(c)iStock.com/binabina
Everyone is looking for that silver bullet which makes their cloud implementations a success. It’s little wonder that events such as Cloud World Forum and Cloud Expo are at the forefront of executives’ minds.
At Cloud World Forum in London last week, MongoDB VP strategy Kelly Stirman spoke of five directions in which the industry is changing, and how companies can make their cloud success a reality: embrace failure, double down on ops, and pick your partners wisely.
Rule number one, Stirman argued, was to embrace failure. To be able to embrace failure, you need to iterate quicker. “Today, customers expect applications faster,” he tells CloudTech. “When you wake up in the morning, you look for apps to have been installed on your phone, and to do that people need to embrace a more iterative project development lifecycle.
“Cloud makes this possible because the lead time to do things and the cost to do things is so much less, that it becomes reasonable to screw up and then recover,” he adds. “Most people aren’t comfortable with that.”
Quick iteration links in to point number two – move beyond lift and shift. Amazon is not doing releases every few days, but every few seconds. It’s a world away from where most companies are, but Stirman argues simply taking your existing application stack and moving it to the cloud isn’t going to give you a huge amount of value.
“Most people think ‘it’s going to take eight weeks for IT to get my infrastructure ready for my application, and the cloud can be just a couple of minutes’, but that’s kind of it,” he says. “If your application didn’t scale, or you had some limitation in your data centre, it’s going to be the exact same problem – or worse – with cloud.”
Stirman adds: “Things that are really important about the cloud, like elasticity, paying for what you actually use, different storage products so you can optimise for hot and cold data, programmable automated infrastructure – that’s what really is powerful and valuable about the cloud, and you need to make that part of how you think about using [it], not just lift and shift.”
Vendor lock-in, although a long-term enterprise IT issue, has long since been a worry for businesses moving to the cloud, finding the implementation is not for them as their business needs change and then finding they can’t get out of it. But it still warrants a warning note from Stirman.
He explains: “The slightly controversial assertion I made [was] in my lifetime, the tow most proprietary technologies that have come to market have been Apple and cloud. These cloud vendors [are] all based on open source software, and different types of standards – but they themselves are highly proprietary. It is inevitable that you are going to get locked in.”
Cloud products are ‘carefully designed’ for a lock in, he argues. The pricing models are the same, you can’t compare prices between offerings, there are charges to get data in, there are charges to get data out, and so on. “All of these players are adding value on top of the core stack, and the more you use those, the more you get locked in,” Stirman says.
Stirman took to burst two myths around the cloud. The first was regarding the security of the cloud – a subject which is often a bugbear for executives as survey data bears out. “It’s just not true,” he says. “Most of these guys are vastly more secure than any of us can design in our own systems.” Secondly, the idea that if you move to the cloud your operations teams halve is also a red herring. “You need them more than ever, because what you’re going to do is use way more infrastructure,” he says. “The operational obligations you have are directly proportional to the number of operating systems. So if you take a big server and slice it up for virtualisation, you are increasing your operational overhead.”
The MongoDB exec saved his most controversial comments for last; there will only be three players in the cloud infrastructure as a service (IaaS) space; Amazon, Google, and Microsoft. Everyone else will run out of money. “The real cloud opportunity is going to be products that are mostly infrastructure as a service agnostic; they are layers you could play in a cloud vendor, or across those three that give value add services,” he explains.
“What’s going to be interesting to see is, are Amazon and Google and Microsoft going to let other vendors come in and play in their infrastructure stack, or are they going to hold people out, the way Apple does, and really own the value added services on top of the stack?
“You’ve got to pick your partner carefully,” he adds.