Sign up to save your podcasts
Or
Share Episode 51: Cloud Native Agility, Reliability and Stability with Weaveworks CTO Cornelia Davis
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
Episode 51: Cloud Native Agility, Reliability and Stability with Weaveworks CTO Cornelia Davis
Intro
Mike Schwartz: Hello and welcome to Open Source Underdogs. I’m your host Mike Schwartz, and this is episode 51, with Cornelia Davis, Chief Technology Officer of Weaveworks and author of the Manning book Cloud Native Patterns.
If you’re like me and your whole business has
been realigned around distributed cloud computing, Cornelia’s book can help you
make sense of it.
Prior to joining Weaveworks, Cornelia was an engineering leader at Pivotal, where she was active developing the Cloud Foundry platform. If you want to learn more about Pivotal, you might remember James Watters was a guest on Episode 40 of Underdogs.
If you’re a fan of the podcast, help us get the word out. The goal is to help startups figure out how to use open-source software as part of their business model. But we can only do that if they know about us. So, take a minute, if you can, to comment on Hacker News, tweet out a link to an episode, or follow me and share one of my posts on LinkedIn.
Cornelia just had so many interesting things to say–I’m sure you’ll be super impressed. So, without further ado, let’s get on with the show.
Origin
Mike Schwartz: Cornelia, thank you so much for joining us today.
Mike Schwartz: At a high level, can you talk about how Weaveworks got started,
and what’s the mission of the company?
Cornelia Davis: Mike, it’s great to be here. Thank you so much for having me.
Cornelia Davis: I’ll go all the way back to the founding of Weaveworks, which has
been already about five years ago, and the initial thing that Weaveworks went
off and did, was they created a networking product, and it was around container
networking. We still have that product, or I should say project, because it’s
an open-source project, it’s called Weave Net.
And this was pre-Kubernetes, so, it was
post-Docker. So, people were moving towards container-based solutions, but then
this challenge of how do you do networking across these various containers that
are running on various hosts, and how do you do what we now refer to as overlay
networks across Kubernetes. These are problems, and it’s a new set of
abstractions. It used to be everything was abstracted around the hosts, now, how
you do networking across this new abstraction of containers. And that was the genesis,
the company.
And if we think about who the founders of the
company were, Alexis Richardson and Matthias Radestock, and they had worked
together before, they were the cofounders of Rabbit, RabbitMQ, which had been
acquired by VMware. And then, VMware had done the Pivotal spin out, and both of
them went over to Pivotal, so they both were working on Cloud Foundry at the
time.
Now, Cloud Foundry has been container-based from
the beginning, and so we needed to solve those problems around networking
across containers and how do you do that, and that was kind of the genesis of
them going off in creating Weaveworks. And it had a different name, I wasn’t
with him at the time, but that was the genesis of Weaveworks.
Now, I’m going to fast-forward a lot to tell you a little bit more about who we are today, because we are not a networking company. We did some things with Weave Net, and Kubernetes came along, and we started kind of branching out into more than just networking for containers, but we’ve got this — there’s other problems that need to be solved around containers. And then, Kubernetes came along, and how do you manage that platform. And so, we started moving into the space of a set of services that assist kind of in this container-based platform space.
And our first product was a product that we call Weave Cloud, which is a SaaS product that allows a customer to effectively sign up for that service, which drops an agent on their Kubernetes cluster, and then it gives them services like observability, and uses things like Prometheus and Grafana to provide those services to organizations.
So, that was kind of Step 2. And then, we were
of course running this SaaS product, which was running itself on Kubernetes.
So, we had developed a whole bunch of skills, if you will, SRE or CRE skills,
around managing this Kubernetes platform in this application on top of Kubernetes.
And you can find Alexis telling the story, we had a production outage, and our
meantime recovery was crazy fast because of some of the things that we had put
in place. And that was the birth of GitOps.
It was this new set of operational practices that we were employing to run our own platform,
that allowed us to have a very short meantime to recovery when we had an outage on Weave Cloud. And that was when we had the light bulb moment that takes us to where we are today, which is, oh my gosh there’s a set of tools that support modern cloud-native operational practices, the term we use for that is GitOps, and that’s where the company is focused now.
Why Join Weaveworks?
Mike Schwartz: Cornelia, you worked with James Watters, who
is actually a guest on episode 40th of Underdogs. And obviously,
Pivotal is a great company. What was it about Weaveworks that made you want to
join their team?
Cornelia Davis: There was a great story there as well. I consider James — just working with him in the last seven years, or so, at Pivotal, and I went to work FOR James — in fact, the way that I went to work for James was that I had started, I was in the EMC/CTO office, and I was playing with Cloud Foundry, because I worked on emerging tech. And so, that was an emerging tech space that I was working with, and so, I had started learning about Cloud Foundry, and I had started blogging about it. I’d started blogging about Bosch, which is one of the components under the covers.
And then, I met a few people on the Cloud Foundry team, and I started becoming interested in Pivotal. EMC was spinning Pivotal off EMC and VMware, and I started becoming interested in joining that team in particular, as a part of the spinoff. And they introduced this person that I had met, Elizabeth Hendrix introduced me to James Watters, and I walked into his office, and he said, “Wait a minute. Are you the woman who’s been blogging about Bosch?” And I said, “That’s – guilty!” And so, that was how we met, and we’ve been working together for some time. And it has been just a complete delight, and he’s a good friend as well.
But how I ended up at Weave, it’s definitely connected. So, James and I, James was at VMware before I wasn’t. I started working on Cloud Foundry shortly before the spin-off, but we were both there in the early days of Cloud Foundry. Cloud Foundry was this platform, this application CloudNative application platform that was highly opinionated. And those strong opinions brought to Enterprise customers, and if you found an Enterprise customer, who was willing to absorb those strong opinions, their outcomes were PHE-NO-ME-NAL, absolutely phenomenal.
Now, for some Enterprises, those opinions were
too strong. And Cloud Foundry by design was a platform that was designed with
these strong — you know, I used to say we’re unapologetically opinionated in
our platform. And, again, that really yielded fantastic outcomes.
And then, what happened is that Kubernetes came
along. And Kubernetes was a tool kit, if you will, I considered Kubernetes a toolkit
for building platforms. And famous people like Kelsey Hightower said it’s a platform
for building platforms. And it really is. It is a toolkit for building
platforms. And while I was still at Pivotal, we kind of toyed with this idea of,
well, what people really want is an opinionated Kubernetes platform. And while
I’m not sure that everybody at VMware and Pivotal would agree with this, my two
and a half years of working on Kubernetes at Pivotal, I started forming this
opinion that an opinionated Kubernetes isn’t what customers wanted, they wanted
their own flavor of Kubernetes.
And that is what has brought me to Weaveworks. It is because what we do is, we are really embracing – we’re building out a certain set of tools that will help the platform teams build out the platforms that their developers need, instilling the opinions that their developers need or that their company – sometimes those opinions are not developer opinions, they are compliance and regulatory and security opinions. But it’s more of let the platform teams at these various Enterprises build THEIR platform. And I see that the way that Weaveworks is embracing that and providing that capability to enterprises through declarative configuration, multiple reconciliation loops – we will talk a little bit more about those things – that was what excited me about it. And that’s why I’m at Weaveworks now.
Value Prop
Mike Schwartz: From a marketing
perspective, I don’t envy the marketing team who has to consolidate this group
of products and services into one understandable business proposition. What do
you think are the most important value propositions for Weawork customers?
Cornelia Davis: One of my favorite bits of work, and this isn’t my work but it
is such good work in the industry, is the work that the DevOps research
assessment group has done. So, DORA – I’m talking about Nicole Forsgren, Gene
Kim, Jez Humble. And what they’ve done is, they’ve done this work – I’m sure
many of your listeners, and you’re already familiar with it – they did the work
to tie specific IT practices to business outcomes. So, high performing
organizations, those that are gaining market share that are profitable and so
on, tend to do certain things in their IT practices a certain way. They release
software more frequently, they have lower meantime to recovery, they have lower
change failure rates, and so on.
So, the business proposition here is exactly
that – it is to enable your application teams to achieve those metrics. Because,
again, DORA – and your listeners might also, if they don’t know DORA, many of
them probably do know the state of the DevOps report – there they’ve done the
work to prove that these IT practices really correlate to business strength, either
negatively or positively. So, that’s the fundamental thing.
What we’ve been doing is, as an industry, we’ve
gotten pretty mature on understanding how the cloud-native architectural
patterns and software support those things, where we are still developing is
cloud-native operational practices, not software design but operational
practices. And that’s really the value prop is cloud-native operational
practices that are going to contribute to these IT practices which have a correlation
to business outcomes.
How Do Customers Find Weaveworks?
Mike Schwartz: Would you say that customers are finding your product sort of through the operations
group, or is it more common that maybe the IT leadership of the CIO/CTO office is
saying, “Okay, we have to figure out how to make this more scalable?”
Cornelia Davis: So far, a great deal of our traction has come from our presence
in the open-source community. And you talked about this set of different
projects and challenge that marketing faces, we have a lot of things going on
in the open source, that are all related. But you have to read the tea leaves a
little bit, I think that’s what you were getting to earlier. And so, we have
some very successful open-source projects. And to date, I would say that a good
portion of our kind of entrance into commercial things has come from somebody
who has been playing with Flux, which is one of the components, one of the kind
of core GitOps components that we have out in the open-source community. Or
they want to do canary style deployments because that is something that is
hugely valuable to reducing your change failure rate.
And people like James Governor from RedMonk is
talking a lot about progressive delivery this year. So, there’s a lot happening
there. They are looking at Flagger and so on. And so, there’s been a fair bit
of that. And I also think, to some extent, that we are still kind of the tip of
the spear. I don’t believe that CIOs right now, CIOs in many cases, what
they’re trying to do is “move to the cloud”, but they’re not yet at the level
of detail.
And by the way, these CIOs understand that they
need things like microservice architectures, but they don’t know yet what to
ask for when it comes to cloud-native operations. Yes, they’re talking about
infrastructure as code, which is something that we’ve had for a while, and
they’re starting to look at some of those things. You know, CI, well, that’s
something that’s already quite mature, but when it comes to the very things
that we are selling, if you will – and I don’t necessarily mean just from a
monetary perspective, but that as well – these concepts are still relatively
emerging.
And so, I don’t think the CIOs are asking just yet, it’s either the developers or the platform teams who are saying, “Oh, well, I can actually create my platform and create repeatability and reduce my meantime to recovery because I have this completely non-snowflaked expression of what my platforms look like, which means I can stand one up on, you know, in moments.
How Much Of The Code Is Open Source?
Mike Schwartz: I was looking on
your website, and there’s eight open-source projects that are listed. I bet you
there’s even a couple more that maybe aren’t listed. In rough terms, what
percentage of your developers are maybe of your engineering team is working on
these projects?
Cornelia Davis: That is a great question. And in fact, that’s one of the things
that I’m working with our VP of engineering, David Turner, to kind of rejigger,
if you will, that. Well, if we look at the number of places that code is
committed, I would say that 80% of the code that we commit into a git
repository goes into one of those open-source projects. Even Enterprise
product, Weave Kubernetes platform, which I’m sure we’ll talk about a little
bit more, a good portion of the Weave Kubernetes platform is based on a project
called WKSCTL, which is an open-source project.
So, even that is pretty significant. We have
another open-source project that we partner with on Amazon, which is EKSCTL, which
is the CLI, it is the UX4EKS that provides a higher level set of abstractions
than the base APIs for EKS does. We have Flux, which is kind of our flagship GitOps
project. We have a developer experience team that builds a number of getting
open-source projects as well. So, again, I would say that 80% of the code that
we commit, gets committed to those repositories.
Now, in terms of the teams working on the commercial product, we have people working on a commercial product that could make quite a bit into the open-source projects. Yes, we have some private repositories as well.
Balancing R&D Investment In Open Source
Mike Schwartz: Lot of open-source
companies struggle with, well, how do I justify investing in open source versus,
let’s say, the cloud platform or the Enterprise, if there’s an Enterprise
product. How do you make that balance between devoting resources?
Cornelia Davis: I’ll say that part of the answer here is that is seated in our
bias towards community and open-source. The founders, everybody on the
leadership team, I daresay just about everybody in the company really feels
this draw to open-source and this commitment to a broader industry. I don’t consider
Liz Rice – I consider her my colleague, she works at Aqua, she doesn’t work at
Weaveworks, but I consider her my colleague. I even consider everybody at Pivotal
still my colleagues.
And so, I think that we have this kind of
philosophy, if you will, that, we’re community first or industry first. And so,
to some extent, it’s seated in that. I used to joke around that my
father-in-law years ago always said, “If I ever win the lottery, I’m going to
share my winnings with all my kids.” And so, I used to say, “Well, when he wins
the lottery – and he has that luck, that’s why I say when – that I’m going to
quit my job and work on open source.” I said that 15 years ago, and now, I am
gainfully employed and I work on open source, which is great, it’s been a fantastic
shift. So, this bias that I just
described around, you know, we have this feel to connect with community is not
unique to us or to me personally.
So, that’s kind of, we have that bias towards
this, but again, we, of course, in order for them to pay me my salary so that I
can work on open source, we have to make money. And one of the areas that we’re
super committed to is that we do not want to have Flux for example, which is
this reconciler, this continuous delivery reconciler, we don’t want to have Flux
features that are open-source and Flux features that are closed-source. So, we
believe that everybody should have the ability to do the basic capabilities of
these GitOps workflows and these cloud-native operational ways of working.
And what we want to do is we want to help
Enterprises, and Enterprises are our target market here. We want to help them
do that at scale. That’s where we decide. So, we spend a lot of time, we have
this kind of view of what goes into the commercial product is where you’re not
applying these things to one or two projects or 10 projects, but you’re
applying these to an organization that has 5,000 developers and 3,000
applications. And that’s where we’re helping organizations do those things at
scale.
And so, having that kind of as a overall governor of how we decide what goes where, it maybe doesn’t help us decide how much effort to put in places, but it certainly helps us when we’re thinking about a capability, and we’re talking to a customer who says, “I need this.”, we’d look at it and say, “Is it a scale problem, or is it something that we really want to enable the entire community to work on?”
Revenues: SaaS V. Self-Hosted?
Mike Schwartz: Switching into the products space a little bit. So,
Weaveworks has a product in the, let’s say self-hosted space called Weave
Kubernetes platform, and there’s also Weave Cloud – which of these projects or
products are actually more important to you from a revenue perspective?
Cornelia Davis: I’ll answer that question in two ways. I’ll answer the question you’ve asked, which is, from a revenue perspective, but I’ll also answer from more of a strategic perspective. Weave Cloud platform, which is the one that I talked about earlier, which is the SaaS offering that allows somebody to say, “I’ve got my own Kubernetes cluster, and I want a set of services that are going to allow me to manage that Kubernetes cluster and the workloads that are on it in a better way.” So, around observability, metrics, logging and those types of things.
Again, this is before my time, but I think that
at the time that we created the Weave Cloud, we felt like there was this need
to understand how to help people operate both the platform and the workloads
running on the platform. And we have gotten a certain amount of traction but
it’s kind of the small to medium business type of thing, or it might be people
that are into Enterprises, but they’re still operating at that smaller-scale
that I talked about.
So, it’s still a project, where somebody says,
“Oh, well, I’m working on a product. I’m within a larger organization, but I’m
solving this particular problem just for myself, just for my project. And
that’s where we’ve gotten the traction. And that’s generated a modest amount of
revenue. But really back to what I was saying earlier, is that what we really
want to do, the value that we want to bring to Enterprises, is to help them
solve those problems at scale. And so, while the Weave Kubernetes platform is a
newer platform, I’m not at liberty to talk about where we sit with revenue
today. We absolutely anticipate that the Weave Kubernetes platform, that’s what
we’re banking on, that is where we are going to see the revenue opportunity.
And so, therefore, it’s also kind of our strategic platform moving forward.
Now, Weave Cloud folks, don’t worry, it’s not
that we are going to ditch Weave Cloud. I think that Weave Cloud is still one
of the best places where you can start to experiment with some of the things at
that lower level of scale, to start to get some of those capabilities. Because,
frankly, what we do in the Weave Kubernetes platform, which is the self-hosted,
that you talked about, and I like how you said it self-hosted, it’s not
necessarily on-premise self-hosted, and it might be self-hosted in the cloud,
or it might be self-hosted in your own data center, but it is self-hosted. So,
while WKP is strategically, and from a revenue perspective, definitely the
direction that we’re going, Weave Cloud folks, don’t worry, we still think that
Weave Cloud is a great place for you to start to, you know, get your legs under
you and solve some of these problems maybe at the lower scale.
And so, we see both Weave Cloud as a great entry point to understand the services that we offer as open-source is as well, I described earlier how using the open-source projects is also another way for you to understand the set of services that we offer, on your road to solving these problems at scale, which is where the WKP platform comes in.
Is K8 A New OS Platform For The Cloud?
Mike Schwartz: Okay, so, I’m not
a cloud native expert here, so, you maybe have to cut me a little slack on this
analogy, but would you say that perhaps the Weave Kubernetes platform is sort
of like a Linux distribution, where, for example, there are a lot of ways to
deploy Linux to there’s room for different distributions? Could you say the
same for Kubernetes? That maybe people can think of Kubernetes as a new type of
cloud operating system?
Cornelia Davis: Oh, absolutely. And I think that the industry is a whole thing’s
of Kubernetes is a new Cloud operating system, but your analogy is actually
more — it goes beyond that, it’s deeper than that. Because you said that it is
a way of managing perhaps a number of different distributions of Kubernetes. And
you nailed something that is really, really subtle there. And that is that we
are not aiming with WKP to be the Kubernetes distribution.
While we do have customers who are using WKP as
their Kubernetes distribution, what we are really doing is, we are providing a
platform that allows you to manage whatever Kubernetes distribution you want. Now,
I’m not going to suggest that you can just come along and say, “Hey, I’ve got
my own bespoke fork of Kubernetes and all this stuff in it.” And we can just
say, “Oh, just point us to that repository, and we can go ahead and install it.
We’re not there, but we do support in WKP a number of different flavors of Kubernetes
under the covers.” And, you know, for example, we do, out-of-the-box, provide
you the mechanics to support upstream Kubernetes. You know, the releases that
are put out on the Kubernetes in the GitHub repository under the releases tab.
We also support EKS in there, and that list will continue to evolve. And we are working on ways of bringing other Kubernetes releases in there. And WKP also allows you to say, “You know what? We’re actually going to treat the Kubernetes thing out of band.” Because once you get that Kubernetes dial tone, if you will, there’s still an awful lot that you need to do on top of that to configure Kubernetes, to configure it against the storage systems, to configure it against your networking and your ingress controllers, and to set up the access controls and the pod security policies and all of those things, install the services that you use on top, very services that I talked about on Weave Cloud, like monitoring, and logging, and observability, and all of those things. And, by the way, Flux, which is the GitOps capabilities for your application teams, we are very focused on helping you manage all of that complexity on top of Kubernetes as well.
Partnerships
Mike Schwartz: You’ve mentioned a
number of companies in adjacent markets and Amazon for example – can you talk a
little bit about the partnerships that have been most important and that you
see as being most strategic for Weaveworks going forward?
Cornelia Davis: So, Amazon of course is a close partner of ours, it’s really very
interesting – and again, I always come at it from a little bit of a technical
angle – but I’ll go back to my days at Pivotal. So, I worked on Pivotal Container
service, which was Pivotal ‘s Kubernetes distribution we were working with
VMware, bringing that to market. And when I was out talking with customers,
they were like, “Well, okay, how this compares to…”, and one of the things that
was often asked was, “Tell me how this compares to EKS.”
And one of the things that, at the time, I
talked about was how EKS was this — it wasn’t at all a turnkey Kubernetes user
experience. You couldn’t go to EKS and say, “Hey, give me a cluster, give me a
cluster with three masters and 10 workers.” And it would make it so. It was
much lower-level protocols.
And so, our partnership with Amazon really
launched when Weaveworks said, “Okay, we believe that we can use these
practices that we’re starting to become experts in to provide a better UX.” Or,
not better but easier UX. We will use those low-level building blocks to create
a user experience on top of EKS, that make EKS accessible to a larger market
and a larger set of people. And so, that was the genesis of our relationship
with AWS.
Now, the interesting thing is that I can use AWS
as a bit of an exemplar. Because AWS has been very successful in – well,
massively successful across a huge market, and certainly Enterprises are moving
in that direction. But Enterprises still are doing an awful lot of their own
stuff, their own self-hosted stuff, so they haven’t embraced kind of the whole cloud-native
operational model.
And so, working with companies like AWS and
Google, and we have a very strong partnership with AWS – we also have a relationship
with Google – to help those organizations connect with the Enterprise in a way
that we — you know, our legacy, again, remember that Alexis, Matthias, myself,
we all come from the VMware, EMC, Pivotal Enterprise market. So, that’s been
certainly part of that partnership is to help the cloud providers move in that
direction.
You’ll notice that I haven’t said anything about Microsoft, and I put Microsoft in kind of a little bit of a different category because they also have that Enterprise genesis, if you will. So, they understand the Enterprise, their history is in the Enterprise as well, sure, PCS, and all of that stuff, but they may arguably become very successful in the Enterprise Windows server exchange. You know, they were heavily entrenched in the Enterprise. So, we are very much partnering with Microsoft there as well.
I think the thing that’s given us traction the most in that is that Microsoft, who I have been so impressed with – who hasn’t in the last five years, kind of five to eight years, they’ve totally reinvented themselves – is that they have very early on latched onto this, again, this cloud-native operational model, which is what we represent, and the term we use for it is GitOps.
And so, that is really kind of the core of our partnership there is that all of these cloud service providers are recognizing that were at an inflection point in the industry, where we are really starting to revolutionize the way we do operations. And making that cloud native, SRE was just the beginning of that. There’s a whole set of tools and practices around that.
Customer Perception Of Open Versus Commercial License
Mike Schwartz: You are very much
on the open-source bandwagon, but I’m going to play devil’s advocate a little
bit and say that there are still some negative connotations around open-source
software. And, for example, if there’s a commercial and open-source version of
a certain segment of software, the commercial version is often viewed as better
or more secure or justifying a higher price. And also, when you look in terms
of dollars of the software market, there’s more dollars spent on commercial
software than on open-source software.
Obviously, you’re very pro open source, but do
you think that perception is changing? And is it changing in all the segments
or in certain segments but not others?
Cornelia Davis: Well, I think that, overall, the perception of open-source has clearly changed. It dramatically goes back to that story when I used to think I had to quit my job to be able to work on open source. And now, the reason that people work on open source is because there’s a demand for it. And that demand is driven by revenue as well, because most of everybody who’s working on open source is not independently wealthy. They’re getting paid through something, and they get paid through – you know, the traditional open-source business models are, we’re going to provide support over open source. And that’s a great business. And people definitely do well with that. And then, there’s the, “Okay, we’re going to actually provide a different set of capabilities.
And the former of those models is really where it comes to this notion of open source as “free”, free like a free puppy is, is that, yeah, even though you’re not paying for the source code, to be able to operationalize that open-source project, you need a whole lot of skills, you need support. And you might even need a supply chain, a software supply chain to help you deal with the releases that are coming out.
Years ago, I worked with a colleague in the EMC days, who had really spent his whole career in mainframes, and when we first started working on open source together, he was like, “This is never going to work! Things are changing all the time! I had something running on Friday, there’s a new release on Monday, and everything broke.” But, obviously, we’ve built actually business models around kind of being able to deal with that variability.
But I also think it’s completely fair to have a commercial model that says, “We’re providing some capabilities.” I don’t like the open-core model myself, and I wouldn’t consider us having that open core model, which is to say, “Hey, there are some features that you just don’t get unless you buy the commercial product.”
I know it’s a little bit of a slippery slope. You might argue that the model that I described earlier, which is, “Hey, we’re going to help you solve these problems that scale is a little bit of that, but I see it as a little bit different, which is to say, “No, the core capabilities — it’s democratized, everybody can do that. We’re just going to help you solve a different set of problems.” And that’s where we’re going to put our commercial efforts.
So, the other thing that I’ll share is, I’ve spent a lot of time over the last six, seven years with Enterprise customers, and more and more of them, they want to move to open source. Now, sometimes, they believe that by going to open source, it’s all about vendor lock-in for them. And that is tricky. Because as soon as you engage with a vendor for anything, even kind of the tooling that you’re going to use to keep up-to-date with the changes on open-source, at some point, you end up with some level of vendor lock-in, even if it’s just consulting.
So, I’m not sure that that’s the best reason,
but I have found that enterprises, even if they don’t understand exactly how,
and think that it’s about vendor lock-in, they have this really good intuition and
this instinct that open source actually is a better way for them to go.
And, yes, but at the same time, they’re like, “Look, I want my developers delivering value to my customers. I don’t want them doing all the things that you can do for us. And so, we will enter in a commercial agreement with you.”
What Can We Do To Help More Women Join The Open-Source Community?
Mike Schwartz: Putting on your activist hat for a minute, you’ve probably
noticed that the male/female ratio on the podcast hasn’t been that great,
although we’re trying to make it better this year. But I listened to one of
your podcasts, or one of your previous presentations, and you said it’s up to us
to do something about this, to foster a better tech environment that results in
more women in the industry. So, what are some of the most important things we
can do in the open-source business community to affect that change?
Cornelia Davis: So, you asked specifically in open source, and that one is even
trickier. So, by the way, I, first of all, want to congratulate you on. Maybe
over the course of your entire podcast, your ratio hasn’t been that great, but
you’re knocking it out of the park this year, Mike. So, doing really well. And
I really appreciate your efforts there.
And, in a way, what you’re doing is the
exemplar, and to answer that question, whether it’s in the open-source community
or in the industry at large, which is, you just have to pay attention. And we
all have to pay attention, and we all have to make those decisions. So, one of
the things that I regularly ask my male counterparts is that when you are
invited to sit on a panel, please, decline until there is a woman that is
sitting on the panel with you. Do not sit on all-male panels.
So, here’s a very concrete pragmatic thing that
we can do.
Now, open source, my perception is that it is
even trickier there, is that, in a way, open-source – we kind of think of it as
the total democratization of software, all the software is in the open, anybody
can participate, anybody can pick up an issue from an issue’s list and study
that, come up with a solution, submit a pull request, and those types of
things.
But the reality is that the environment, so, one
of the challenges of women in technology is that we are constantly getting
triggered, signals all day long, all day every day, I always say hashtag #alldayeveryday
we’re getting signals that we don’t belong. It could be anything from surprise,
like, “Oh, really, you’re a software engineer?!” You know, those types of things
to even more overt stuff. And, unfortunately, it still does happen.
And so, we tend to – and again, this is a gross
generalization, but I think the studies have shown this that women tend to
maybe hold back a little bit and don’t speak out as much. And that ends up just
perpetuating itself. Because to participate in open source, you really have to
put yourself out there. And it’s harder for women to put themselves out there.
So, I think that – going back to you quoting me
– it’s up to all of us. By us, I mean ALL of us. I mean, men and women to use
— you know, men in this technology industry have a privilege because they
outnumber women so significantly, is, be conscious about that and then lend
your privilege. Look for elevating the voices of women.
If you see somebody who is participating in a Kubernetes SIG, who is on the calls every week, and is just listening in and isn’t speaking up, take a moment to ask her opinion. Also, maybe provide encouragement to, “Hey, you’ve got some great ideas here. You should really issue a pull request against this.” Help turn up the volume for the women who are participating. Because I think the numbers in open source have shown that the numbers are even worse in open source than they are in the industry at large.
Advice For New Open Source Startups?
Mike Schwartz: This is the same question I asked everyone last, which is, do you have any advice for the poor entrepreneurs who are launching a business around an open-source software project?
Cornelia Davis: One of the things, and this is quite personal, is that the first
bit of advice I would say very directly is, trust your gut, trust your intuition.
However, that’s not enough. And just be very aware of the fact that your
intuition is surely influenced by the experience that you have to date, what
you’ve done in your previous jobs, what you’re seeing happening out in the
open-source industry and those types of things, but make sure that you treat
that intuition as a hypothesis. And look for ways to validate that.
Now, I’ve spent my entire career in emerging
tech, and so that validation step is very tricky. Because — and one of my
favorite quotes, and everybody’s heard it, is the Henry Ford quote, which is,
“If I had delivered what people asked for, it would have been faster horses.”,
or something along those lines. I think I just totally botched that, but I think
people get the idea.
So, in many cases, as an entrepreneur, you are really
at the tip of the spear. And so, how do you do that market validation. You want
to – again, another cliché – but you want to go to where the pack is going to
be, not where it is now.
I think intuition is something that is really,
really very good. And then, my experience at Pivotal over the last seven or
eight years is that they really did a fantastic job on getting there incrementally.
So, get lots of feedback, put things out there, find some trusted advisors, find some people who are willing to go on the journey with you. You are not going to want to go into the very well-established Enterprise organization that has done things the same way for the last 50 years. They’re not your partner on this. Look for the partners who are looking to go on the journey with you and co-invent with you. I always think of the people that are customers as people that I’m co-inventing with.
Closing
Mike Schwartz: That’s great
advice. Thank you so much for sharing all of your experiences today, and best
of luck at Weaveworks.
Cornelia Davis: Thank you. It’s been a pleasure to be here.
Mike Schwartz: Big thanks to Radmila Ercegovac from Manning for bringing Cornelia’s book to my attention. And also, thanks to Alexis Richardson, the CEO and Co-Founder at Weaveworks for helping us coordinate.
Audio editing by Ines Cetenji, transcription and episode website
by Marina Andjelkovic, cool graphics by Kemal Bhattacharjee. Music from Brooke
For Free, Chris Zabriskie and Lee Rosevere.
Next episode, I was really honored to get the chance to interview Melissa Di Donato, the CEO of SUSE. Since being spun out of MicroFocus, SUSE is one of the world’s largest independent open-source companies. Don’t miss it. Melissa had a ton of insights into the industry, and how SUSE is positioning itself to provide a leadership role. Until next time, stay safe and thanks for listening.
The post Episode 51: Cloud Native Agility, Reliability and Stability with Weaveworks CTO Cornelia Davis first appeared on Open Source Underdogs.
View all episodes
By Open Source Underdogs
5
1111 ratings
Intro
Mike Schwartz: Hello and welcome to Open Source Underdogs. I’m your host Mike Schwartz, and this is episode 51, with Cornelia Davis, Chief Technology Officer of Weaveworks and author of the Manning book Cloud Native Patterns.
If you’re like me and your whole business has
been realigned around distributed cloud computing, Cornelia’s book can help you
make sense of it.
Prior to joining Weaveworks, Cornelia was an engineering leader at Pivotal, where she was active developing the Cloud Foundry platform. If you want to learn more about Pivotal, you might remember James Watters was a guest on Episode 40 of Underdogs.
If you’re a fan of the podcast, help us get the word out. The goal is to help startups figure out how to use open-source software as part of their business model. But we can only do that if they know about us. So, take a minute, if you can, to comment on Hacker News, tweet out a link to an episode, or follow me and share one of my posts on LinkedIn.
Cornelia just had so many interesting things to say–I’m sure you’ll be super impressed. So, without further ado, let’s get on with the show.
Origin
Mike Schwartz: Cornelia, thank you so much for joining us today.
Mike Schwartz: At a high level, can you talk about how Weaveworks got started,
and what’s the mission of the company?
Cornelia Davis: Mike, it’s great to be here. Thank you so much for having me.
Cornelia Davis: I’ll go all the way back to the founding of Weaveworks, which has
been already about five years ago, and the initial thing that Weaveworks went
off and did, was they created a networking product, and it was around container
networking. We still have that product, or I should say project, because it’s
an open-source project, it’s called Weave Net.
And this was pre-Kubernetes, so, it was
post-Docker. So, people were moving towards container-based solutions, but then
this challenge of how do you do networking across these various containers that
are running on various hosts, and how do you do what we now refer to as overlay
networks across Kubernetes. These are problems, and it’s a new set of
abstractions. It used to be everything was abstracted around the hosts, now, how
you do networking across this new abstraction of containers. And that was the genesis,
the company.
And if we think about who the founders of the
company were, Alexis Richardson and Matthias Radestock, and they had worked
together before, they were the cofounders of Rabbit, RabbitMQ, which had been
acquired by VMware. And then, VMware had done the Pivotal spin out, and both of
them went over to Pivotal, so they both were working on Cloud Foundry at the
time.
Now, Cloud Foundry has been container-based from
the beginning, and so we needed to solve those problems around networking
across containers and how do you do that, and that was kind of the genesis of
them going off in creating Weaveworks. And it had a different name, I wasn’t
with him at the time, but that was the genesis of Weaveworks.
Now, I’m going to fast-forward a lot to tell you a little bit more about who we are today, because we are not a networking company. We did some things with Weave Net, and Kubernetes came along, and we started kind of branching out into more than just networking for containers, but we’ve got this — there’s other problems that need to be solved around containers. And then, Kubernetes came along, and how do you manage that platform. And so, we started moving into the space of a set of services that assist kind of in this container-based platform space.
And our first product was a product that we call Weave Cloud, which is a SaaS product that allows a customer to effectively sign up for that service, which drops an agent on their Kubernetes cluster, and then it gives them services like observability, and uses things like Prometheus and Grafana to provide those services to organizations.
So, that was kind of Step 2. And then, we were
of course running this SaaS product, which was running itself on Kubernetes.
So, we had developed a whole bunch of skills, if you will, SRE or CRE skills,
around managing this Kubernetes platform in this application on top of Kubernetes.
And you can find Alexis telling the story, we had a production outage, and our
meantime recovery was crazy fast because of some of the things that we had put
in place. And that was the birth of GitOps.
It was this new set of operational practices that we were employing to run our own platform,
that allowed us to have a very short meantime to recovery when we had an outage on Weave Cloud. And that was when we had the light bulb moment that takes us to where we are today, which is, oh my gosh there’s a set of tools that support modern cloud-native operational practices, the term we use for that is GitOps, and that’s where the company is focused now.
Why Join Weaveworks?
Mike Schwartz: Cornelia, you worked with James Watters, who
is actually a guest on episode 40th of Underdogs. And obviously,
Pivotal is a great company. What was it about Weaveworks that made you want to
join their team?
Cornelia Davis: There was a great story there as well. I consider James — just working with him in the last seven years, or so, at Pivotal, and I went to work FOR James — in fact, the way that I went to work for James was that I had started, I was in the EMC/CTO office, and I was playing with Cloud Foundry, because I worked on emerging tech. And so, that was an emerging tech space that I was working with, and so, I had started learning about Cloud Foundry, and I had started blogging about it. I’d started blogging about Bosch, which is one of the components under the covers.
And then, I met a few people on the Cloud Foundry team, and I started becoming interested in Pivotal. EMC was spinning Pivotal off EMC and VMware, and I started becoming interested in joining that team in particular, as a part of the spinoff. And they introduced this person that I had met, Elizabeth Hendrix introduced me to James Watters, and I walked into his office, and he said, “Wait a minute. Are you the woman who’s been blogging about Bosch?” And I said, “That’s – guilty!” And so, that was how we met, and we’ve been working together for some time. And it has been just a complete delight, and he’s a good friend as well.
But how I ended up at Weave, it’s definitely connected. So, James and I, James was at VMware before I wasn’t. I started working on Cloud Foundry shortly before the spin-off, but we were both there in the early days of Cloud Foundry. Cloud Foundry was this platform, this application CloudNative application platform that was highly opinionated. And those strong opinions brought to Enterprise customers, and if you found an Enterprise customer, who was willing to absorb those strong opinions, their outcomes were PHE-NO-ME-NAL, absolutely phenomenal.
Now, for some Enterprises, those opinions were
too strong. And Cloud Foundry by design was a platform that was designed with
these strong — you know, I used to say we’re unapologetically opinionated in
our platform. And, again, that really yielded fantastic outcomes.
And then, what happened is that Kubernetes came
along. And Kubernetes was a tool kit, if you will, I considered Kubernetes a toolkit
for building platforms. And famous people like Kelsey Hightower said it’s a platform
for building platforms. And it really is. It is a toolkit for building
platforms. And while I was still at Pivotal, we kind of toyed with this idea of,
well, what people really want is an opinionated Kubernetes platform. And while
I’m not sure that everybody at VMware and Pivotal would agree with this, my two
and a half years of working on Kubernetes at Pivotal, I started forming this
opinion that an opinionated Kubernetes isn’t what customers wanted, they wanted
their own flavor of Kubernetes.
And that is what has brought me to Weaveworks. It is because what we do is, we are really embracing – we’re building out a certain set of tools that will help the platform teams build out the platforms that their developers need, instilling the opinions that their developers need or that their company – sometimes those opinions are not developer opinions, they are compliance and regulatory and security opinions. But it’s more of let the platform teams at these various Enterprises build THEIR platform. And I see that the way that Weaveworks is embracing that and providing that capability to enterprises through declarative configuration, multiple reconciliation loops – we will talk a little bit more about those things – that was what excited me about it. And that’s why I’m at Weaveworks now.
Value Prop
Mike Schwartz: From a marketing
perspective, I don’t envy the marketing team who has to consolidate this group
of products and services into one understandable business proposition. What do
you think are the most important value propositions for Weawork customers?
Cornelia Davis: One of my favorite bits of work, and this isn’t my work but it
is such good work in the industry, is the work that the DevOps research
assessment group has done. So, DORA – I’m talking about Nicole Forsgren, Gene
Kim, Jez Humble. And what they’ve done is, they’ve done this work – I’m sure
many of your listeners, and you’re already familiar with it – they did the work
to tie specific IT practices to business outcomes. So, high performing
organizations, those that are gaining market share that are profitable and so
on, tend to do certain things in their IT practices a certain way. They release
software more frequently, they have lower meantime to recovery, they have lower
change failure rates, and so on.
So, the business proposition here is exactly
that – it is to enable your application teams to achieve those metrics. Because,
again, DORA – and your listeners might also, if they don’t know DORA, many of
them probably do know the state of the DevOps report – there they’ve done the
work to prove that these IT practices really correlate to business strength, either
negatively or positively. So, that’s the fundamental thing.
What we’ve been doing is, as an industry, we’ve
gotten pretty mature on understanding how the cloud-native architectural
patterns and software support those things, where we are still developing is
cloud-native operational practices, not software design but operational
practices. And that’s really the value prop is cloud-native operational
practices that are going to contribute to these IT practices which have a correlation
to business outcomes.
How Do Customers Find Weaveworks?
Mike Schwartz: Would you say that customers are finding your product sort of through the operations
group, or is it more common that maybe the IT leadership of the CIO/CTO office is
saying, “Okay, we have to figure out how to make this more scalable?”
Cornelia Davis: So far, a great deal of our traction has come from our presence
in the open-source community. And you talked about this set of different
projects and challenge that marketing faces, we have a lot of things going on
in the open source, that are all related. But you have to read the tea leaves a
little bit, I think that’s what you were getting to earlier. And so, we have
some very successful open-source projects. And to date, I would say that a good
portion of our kind of entrance into commercial things has come from somebody
who has been playing with Flux, which is one of the components, one of the kind
of core GitOps components that we have out in the open-source community. Or
they want to do canary style deployments because that is something that is
hugely valuable to reducing your change failure rate.
And people like James Governor from RedMonk is
talking a lot about progressive delivery this year. So, there’s a lot happening
there. They are looking at Flagger and so on. And so, there’s been a fair bit
of that. And I also think, to some extent, that we are still kind of the tip of
the spear. I don’t believe that CIOs right now, CIOs in many cases, what
they’re trying to do is “move to the cloud”, but they’re not yet at the level
of detail.
And by the way, these CIOs understand that they
need things like microservice architectures, but they don’t know yet what to
ask for when it comes to cloud-native operations. Yes, they’re talking about
infrastructure as code, which is something that we’ve had for a while, and
they’re starting to look at some of those things. You know, CI, well, that’s
something that’s already quite mature, but when it comes to the very things
that we are selling, if you will – and I don’t necessarily mean just from a
monetary perspective, but that as well – these concepts are still relatively
emerging.
And so, I don’t think the CIOs are asking just yet, it’s either the developers or the platform teams who are saying, “Oh, well, I can actually create my platform and create repeatability and reduce my meantime to recovery because I have this completely non-snowflaked expression of what my platforms look like, which means I can stand one up on, you know, in moments.
How Much Of The Code Is Open Source?
Mike Schwartz: I was looking on
your website, and there’s eight open-source projects that are listed. I bet you
there’s even a couple more that maybe aren’t listed. In rough terms, what
percentage of your developers are maybe of your engineering team is working on
these projects?
Cornelia Davis: That is a great question. And in fact, that’s one of the things
that I’m working with our VP of engineering, David Turner, to kind of rejigger,
if you will, that. Well, if we look at the number of places that code is
committed, I would say that 80% of the code that we commit into a git
repository goes into one of those open-source projects. Even Enterprise
product, Weave Kubernetes platform, which I’m sure we’ll talk about a little
bit more, a good portion of the Weave Kubernetes platform is based on a project
called WKSCTL, which is an open-source project.
So, even that is pretty significant. We have
another open-source project that we partner with on Amazon, which is EKSCTL, which
is the CLI, it is the UX4EKS that provides a higher level set of abstractions
than the base APIs for EKS does. We have Flux, which is kind of our flagship GitOps
project. We have a developer experience team that builds a number of getting
open-source projects as well. So, again, I would say that 80% of the code that
we commit, gets committed to those repositories.
Now, in terms of the teams working on the commercial product, we have people working on a commercial product that could make quite a bit into the open-source projects. Yes, we have some private repositories as well.
Balancing R&D Investment In Open Source
Mike Schwartz: Lot of open-source
companies struggle with, well, how do I justify investing in open source versus,
let’s say, the cloud platform or the Enterprise, if there’s an Enterprise
product. How do you make that balance between devoting resources?
Cornelia Davis: I’ll say that part of the answer here is that is seated in our
bias towards community and open-source. The founders, everybody on the
leadership team, I daresay just about everybody in the company really feels
this draw to open-source and this commitment to a broader industry. I don’t consider
Liz Rice – I consider her my colleague, she works at Aqua, she doesn’t work at
Weaveworks, but I consider her my colleague. I even consider everybody at Pivotal
still my colleagues.
And so, I think that we have this kind of
philosophy, if you will, that, we’re community first or industry first. And so,
to some extent, it’s seated in that. I used to joke around that my
father-in-law years ago always said, “If I ever win the lottery, I’m going to
share my winnings with all my kids.” And so, I used to say, “Well, when he wins
the lottery – and he has that luck, that’s why I say when – that I’m going to
quit my job and work on open source.” I said that 15 years ago, and now, I am
gainfully employed and I work on open source, which is great, it’s been a fantastic
shift. So, this bias that I just
described around, you know, we have this feel to connect with community is not
unique to us or to me personally.
So, that’s kind of, we have that bias towards
this, but again, we, of course, in order for them to pay me my salary so that I
can work on open source, we have to make money. And one of the areas that we’re
super committed to is that we do not want to have Flux for example, which is
this reconciler, this continuous delivery reconciler, we don’t want to have Flux
features that are open-source and Flux features that are closed-source. So, we
believe that everybody should have the ability to do the basic capabilities of
these GitOps workflows and these cloud-native operational ways of working.
And what we want to do is we want to help
Enterprises, and Enterprises are our target market here. We want to help them
do that at scale. That’s where we decide. So, we spend a lot of time, we have
this kind of view of what goes into the commercial product is where you’re not
applying these things to one or two projects or 10 projects, but you’re
applying these to an organization that has 5,000 developers and 3,000
applications. And that’s where we’re helping organizations do those things at
scale.
And so, having that kind of as a overall governor of how we decide what goes where, it maybe doesn’t help us decide how much effort to put in places, but it certainly helps us when we’re thinking about a capability, and we’re talking to a customer who says, “I need this.”, we’d look at it and say, “Is it a scale problem, or is it something that we really want to enable the entire community to work on?”
Revenues: SaaS V. Self-Hosted?
Mike Schwartz: Switching into the products space a little bit. So,
Weaveworks has a product in the, let’s say self-hosted space called Weave
Kubernetes platform, and there’s also Weave Cloud – which of these projects or
products are actually more important to you from a revenue perspective?
Cornelia Davis: I’ll answer that question in two ways. I’ll answer the question you’ve asked, which is, from a revenue perspective, but I’ll also answer from more of a strategic perspective. Weave Cloud platform, which is the one that I talked about earlier, which is the SaaS offering that allows somebody to say, “I’ve got my own Kubernetes cluster, and I want a set of services that are going to allow me to manage that Kubernetes cluster and the workloads that are on it in a better way.” So, around observability, metrics, logging and those types of things.
Again, this is before my time, but I think that
at the time that we created the Weave Cloud, we felt like there was this need
to understand how to help people operate both the platform and the workloads
running on the platform. And we have gotten a certain amount of traction but
it’s kind of the small to medium business type of thing, or it might be people
that are into Enterprises, but they’re still operating at that smaller-scale
that I talked about.
So, it’s still a project, where somebody says,
“Oh, well, I’m working on a product. I’m within a larger organization, but I’m
solving this particular problem just for myself, just for my project. And
that’s where we’ve gotten the traction. And that’s generated a modest amount of
revenue. But really back to what I was saying earlier, is that what we really
want to do, the value that we want to bring to Enterprises, is to help them
solve those problems at scale. And so, while the Weave Kubernetes platform is a
newer platform, I’m not at liberty to talk about where we sit with revenue
today. We absolutely anticipate that the Weave Kubernetes platform, that’s what
we’re banking on, that is where we are going to see the revenue opportunity.
And so, therefore, it’s also kind of our strategic platform moving forward.
Now, Weave Cloud folks, don’t worry, it’s not
that we are going to ditch Weave Cloud. I think that Weave Cloud is still one
of the best places where you can start to experiment with some of the things at
that lower level of scale, to start to get some of those capabilities. Because,
frankly, what we do in the Weave Kubernetes platform, which is the self-hosted,
that you talked about, and I like how you said it self-hosted, it’s not
necessarily on-premise self-hosted, and it might be self-hosted in the cloud,
or it might be self-hosted in your own data center, but it is self-hosted. So,
while WKP is strategically, and from a revenue perspective, definitely the
direction that we’re going, Weave Cloud folks, don’t worry, we still think that
Weave Cloud is a great place for you to start to, you know, get your legs under
you and solve some of these problems maybe at the lower scale.
And so, we see both Weave Cloud as a great entry point to understand the services that we offer as open-source is as well, I described earlier how using the open-source projects is also another way for you to understand the set of services that we offer, on your road to solving these problems at scale, which is where the WKP platform comes in.
Is K8 A New OS Platform For The Cloud?
Mike Schwartz: Okay, so, I’m not
a cloud native expert here, so, you maybe have to cut me a little slack on this
analogy, but would you say that perhaps the Weave Kubernetes platform is sort
of like a Linux distribution, where, for example, there are a lot of ways to
deploy Linux to there’s room for different distributions? Could you say the
same for Kubernetes? That maybe people can think of Kubernetes as a new type of
cloud operating system?
Cornelia Davis: Oh, absolutely. And I think that the industry is a whole thing’s
of Kubernetes is a new Cloud operating system, but your analogy is actually
more — it goes beyond that, it’s deeper than that. Because you said that it is
a way of managing perhaps a number of different distributions of Kubernetes. And
you nailed something that is really, really subtle there. And that is that we
are not aiming with WKP to be the Kubernetes distribution.
While we do have customers who are using WKP as
their Kubernetes distribution, what we are really doing is, we are providing a
platform that allows you to manage whatever Kubernetes distribution you want. Now,
I’m not going to suggest that you can just come along and say, “Hey, I’ve got
my own bespoke fork of Kubernetes and all this stuff in it.” And we can just
say, “Oh, just point us to that repository, and we can go ahead and install it.
We’re not there, but we do support in WKP a number of different flavors of Kubernetes
under the covers.” And, you know, for example, we do, out-of-the-box, provide
you the mechanics to support upstream Kubernetes. You know, the releases that
are put out on the Kubernetes in the GitHub repository under the releases tab.
We also support EKS in there, and that list will continue to evolve. And we are working on ways of bringing other Kubernetes releases in there. And WKP also allows you to say, “You know what? We’re actually going to treat the Kubernetes thing out of band.” Because once you get that Kubernetes dial tone, if you will, there’s still an awful lot that you need to do on top of that to configure Kubernetes, to configure it against the storage systems, to configure it against your networking and your ingress controllers, and to set up the access controls and the pod security policies and all of those things, install the services that you use on top, very services that I talked about on Weave Cloud, like monitoring, and logging, and observability, and all of those things. And, by the way, Flux, which is the GitOps capabilities for your application teams, we are very focused on helping you manage all of that complexity on top of Kubernetes as well.
Partnerships
Mike Schwartz: You’ve mentioned a
number of companies in adjacent markets and Amazon for example – can you talk a
little bit about the partnerships that have been most important and that you
see as being most strategic for Weaveworks going forward?
Cornelia Davis: So, Amazon of course is a close partner of ours, it’s really very
interesting – and again, I always come at it from a little bit of a technical
angle – but I’ll go back to my days at Pivotal. So, I worked on Pivotal Container
service, which was Pivotal ‘s Kubernetes distribution we were working with
VMware, bringing that to market. And when I was out talking with customers,
they were like, “Well, okay, how this compares to…”, and one of the things that
was often asked was, “Tell me how this compares to EKS.”
And one of the things that, at the time, I
talked about was how EKS was this — it wasn’t at all a turnkey Kubernetes user
experience. You couldn’t go to EKS and say, “Hey, give me a cluster, give me a
cluster with three masters and 10 workers.” And it would make it so. It was
much lower-level protocols.
And so, our partnership with Amazon really
launched when Weaveworks said, “Okay, we believe that we can use these
practices that we’re starting to become experts in to provide a better UX.” Or,
not better but easier UX. We will use those low-level building blocks to create
a user experience on top of EKS, that make EKS accessible to a larger market
and a larger set of people. And so, that was the genesis of our relationship
with AWS.
Now, the interesting thing is that I can use AWS
as a bit of an exemplar. Because AWS has been very successful in – well,
massively successful across a huge market, and certainly Enterprises are moving
in that direction. But Enterprises still are doing an awful lot of their own
stuff, their own self-hosted stuff, so they haven’t embraced kind of the whole cloud-native
operational model.
And so, working with companies like AWS and
Google, and we have a very strong partnership with AWS – we also have a relationship
with Google – to help those organizations connect with the Enterprise in a way
that we — you know, our legacy, again, remember that Alexis, Matthias, myself,
we all come from the VMware, EMC, Pivotal Enterprise market. So, that’s been
certainly part of that partnership is to help the cloud providers move in that
direction.
You’ll notice that I haven’t said anything about Microsoft, and I put Microsoft in kind of a little bit of a different category because they also have that Enterprise genesis, if you will. So, they understand the Enterprise, their history is in the Enterprise as well, sure, PCS, and all of that stuff, but they may arguably become very successful in the Enterprise Windows server exchange. You know, they were heavily entrenched in the Enterprise. So, we are very much partnering with Microsoft there as well.
I think the thing that’s given us traction the most in that is that Microsoft, who I have been so impressed with – who hasn’t in the last five years, kind of five to eight years, they’ve totally reinvented themselves – is that they have very early on latched onto this, again, this cloud-native operational model, which is what we represent, and the term we use for it is GitOps.
And so, that is really kind of the core of our partnership there is that all of these cloud service providers are recognizing that were at an inflection point in the industry, where we are really starting to revolutionize the way we do operations. And making that cloud native, SRE was just the beginning of that. There’s a whole set of tools and practices around that.
Customer Perception Of Open Versus Commercial License
Mike Schwartz: You are very much
on the open-source bandwagon, but I’m going to play devil’s advocate a little
bit and say that there are still some negative connotations around open-source
software. And, for example, if there’s a commercial and open-source version of
a certain segment of software, the commercial version is often viewed as better
or more secure or justifying a higher price. And also, when you look in terms
of dollars of the software market, there’s more dollars spent on commercial
software than on open-source software.
Obviously, you’re very pro open source, but do
you think that perception is changing? And is it changing in all the segments
or in certain segments but not others?
Cornelia Davis: Well, I think that, overall, the perception of open-source has clearly changed. It dramatically goes back to that story when I used to think I had to quit my job to be able to work on open source. And now, the reason that people work on open source is because there’s a demand for it. And that demand is driven by revenue as well, because most of everybody who’s working on open source is not independently wealthy. They’re getting paid through something, and they get paid through – you know, the traditional open-source business models are, we’re going to provide support over open source. And that’s a great business. And people definitely do well with that. And then, there’s the, “Okay, we’re going to actually provide a different set of capabilities.
And the former of those models is really where it comes to this notion of open source as “free”, free like a free puppy is, is that, yeah, even though you’re not paying for the source code, to be able to operationalize that open-source project, you need a whole lot of skills, you need support. And you might even need a supply chain, a software supply chain to help you deal with the releases that are coming out.
Years ago, I worked with a colleague in the EMC days, who had really spent his whole career in mainframes, and when we first started working on open source together, he was like, “This is never going to work! Things are changing all the time! I had something running on Friday, there’s a new release on Monday, and everything broke.” But, obviously, we’ve built actually business models around kind of being able to deal with that variability.
But I also think it’s completely fair to have a commercial model that says, “We’re providing some capabilities.” I don’t like the open-core model myself, and I wouldn’t consider us having that open core model, which is to say, “Hey, there are some features that you just don’t get unless you buy the commercial product.”
I know it’s a little bit of a slippery slope. You might argue that the model that I described earlier, which is, “Hey, we’re going to help you solve these problems that scale is a little bit of that, but I see it as a little bit different, which is to say, “No, the core capabilities — it’s democratized, everybody can do that. We’re just going to help you solve a different set of problems.” And that’s where we’re going to put our commercial efforts.
So, the other thing that I’ll share is, I’ve spent a lot of time over the last six, seven years with Enterprise customers, and more and more of them, they want to move to open source. Now, sometimes, they believe that by going to open source, it’s all about vendor lock-in for them. And that is tricky. Because as soon as you engage with a vendor for anything, even kind of the tooling that you’re going to use to keep up-to-date with the changes on open-source, at some point, you end up with some level of vendor lock-in, even if it’s just consulting.
So, I’m not sure that that’s the best reason,
but I have found that enterprises, even if they don’t understand exactly how,
and think that it’s about vendor lock-in, they have this really good intuition and
this instinct that open source actually is a better way for them to go.
And, yes, but at the same time, they’re like, “Look, I want my developers delivering value to my customers. I don’t want them doing all the things that you can do for us. And so, we will enter in a commercial agreement with you.”
What Can We Do To Help More Women Join The Open-Source Community?
Mike Schwartz: Putting on your activist hat for a minute, you’ve probably
noticed that the male/female ratio on the podcast hasn’t been that great,
although we’re trying to make it better this year. But I listened to one of
your podcasts, or one of your previous presentations, and you said it’s up to us
to do something about this, to foster a better tech environment that results in
more women in the industry. So, what are some of the most important things we
can do in the open-source business community to affect that change?
Cornelia Davis: So, you asked specifically in open source, and that one is even
trickier. So, by the way, I, first of all, want to congratulate you on. Maybe
over the course of your entire podcast, your ratio hasn’t been that great, but
you’re knocking it out of the park this year, Mike. So, doing really well. And
I really appreciate your efforts there.
And, in a way, what you’re doing is the
exemplar, and to answer that question, whether it’s in the open-source community
or in the industry at large, which is, you just have to pay attention. And we
all have to pay attention, and we all have to make those decisions. So, one of
the things that I regularly ask my male counterparts is that when you are
invited to sit on a panel, please, decline until there is a woman that is
sitting on the panel with you. Do not sit on all-male panels.
So, here’s a very concrete pragmatic thing that
we can do.
Now, open source, my perception is that it is
even trickier there, is that, in a way, open-source – we kind of think of it as
the total democratization of software, all the software is in the open, anybody
can participate, anybody can pick up an issue from an issue’s list and study
that, come up with a solution, submit a pull request, and those types of
things.
But the reality is that the environment, so, one
of the challenges of women in technology is that we are constantly getting
triggered, signals all day long, all day every day, I always say hashtag #alldayeveryday
we’re getting signals that we don’t belong. It could be anything from surprise,
like, “Oh, really, you’re a software engineer?!” You know, those types of things
to even more overt stuff. And, unfortunately, it still does happen.
And so, we tend to – and again, this is a gross
generalization, but I think the studies have shown this that women tend to
maybe hold back a little bit and don’t speak out as much. And that ends up just
perpetuating itself. Because to participate in open source, you really have to
put yourself out there. And it’s harder for women to put themselves out there.
So, I think that – going back to you quoting me
– it’s up to all of us. By us, I mean ALL of us. I mean, men and women to use
— you know, men in this technology industry have a privilege because they
outnumber women so significantly, is, be conscious about that and then lend
your privilege. Look for elevating the voices of women.
If you see somebody who is participating in a Kubernetes SIG, who is on the calls every week, and is just listening in and isn’t speaking up, take a moment to ask her opinion. Also, maybe provide encouragement to, “Hey, you’ve got some great ideas here. You should really issue a pull request against this.” Help turn up the volume for the women who are participating. Because I think the numbers in open source have shown that the numbers are even worse in open source than they are in the industry at large.
Advice For New Open Source Startups?
Mike Schwartz: This is the same question I asked everyone last, which is, do you have any advice for the poor entrepreneurs who are launching a business around an open-source software project?
Cornelia Davis: One of the things, and this is quite personal, is that the first
bit of advice I would say very directly is, trust your gut, trust your intuition.
However, that’s not enough. And just be very aware of the fact that your
intuition is surely influenced by the experience that you have to date, what
you’ve done in your previous jobs, what you’re seeing happening out in the
open-source industry and those types of things, but make sure that you treat
that intuition as a hypothesis. And look for ways to validate that.
Now, I’ve spent my entire career in emerging
tech, and so that validation step is very tricky. Because — and one of my
favorite quotes, and everybody’s heard it, is the Henry Ford quote, which is,
“If I had delivered what people asked for, it would have been faster horses.”,
or something along those lines. I think I just totally botched that, but I think
people get the idea.
So, in many cases, as an entrepreneur, you are really
at the tip of the spear. And so, how do you do that market validation. You want
to – again, another cliché – but you want to go to where the pack is going to
be, not where it is now.
I think intuition is something that is really,
really very good. And then, my experience at Pivotal over the last seven or
eight years is that they really did a fantastic job on getting there incrementally.
So, get lots of feedback, put things out there, find some trusted advisors, find some people who are willing to go on the journey with you. You are not going to want to go into the very well-established Enterprise organization that has done things the same way for the last 50 years. They’re not your partner on this. Look for the partners who are looking to go on the journey with you and co-invent with you. I always think of the people that are customers as people that I’m co-inventing with.
Closing
Mike Schwartz: That’s great
advice. Thank you so much for sharing all of your experiences today, and best
of luck at Weaveworks.
Cornelia Davis: Thank you. It’s been a pleasure to be here.
Mike Schwartz: Big thanks to Radmila Ercegovac from Manning for bringing Cornelia’s book to my attention. And also, thanks to Alexis Richardson, the CEO and Co-Founder at Weaveworks for helping us coordinate.
Audio editing by Ines Cetenji, transcription and episode website
by Marina Andjelkovic, cool graphics by Kemal Bhattacharjee. Music from Brooke
For Free, Chris Zabriskie and Lee Rosevere.
Next episode, I was really honored to get the chance to interview Melissa Di Donato, the CEO of SUSE. Since being spun out of MicroFocus, SUSE is one of the world’s largest independent open-source companies. Don’t miss it. Melissa had a ton of insights into the industry, and how SUSE is positioning itself to provide a leadership role. Until next time, stay safe and thanks for listening.
The post Episode 51: Cloud Native Agility, Reliability and Stability with Weaveworks CTO Cornelia Davis first appeared on Open Source Underdogs.