I currently use those very nice people at CobWeb for my email service. I have a singe user account, utilising Exchange 2007 (2gb mailbox) which costs a mere £6.00 a month. The service from CobWeb, for mail, has been outstanding – not a single noticeable glitch in over a year. It took a couple of days to get mail flowing when I first signed up, due to long DNS, but that was quickly resolved. Since then, it’s been flawless. And that’s the level of service that I guess I expect. Not being able to get to my mail is, as someone who is self employed, simply bad news!
I’ve been playing with Office 365 and am very impressed with the overall service – for not much more money than I currently pay, I could move up to a 25gb mailbox (not that I actually need such a large mailbox, but it’s the principle!), and get SharePoint and Lync thrown in. It’s Lync that is especially interesting – the new Lync client is so good, that I’m tempted to move just for that!
Then I read this: http://www.bbc.co.uk/news/technology-14851455.
While this explains the downtime, I suppose you could argue that a couple of hours downtime is no big deal especially as it took place overnight. But look at the Office 365 service’s reliability record since GA at the very end of June – it was down briefly in August, then again this week. In just over 2 months of live service, two outages in 2 months.
Of course, Microsoft is not alone in having challenges. Google had an outage this week as well. And in April, Amazon’s infrastructure failed taking down a number of sites, including Foursquare.
So is this latest outage just a bit of teething/growing pains on the part of Office 365 or something else? I remember the trials and tribulations Demon had in the very early days – scaling massively a service that’s growing massively is very hard work! So, I suspect it’s a combination of factors. Certainly, developing and delivering scalable and highly reliable solutions, especially to Internet scale, is just plain difficult. Things that probably shouldn’t go wrong do – at least until there’s enough experience to make those problems a thing of the past. So I feel that at least a goodly portion of the ‘blame’ must lie in growing pains – which one would expect (hope) die off. Certainly taking the latest and greatest versions of the software has risks. Cobweb is still running Exchange 2007 while O365 is using Office 2010 and related servers. But the features are so much better, I hear you say!
In my view, Microsoft released Office 365 a tad early. The outages, the problems with federation with Live, and of course the still missing PowerShell cmdlets for Lync Online and SharePoint online do not make for a perfect story. And I’d really have liked to see a full voice solution in terms of Lync online (but I know that such a feature is likely to reveal a number of challenges both technical and legal). The outages (yesterday’s and the one in August) and lack of tools would have been noteworthy but not complaint worthy had the service still been in beta. Maybe Microsoft should have considered a longer beta? I certainly think another 6 months would have been appropriate and might have enabled Microsoft to get better at running this vast service.
So what’s to be done. Certainly, as the BBC article points out, “the number of high profile failures have dented confidence in cloud computing”. But I do believe that eventually, MIcrosoft and the rest of the cloud vendors will get it right and we will see cloud computing as an everyday thing. I suppose it pays to be cautious. I’ve put on hold my plans to move my business email to Office 365, but I’m watching things carefully.