I sat down last week with Chris Kinsman, Chief Architect at PushSpring and a 20+ year veteran of software development, to talk about the factors that go into deciding when to migrate to the cloud. In Part 2 of our conversation we talked about the resistance and arguments against moving to the cloud. Once again I’ll let Chris do most of the talking:
“What I struggle with is those companies who are trying to decide whether they go to the cloud or not even attempt it. For someone starting out, I don’t think there’s any choice – unless there is one specific requirement that pushes you out of the cloud. Or, and this is a little different, they may be evaluating a new project in an existing company. If you’re an existing company that has in-house already – whether it’s co-location or your own data center and you’re already invested in servers and storage – then the default always seems to be continue as is.”
Well, if the business has existing business with the in-house model, to the extent that they’re successful with that model they may be hesitant to impact that revenue stream.
“Maybe or maybe not – it could be that you’re just selling a service and they don’t really care whether it’s your hardware or someone else’s. As an example, if you’re paying for a $50/month MailChimp subscription do you really care whose equipment it runs on? Because you’ve already bought into that whole idea of paying for a service and not really caring about what’s behind it. So if I already have in-house infrastructure, at what point do I transition to the cloud, and that might be a more gradual decision.
One common justification for remaining in-house is avoiding any forced migration. While it’s true that the cloud can – and will! – upgrade your infrastructure, you can also use the cloud to run whatever you want to run.
“As a consumer of a service bus, I don’t care if they’re running it on 2008 or 2012. As long as it provides the same service, I don’t care what they run under the covers and that’s the beauty of it. Because if I care, that means I have guys in my org who are worried about that that I’m paying to worry about that, and how to keep up with that schedule, and everything else.
So let’s go back to machine instances. That’s where it’s now on me to maintain, care about migration. And that’s also a choice. There are services that allow me to run my service as service bus, i.e. if I’m running Azure websites, I’ve got a way to get deployed – I need to make sure that my code runs – but I shouldn’t care what server OS they’re running. So these machine instances, that’s the difference between Platform as a Service (PaaS) and Infrastructure as a Service (IaaS), and Infrastructure as a Service. A lot of folks gravitate towards IaaS because they give up LESS control – they haven’t totally abdicated. But the problem I’ve found is that while that’s an easy baby step to lift and step your infrastructure from your internal data center into the cloud, that’s not really taking advantage of the Cloud. That’s continuing to do the thing you’ve always done, just with a machine that’s somewhere else. And I would also argue that that’s not the way you’re going to realize cost savings by moving to the cloud – because you’re continuing to do things the same way you did them before.”
So, what are the cases you’ve seen where someone should not be using the cloud?
“This is controversial, but I can think of a few. Where technology is not their primary business. Where they are potentially in a declining market. Where they’re not rolling out new software, they’re not trying to gain new users. They’re just rolling something along to the end of its product cycle. They don’t need scalability – they just want the absolute lowest cost. They don’t even really care about the experience of the users. Then, absolutely find the lowest cost place to host your server. But when I look at people who are making their money off of software, who are trying to grow their customer base and get more users, it’s almost never the right decision.
And that’s not just me saying that. I saw a stat the other day that 10 years ago, for an Internet startup 80% of their Series A money went to infrastructure. Today, if you tried to pitch a VC that 70-80 of their money would go to infrastructure they would never fund you. They fully expect that 70-80% of their investment will go to salaries for the individuals who are building the product.”
How about a more mature operation already hosted in the cloud? Is there a point where, for cost or other reasons, it might make sense to move to an in-house model?
“Even those middle stage startups, I’d say help me understand your case. Maybe their case is they’re doing real-time something and they’re very sensitive to latencies. But I’d say that even guys who are doing VOIP are starting to do more virtualization – and VOIP is about as latency-sensitive as you can get when you’re actively listening for any pauses. So what are those cases? Perhaps there are some out there, but I haven’t encountered them.
Take telemetry, if you’re a company going out with an Internet of Things story – you’re the next FitBit, or a wearable, or put data collection of your car in the cloud – all of those business models are built on the cloud.”
Well, you used to work for casinos. They have massive data, real-time needs, and obviously super-tight security. What about them?
“There I can see a lot of resistance to moving to the Cloud. Legally there are some issues. They are very controlled by the gaming authorities – so maybe there are some barriers they have to push through. But I look at some of the challenges they also have – one of the things I thought was very interesting when I worked for them 10-15 years ago was progressive slots, which were one of the “big things” at that time. Where a bunch of machines were aggregated against a single jackpot. And this required dedicated T1 and T3 lines back to the slot manufacturers so they could do this real time aggregation of data and real-time payouts. That would obviously be so much easier if you could do it in the cloud. So the questions would be could you get this past the gaming commissions and could you make it secure enough. Well…Amazon right now is building a Cloud facility for the CIA, if they can convince the CIA to use the Cloud…”
But you’ve seen the movies and read the stories – they make you think that the CIA is run by clowns. Of course the conspiracy side of that is they want you to underestimate them, but still it makes you wonder.
“I go back to argument I made earlier. I have a hard time believing that an environment set up in the cloud is any less secure than your data center that’s connected to the Internet. In fact, it’s probably more secure.
I think the thing that security guys are concerned with is their whole strategy in the past has been how do I wall everything off? How do I provide defense in layers? At the point where more of your infrastructure is shared, it becomes harder to know where those walls are being crossed, which makes them more and more uncomfortable. But the challenge is, what if a competitor figures out how to do a secure solution in the cloud? There’s a real possibility that they traditional technology folks may not be in business down the road because the costs of operating in the Cloud will be so much lower. And that’s what’s going to end up happening – those guys who are staying in-house, staying with their co-lo, and some point in time there’s going to be some startup that says “we’re going to be disruptive, we’re going to blow up this entire industry, we’re going to look at it differently, our cost basis is going to be so mucfh lower for providing this service that will out-compete them. And over time, that incumbent goes away.
We’ve seen that again and again, where it isn’t even a brand new business. Heck, Amazon – selling books is not a brand new business – it’s a really, really old business. But they came in and did some pretty interesting things driving their costs way. And figuring out all those things they could do to bring on additional ancillary revenue. What happened to all those big booksellers?
Or shipping. Look at Fedex, the whole concept that someone could build an airport outside a city in the Midwest…and now the Post Office is almost out of the business of delivering packages.”
So if you are not actively and continuously thinking about how to use the advantages the cloud can provide, then you are living on borrow time.