome.
905
00:51:27,931 -->
00:51:32,101
When you're running small stuff and
you're a small company and every single
906
00:51:32,101 -->
00:51:36,161
workload you add has, this minimum
cost to it, and Even when you try to
907
00:51:36,161 -->
00:51:40,701
run, small instances in an EC2 cluster,
you're limited on how small you can go,
908
00:51:40,721 -->
00:51:46,181
and then you're inherently bottlenecked
there if you ever get any decent traffic
909
00:51:46,201 -->
00:51:48,321
because it, it doesn't know how to…
910
00:51:48,361 -->
00:51:50,541
You, you have to build a lot of
intelligence in in order for it to
911
00:51:50,541 -->
00:51:54,771
auto-scale up and correctly change
the node size, if you need more memory
912
00:51:54,771 -->
00:51:59,171
because, a classic example is, a legacy
app with a PDF export and someone
913
00:51:59,171 -->
00:52:03,481
decides to PDF export a report that isn't
batched properly, so then it consumes
914
00:52:03,481 -->
00:52:06,761
all the memory on the server for one
report, and then the, and then your tiny
915
00:52:06,761 -->
00:52:09,631
little app crashes the whole thing all
because you wanted to print a report.
916
00:52:09,631 -->
00:52:13,101
So there's always, there's always, sizing
challenges because you just want things,
917
00:52:13,111 -->
00:52:17,331
especially when, when it's, a small
business kind of situation where, they,
918
00:52:17,341 -->
00:52:23,531
you tell them, $500 a month for something
that they consider not worth all that is a
919
00:52:23,531 -->
00:52:27,131
hard sell when you're like, "Well, that's
what it takes to be minimum redundant
920
00:52:27,131 -->
00:52:29,041
across regions in this particular cloud."
921
00:52:29,141 -->
00:52:32,531
And the idea that I can say, "Well,
you're, you're only using this
922
00:52:32,531 -->
00:52:35,961
app internally, so we don't even
need a single CPU. We need a half
923
00:52:35,961 -->
00:52:37,131
of one. We need a third of one,"
924
00:52:37,231 -->
00:52:40,951
Doron: Yeah, we need a fif-fiftieth
of one or, one, one-twentieth of one.
925
00:52:41,051 -->
00:52:45,701
Yeah, and that's… And you can now
give people redundant availability for
926
00:52:45,701 -->
00:52:47,361
three dollars, five dollars a month.
927
00:52:47,461 -->
00:52:50,941
you can run an app for five
dollars a month that is-- has the
928
00:52:50,941 -->
00:52:52,951
availability of a real enterprise.
929
00:52:53,261 -->
00:52:54,011
Bret (2): how do people get started?
930
00:52:54,021 -->
00:52:57,151
They go to Control Plane.com
and sign up for an account?
931
00:52:57,251 -->
00:52:57,616
Doron: Yeah.
932
00:52:57,616 -->
00:52:58,971
Yeah, it's very easy.
933
00:52:59,071 -->
00:53:02,901
so they go to Control Plane, they
click sign up, they follow the prompts.
934
00:53:02,941 -->
00:53:06,721
we ask for a credit card, and I
find some people get intimidated.
935
00:53:06,821 -->
00:53:10,331
we're happy to give you,
especially to your listeners,
936
00:53:10,431 -->
00:53:15,751
shoot me an email directly, Doron,
D-O-R-O-N, @ Control Plane.com.
937
00:53:16,101 -->
00:53:19,791
I'll give you, basically a, a
free month to experiment with.
938
00:53:19,791 -->
00:53:23,331
Of course, don't do crypto mining and
don't do crazy stuff, but happy to
939
00:53:23,331 -->
00:53:25,231
let you experiment on the platform.
940
00:53:25,331 -->
00:53:30,741
Even if we were to charge you, you can
run so cost effectively that unless you
941
00:53:30,741 -->
00:53:34,211
reach five dollars in spend, which is a
lot on Control Plane, we don't charge you.
942
00:53:34,211 -->
00:53:37,921
So if you're under five dollars a
month, we don't even charge you 'cause
943
00:53:37,921 -->
00:53:39,531
it's not worth it to send to Stripe.
944
00:53:39,631 -->
00:53:43,411
So many people can run real projects on
Control Plane for less than five bucks.
945
00:53:43,511 -->
00:53:47,331
There is no platform fee, there's no
per seat fee, there's just usage fee,
946
00:53:47,431 -->
00:53:49,051
and it's very, very cost effective.
947
00:53:49,221 -->
00:53:51,001
Bret (2): That's what I used
to use Heroku for, so now m-
948
00:53:51,261 -->
00:53:52,541
now maybe you're my new Heroku
949
00:53:52,641 -->
00:53:54,301
Doron: Think of us as a non-toy Heroku.
950
00:53:54,401 -->
00:53:54,701
Bret (2): Yeah
951
00:53:54,759 -->
00:53:59,479
Doron: the non-toy, replace your, Heroku
with, the, the, the cloud du jour.
952
00:53:59,579 -->
00:54:03,089
But, we give you all the power
of all the clouds, but, with the
953
00:54:03,119 -->
00:54:05,119
ease of use of Heroku or better,
954
00:54:05,219 -->
00:54:06,529
Bret (2): thank you so
much for being here, Doron.
955
00:54:06,629 -->
00:54:07,179
Doron: thank you.
956
00:54:07,189 -->
00:54:08,759
Really enjoyed, the conversation.
957
00:54:08,769 -->
00:54:10,429
You asked great questions.
958
00:54:10,519 -->
00:54:12,029
super happy we connected
959
00:54:12,129 -->
00:54:13,349
Bret (2): And we'll
see you guys next time.
960
00:54:13,449 -->
00:54:13,689
Ciao.
