Cloud Native ComputingDevelopersDevOpsExplainedFeaturedTFIR InsightsVideo

AWS Outpost : Impact on the industry | Insights by Rob Hirschfeld – RackN


Here is a lightly edited transcript of our discussion with Rob Hirschfeld, CEO & Co-Founder of RackN.

Swapnil Bhartiya: I want to understand from your perspective, what does AWS Outpost mean for Cloud in general.
A retreat for Amazon
Rob Hirschfeld: For AWS, this is really a big change in their thinking from “everything is running in our Cloud” to “you’re going to own and manage the infrastructure.” It really is a redefinition of how they’ve traditionally positioned what Cloud is into something that’s much more hybrid. They are recognizing that people are going to run their infrastructure on their own premises, or in their own colo, and that they want control of that infrastructure.

Amazon is really maintaining very tight control on that infrastructure. It really is a retreat for Amazon in what they think of infrastructure actually – how it’s managed and deployed. It’s a major change from that. I don’t think we should underestimate just how much of an acknowledgement that is and that people do want to hug their servers.

Swapnil Bhartiya: What does it mean for competitors, how are they reacting?
Rob Hirschfeld: We’re really seeing just how dominant a player Amazon has become. Five years ago nobody would have expected them to literally walk into people’s data centers with infrastructure servers and services to sell. The fact that they’re really doing this is really shaking up everybody in the industry. What I see is everybody responding to this announcement in very serious ways. The VMwares of the world have to take this seriously because it’s really bringing Amazon infrastructure into what has traditionally been their last mile. I know that they have a play in here, but it’s not the same as when people really ran their own infrastructure and VMware was a software player. This is a very different services angle.

They got to think about Google with Anthos, which is a very different offering. Microsoft has Azure Edge and a whole bunch of Azure pieces that they’re trying to bring it on premises. So, it’s unclear if Amazon is going to suddenly burst into the scene and make waves. But this is clearly a place where other people thought they had leadership, Google specifically, and then suddenly found that they didn’t, because Amazon’s leap frogged them from a offerings perspective. And then I think traditional software vendors have even a bigger hill to climb.

We have been watching the SaaSification of the industry, where everything runs as a service in a cloud or multiple clouds. We see people in the industry getting a little wary of giving up that much control, and we’re starting to see people say, “You know what? I actually do want to be able to run software and manage my infrastructure.” But in this case, we’re really saying, “hey, if you want to be a SaaS, a SaaS can now come on premises through Amazon.”

Swapnil Bhartiya: There are a lot of AWS customers and if you are on AWS, you really don’t care about those controls. How are these users responding to Outpost?
Rob Hirschfeld: I haven’t seen as much reaction in the industry. I think either the people who are early adopters are being very tight lipped, because they have to be or they are still evaluating. I definitely see a lot of people interested in what they’re doing. The same thing is true with Anthos. I know that there’s a number of people doing Anthos trials, and that stuff hasn’t surfaced as much as a general commentary in the industry either. And it could be that it’s just a much bigger footprint for people to consume. None of these solutions – a cloud coming on premises – are casual solutions. They require big contracts, big lifts, and some cases, heavy consulting footprint.

So these aren’t traditional cloud drop in, buy a server with your credit card or VM with your credit card. In some cases, they’re much more depth than just buying a server, installing and racking it yourself. You’re really committing to the cloud process, the cloud way, the cloud services that are available. It’s a very different story from the days of do-it-yourself.  I personally think that ‘do-it-yourself’ is part of the innovation that’s led to IT working, where people try a lot of different things. This is not showing us with the same innovation up and down the stack. It’s the innovations at the top.

Swapnil Bhartiya: Amazon is breaking away from its stand of ‘everything has to go in the public cloud’. Does it also pose a risk for Amazon because it shows some cracks in its armor?
Rob Hirschfeld: It does. I think that it is an acknowledgement that people have a reason to run infrastructure on premises. They were doing that with Greengrass. In this case, it’s really acknowledging that people have a reason to have their own data centers and they’re trying to thread the needle on that message.

The data’s been very consistent. Only 20% of workloads are moving to the public cloud. There’s plenty of examples of companies that are all in, or cloud first. Word on the street that I’m hearing is that’s not always working out for companies the way they expected. Either it’s harder to drag all their legacy pieces over, they’re finding when they get there, that is very expensive. Sometimes they’re finding that the service reliability of cloud is not everything that’s cracked up to be, and that there’s issues and problems. Security issues that it might have are better than you had done your past in your legacy environment, but it’s still there. It’s still operational concerns and just trying to make it all swept under the rug. It’s not making the problems go away. You’re just ignoring them frankly.

Swapnil Bhartiya: How does it change not only cloud computing, but also computing in general?
Rob Hirschfeld: Some of the things that I think a lot about, and I actually have concerns about, is that Amazon, while it has a lot of services and capabilities, it’s still a very monolithic architecture, and that you’re forced to run things Amazon’s way. That’s what Cloud Native is about. It’s about adopting your processes to the benefit of the cloud vendors. And they are vendors, so they have reasons for what they want to optimize, and how they want you to use their services. So the more we buy into these, and the more people adapt into these infrastructures, the more they’re captive into what those processes are,

Vendor lock-in is a real thing. It’s not lock-in that you’re locked into a vendor, you’re actually much more locked into the processes, and tools, and capabilities, and APIs that that vendor has offered you. It can be very hard to move away from those pieces. Our industry as a whole, by embracing that so fully, has really moved away from saying, “I have a standard open API,” or “I have a way that this is going to work in across vendor-portable way.” In reality, we’ve really slowed down and stopped focusing on that to the same extent. I think that’s a challenge.

Swapnil Bhartiya: Datacenter used to be an escape hatch from public cloud. But that escape hatch now opens into AWS.
Rob Hirschfeld: That’s the industry challenge here. If you look 10 years out, if Amazon is successful in this march, then every computer cycle, every operation would ultimately run on Amazon infrastructure managed by Amazon. Outpost has really a statement that “we’re coming into your data center,” type of thing, which is amazing if you’re all in on Amazon and you’re taking advantage of everything up and down the stack. Things like that tend to be amazing until they’re not, which is, you realize you’ve really depended on a vendor for this type of work. We saw the same thing in manufacturing with supply chains. To me, I have an industrial engineering background, but it’s a supply chain question.

You can do everything yourself in-house and literally build a completely vertically integrated product, and that has challenges too. Or you can go into a much more distributed supply chain. If it’s too distributed, that has problems. But if you depend on just one vendor in your supply chain, those become really serious challenges. Those vendors have a tendency to walk up and down the supply chain and continue to take over industries that are related to them when they get in the middle of things – railroads, oil, telephone are some examples. We’ve seen the story before. Amazon is breaking out of the data center shell that they’ve been in, not that they’ve been constrained very much, but they’re breaking out of that. And for everybody in the industry, we need to be asking, how do we want to respond?

Swapnil Bhartiya: A lot of innovation was happening because of, do-it-yourself. People were scratching their own etches and creating solutions for everyone else to use. That kind of goes away with Outpost. What does it really mean for the startups, for players who are coming up with solving problems, because everything will be locked into AWS ecosystem no matter where you run it.

Rob Hirschfeld: In the 90s, the Windows monopoly was actually very effective for helping a lot of companies have a common place to do work. But it had negative implications also in blocking out competitors. There’s a fine line to work with this. I think that Amazon, as a standard platform, has been very helpful for people to figure out how to do real automation, and really build IT infrastructures that were robust and scalable.

The challenge becomes if everything we do is against that Amazon API, and they start being able to control, tune or limit what people are able to do in that. Then that ends up hurting innovation on the other side, because it becomes impossible to come up with new things. At this point, competition with Amazon is either build it bigger, and bigger, and bigger clouds, and try and compete on sheer footprint, or to find ways that allow people to do what Amazon does by themselves.

There is no doubt there will be competition – Edge Computing is here. We are moving into small footprint infrastructure, where things are redefined, and the standard architectures don’t apply anymore. ARM is much more interesting, and non-traditional vendors have opportunity to come in and sell infrastructure into those markets. So, there are places and that’s why the Outpost is potentially so interesting and threatening because it could be Amazon coming into the edge. It’s a place where we’re about to see this burst of new hardware, new software platforms, new use cases for the industry. We really want those things to go through very fast, tight, iterative cycles where new vendors have a lot of incentive to come out.

If I had to build a new cloud infrastructure, that would be a really hard challenge. You’d have to literally have a global footprint overnight, which is very hard to do. It’s like trying to build a new car company. And to Tesla’s credit, they walked into a place where the incumbents were very dominant, but Musk used billions of dollars to try and do that. So we’ve reached a point where cloud infrastructure is a multibillion dollar entry game. The innovation that’s going to come is either going to come by going on top of that, which entrenches Amazon’s interests, or is disruptive to Amazon’s existence. That’s really where the industry’s ended up.

Swapnil Bhartiya: As AWS becomes more dominant, would the rest of the world – Europe and China – react?
Rob Hirschfeld: It’s fascinating to see. A lot of times, I think the world is chasing what Amazon has created. China is building huge big clouds. They have the resources and the mindset of government establishing big footprint domination. They’re certainly not in favor of a whole bunch of small individuals and small companies doing that work. This to me is actually the interesting challenge. Europe tends to be more regulatory and want to have smaller regions, but not necessarily encourage all of the innovation and small companies popping up that would be less controlled.

The regulatory environment in the United States encourages a lot of growth and innovation in companies where they’re changing the rules, if you will. The VC community has been blowing those up into unicorns that continue to challenge the rules even when they’ve become major players, which is what we see happening in the cloud and social media spaces that there’s continuing this wild West mentality for things that start affecting hundreds of millions of people.

From RackN’s perspective, what we need is to find the right balance. The challenge with innovating around Amazon is, if you’re chasing Amazon, you’re chasing Amazon. You’re following their pattern and their lead. And that’s not a bad thing, but what you need to do is, do things the way Amazon does them, better and more efficiently. It’s going to take a smaller, somebody saying, “you know what? I don’t want to have to buy hundreds of millions of servers in hundreds of locations, in order to compete with Amazon. I want to do it in a way that allows the people in the sites to have control and buy their own infrastructure and use their own assets.”

The way that the industry is going to change is to go forward into an innovation model that we used to see. Where individuals could buy, integrate, do their own work, write more back to a traditional software, and making that possible for people to lift up their own infrastructure.

That’s what Amazon is really working against right now. They’ve changed the baseline to say that “you use our infrastructure; we take care of all that stuff. You build on top of it.” In my mind, we have to solve the problem that says infrastructure has been too hard to use, too hard to buy, too hard to innovate on top of in the small scale. Once we solve that problem, then things should move back to individual operational control. That actually has a lot of benefits, especially for the people who try to innovate and manage their own destiny.

Swapnil Bhartiya: What would be the impact of the Outpost on Open Source?
Rob Hirschfeld: Right now, open source has been moving up in the stack until just the software that you run on top of AWS. When you look at what the cloud has been creating, from an Open Source perspective, open infrastructure code such as OpenStack and even Kubernetes to an extent has not emerged as a disruptor for Amazon’s infrastructure as a service business.

I think that we’re actually in a place where people are reevaluating how it works and what it can do, and what the sustaining model is, behind any of these Open Source projects.

One of the things that Amazon has really showed u, is that operational control and value in sustaining our core values in IT. If people think that they are going to get that from an Open Source project without paying for it, they need to re-evaluate the whole market.

I have seen people downloading free stuff, and then watching Amazon walk all over people, as they realize it’s hard to sustain their data center infrastructure for free. Then we saw Amazon entering the market provide a service where you don’t see how that works behind the scenes, and just crushed the market.

Amazon, Microsoft, Google, Alibaba, Oracle and so on are doing the same thing. There’s no Open Source cloud infrastructure. We really need to think about what Open Source is. It’s an innovation model where people can collaborate. Collaboration is the key thing to me. But the collaboration has to be around a sustained kernel  where people can count on the APIs, count on the quality, count on the support that they need to make the place where they’re innovating around work.

Open Source hasn’t addressed that question in my mind. Dropping code out there and hoping people can use it does not create that core of sustainable innovation that you innovate around. Amazon has shown us that they can provid the core people innovated around us. Outpost, potentially is coming onto your premises to continue delivering that value. We’re going to keep innovating around those APIs. The Open Source communities need to ask themselves, “is it more important to be free, or more important to have these open ecosystems, where people can collaborate, and then start paying for it?” Because you’re paying for Amazon, there’s no free Amazon.