IP Expo 2016: Microsoft CTO, Google And IBM Debate The Future Of Cloud
IP EXPO 2016: Microsoft Azure CTO Mark Russinovich and others discuss the future of cloud and whether security is still an issue
Artificial Intelligence (AI), the Internet of Things (IoT) and big data are but three of the main trends set to revolutionise the world of IT over the coming months and years to come, but all will be underpinned by the cloud.
Some of the sector’s biggest names were present at IP Expo in London to discuss the future of the cloud and security, the increase in number of abstractions and the rise of Infrastructure-as-Code (IaC) would impact businesses.
Mark Russinovich, Microsoft Azure CTO
Microsoft Azure CTO Russinovich said that as Cloud grows so will the number of abstractions. He observed that IaaS and PaaS, as markers at different ends of the Cloud spectrum, offer greatly varying benefits and costs that need to be understood by customers.
“In PaaS you begin to talk about higher and higher levels of abstraction, at the far extreme serverless functions .. such as Amazon Lambda, where you just pay for what you use instead of the infrastructure underneath,” he said.
“What I see is a shift from IaaS on the left to PaaS on the right, if you like, where more and more workloads – and especially newer ones – start out.
“It’s only when you need that control of what the operating system is and the patching and application compatibility, and maybe you’ve got regulatory requirements that don’t allow you to go to the right side, that you won’t go there, because that’s where you’re going to get the value.”
Matt McNeill, Google Cloud Platform’s head for UK & Ireland
Google has made a number of cloud platform announcements in recent weeks, including the launch of dedicated data centres for the UK. But it still thinks there is a bit more fleshing out of the cloud concept to do just yet.
“The whole concept of Cloud has yet to fulfil itself,” said Google Cloud’s UK head Matt McNeill. “I was thinking that the last major disruption in technology was when the mainframe got disrupted by the PC. The first applications that were written for the PC were mainframe terminal emulators. There’s a VT100 terminal emulator, and it took four or five years until Apple and Microsoft started bringing out the sort of sophisticated user interface, the applications like spreadsheets – that was when it really became transformative.
“So much of what we’ve done so far as an industry with Cloud is really to emulate the data centre. I think the potential of Cloud goes way beyond what the kind of architectural concepts of servers and disk and CPU that we’ve been dealing with for 40 years.
“What was one of the interesting things of sitting in the inside of Google for the last seven or eight years is you realise we had the luxury to start from scratch. A lot of challenges had forced us to do that and internally the way we architect doesn’t really include the architectural concept of a server. It’s a completely platform-based architecture.
“The primitives of our architecture are things like containers, like distributed file systems, like distributed data processing. Container management and scheduling is how we work. If you take a VM from our Cl oud Platform offering it’s actually just a job in a container, whereas often we think of containers as things that run on VMs. It’s this opportunity to turn all of our thinking upside down.
“The result is you start getting a ‘platform mentality’, where the capabilities that you start to bring into the market are fundamentally different from what we’ve been able to architect in server-based architectures.
“Platform-as-a-Service (PaaS) has for many years has been marked as a place where you do your prototyping – my view is it’s the future.”
Colin Humphreys, Pivotal CTO
Colin Humphreys, CTO at Pivotal, was in agreement with Russinovich about the issue of abstraction.
“We’ve obviously seen people using IaaS, and similarly many people using PaaS,” he said. “But we’re also seeing emerging models such as containers-as-a-service, such as functions-as-a-service, which makes you think about Amazon Lambda, Google Cloud Functions, Azure Cloud Functions.
“One of the things we need to help educate people about is where to use which abstraction and in the right place. If you ship virtual machines around they’re big, if you’re shipping functions they’re small, but the cost of management with thousands of tiny functions comprising a bigger service is really hard.
“What I hope happens is using the right thing in the right place as we all get experience and knowledge of using these Cloud abstractions. I think we’re only going to continue to see more abstractions emerge.”
