The Cloud – Monkey (from Monkey Magic) had one, should you use it too?
Has your CIO/CEO/IT Manager done this?
Do you trust the cloud?
I would be surprised if you whole-heartedly said ‘yes’. Firstly because you’re talking back to a blog post which is quite strange behaviour, but secondly because there’s a lot of media attention going on in this space.
Just to rehash the last week, there were two major events, one from Google and the other Microsoft.
Wednesday 8th September (ish, it’s hard to gather what timezone they’re all talking about) saw a Google Docs outage. The outage lasted 52 minutes: 23 minutes from being alerted to kick off a rollback proccess which then took 24 minutes to do. Add an extra 5 minutes – the time it took for “the additional capacity restored normal function”.
The cause was due to a change they had implemented to improve real time collaboration, but the heavy load of the real world exposed a memory management bug.
Wednesday 8th September again (although later in the day in America, so the 9th for Aussies) it was Microsoft’s turn. Office 365, Dynamics CRM and some other non-enterprise level services (Hotmail, Skydrive, Live stuff) were down for a few hours. This one was not as clear cut – the outage itself was for the North American data centres, which meant genereally that Australians were fine as we use another based in Singapore. The fix was an update to DNS servers, which means we all have to wait for replication of the new records around the world before everyone everywhere is without issue.
The cause for this is a bit less detailed than Google’s, with ‘DNS issues’ being claimed as the cause.
That’s scary, how do I cope with these outages?
So, would your business complain about these outages? OK yes, you probably have someone who complains about their keyboard clicking too loudly when they type, so of course someone will complain about this.
If you wanted to jump into the cloud, I’d suggest to look at a hybrid solution. This isn’t news to many readers I’m sure – multiple paths of redundancy for “everything” which includes your servers and services. For emails, you can split between your local Exchange (or even hosted) and Google Apps. Postini replays everything that goes through it to Google Apps, so your users can jump onto Google Apps in the event of an Exchange outage and continue working. Then, you’re not missing out on the feature rich options of Exchange, but the business critical emails have full redundancy.
The real summary here is – go ahead, use the Cloud. BUT – do what you should already be doing (i.e. redundancy, are you paying attention here? Good.). A single provider in the cloud is not reliable enough at this stage to be trusted for it’s own inbuilt redundancy. Trust two clouds, or one cloud and the other half on-premesis.
If you’re a small business with a portaloo full of staff, then it becomes much harder to justify. Also, is there a manual process that can be used in the event of a service failure? Business Continuity should dictate what’s required for redundancy. Maybe writing things down for a day is a completely valid way of coping with the outage, with little or no loss to the business? There’s no reason to spend money on redundancy in that sort of situation.
Google Apps and Office 365 both guarantee 99.9% uptime – sounds great, but that’s ~42 minutes a month they’re allowing for. How much would your business lose if nobody could do their computer based work for 42 minutes during the day, every month on average? Over a year, that’s slightly over 1 full working day. If the cost of that outweighs the cost of getting a second cloud service or some other means of redundancy, then it’s already paid for itself. Getting a service refund after the event isn’t really what people care about, they just want it to work.
Google Docs Blog – http://googledocs.blogspot.com/2011/09/what-happened-wednesday.html
Google Postini: http://www.google.com/postini/continuity.html
Windows Blog – http://windowsteamblog.com/windows_live/b/windowslive/archive/2011/09/08/current-hotmail-and-skydrive-issues.aspx
Office 365 Tweet – http://twitter.com/#!/Office365/status/112008132443648000
ZDNet Microsoft Outage – http://www.zdnet.com/blog/microsoft/outage-hits-microsoft-crm-online-office-365-customers/10359