/>
X
Innovation

Cloud fixes on-premise servers too

Not everyone wants to abandon on-premise servers and rush to the cloud. It's not just whether your data is secure or whether there will be an outage when you need the service most; there's the problem of bandwidth and latency - which is always going to be poorer than your local network.

Not everyone wants to abandon on-premise servers and rush to the cloud. It's not just whether your data is secure or whether there will be an outage when you need the service most; there's the problem of bandwidth and latency - which is always going to be poorer than your local network. Cloud is great for scale and (usually) saving money, but there's still a place for servers. But with all the emphasis on developing cloud services to match and surpass server applications, are on-premise server solutions going to fall behind? With so much emphasis on Office 365 and Exchange Online and Azure and Windows InTune as the future, won't Exchange and SharePoint and System Center end up stuck in the mud?

Not according to Brad Anderson, the corporate vice president of Microsoft's management and security division; in fact, he paradoxically says that making better cloud apps makes server apps better.

Yes, it's cloud first for Microsoft, he says. "If you fundamentally believe that the cloud changes everything, and that the design point for the future is public cloud, the entire culture moves to one where you first think about what want to deliver in the cloud and you build it in a way that also allows to you to deliver on premises. I think it's the right change, because you can always scale down - but it's awfully hard to scale up."

The pace of cloud - and the fact that you don't have to produce something major enough for people to pay for a new licence to get what's every time - makes for a faster development pace; and that means on-premise versions get features later, but once they've been debugged thoroughly.

"In the server and tools business, as a design principle, the philosophy is design for the cloud and then bring to the box. Design your services to the scale, to the security, to [be] multi-tenant - all the things that are required for the public cloud - and get it deployed out there - and then also take that and deliver to Windows Server. One of the great things about Azure is we can update every day if we want; we can develop some technology, get it out very quickly, get that real time feedback on it. That strategy allows us to deliver more value in the box because that technology is already tried and tested and delivered.

"What we have found is the majority of things we need to build actually apply both in the public cloud and the private cloud, the majority of the scenarios have to be done in both. So if you have the design principle of designing for the cloud but then using that technology to also deliver in the box you have the advantage from the architecture standpoint how you build once but then use multiple times."   Those Office 365 outages are a good thing as well, at least for in-house Exchange admins (who should only be feeling smug if they haven't had a significant mail server problem in the last year themselves). "This move to the service is the best thing that's ever happened for customer running software on premise. The reason I say that is the Exchange team uses Operations Manager to manage Exchange on line. So the richness of the knowledge that's gets built in to a management pack is far more than it ever was before because the team building it is living and dying with it. The same with SharePoint… Really, delivering these solutions as a service has been the best possible thing for the value we're able to deliver."

That's actually true from what we hear. For example, Office 365 is getting issues fixed in Exchange that have been around for a while. They're not bugs that affect you if you're running it on premise because the exact set of circumstances that trigger the bug just don't happen. Move Exchange to the cloud and they happen frequently, so the Office 365 team is going back and getting the Exchange team to fix the underlying issue rather than working around it.

It's the same as the teams at Microsoft that run IT not just for Microsoft but for customer like Coca-Cola; having a customer who can get the developers to listen ought to make for fewer bugs. That can be a slow process, from what Kristina Ashment, the Senior Program Manager on the Configuration Manager team, was saying at MMS this year. In the past Config Manager has use the WMI repository in Windows to store information. That's been having problems but as she commented "that team doesn't necessarily believe they're the problem". (Incidentally, if you've seen something similar she's collecting data about when repository corrupts on hard reboot, and asking the WMI team to publicly release the API they're using to verify the repository in Windows 7 as well as looking at using a different repository.)

Cloud services run at scales few enterprises can match; that means rare bugs will be seen more often. And with the attention that cloud - and cloud outages - are getting, problems will be getting fixed as fast as possible. As long as the server products get the same fixes, cloud really could make on-premise IT better as well.

Mary Branscombe

Editorial standards