When evaluating a move to cloud-based services for regular business activities, for some systems the benefits will easily outweigh the loss of control. But for others, keeping in-house may be the optimal solution for quite a while yet
As cloud computing trends gain momentum, Gartner has a timely article on considerations when making the move: How long does it take to reboot a cloud? Importantly, it highlights that there are both advantages and disadvantages to switching to cloud-based services. On the one hand, outage times are out of your control. On the other, you will almost certainly benefit from more fault-tolerance making outages less likely to occur. But then, when they do, it’s probably serious and unplanned downtime may be a lot longer than normal. But again, on the positive side, the service provider almost certainly has a lot more resources to throw at fixing the problem.
Examples Gartner includes:
- Google taking 4 days to restore 0.02% of users to a single service
- Amazon taking 4 days to recover from an outage and unable to recover some data at all.
Recently, Microsoft’s business email service – Exchange Online – experienced performance problems during preparation for an upgrade that unexpectedly delayed email for several hours. Reported by Mary-Jo Fokey:
Microsoft officials warned customers… on Monday, May 10, that an upgrade of Exchange Online was slated to begin on May 12. Microsoft didn’t tell users to expect any downtime as a result of the upgrade… But it seems something went wrong before May 12’s upgrade ever began.
Whilst gaming rather than business related, Sony’s online player network was taken offline for over 3 weeks after hackers broke in and stole user information.
And mobile devices aren’t immune either as they increasingly rely on cloud-based services. 18 months ago, Microsoft had an outage during a data centre upgrade that lost user data, including contacts, calendars and photos for T-Mobile users with Danger Sidekick devices.
Back in the world of business services, it’s not just security breaches and outages that businesses need to consider. There may be changes to the service that affect how you work but the decision to upgrade or not is also out of your control.
The email notice at the top of this post is one I’ve just received from Google Apps, notifying us that we will be automatically transitioned to a new version of Google Apps within the next 2 weeks. We already tried it once, last December, and rolled back because it means losing a useful feature. Now there is no choice and the feature will be lost.
Based on these experiences, here are some practical questions to ask when exploring cloud-based service to replace traditional on-premise tools:
- What fault-tolerance is provided for our data and services?
- How is sensitive information protected from hackers?
- Where will our data be stored? (if you are affected by local legal and privacy issues)
- What is the average downtime to date, planned and unplanned?
- How long will it take to recover from backup in the event of a failure?
- What is covered by the backup?
- How flexible is the service to tailor to our needs compared to on-premise equivalents?
- How will the services be upgraded and what notice will we be given to test our systems and prepare for any feature changes?
When you get the answers, compare with your on-premise systems before making a judgement.
To give an example for the last question. Google has been sending out notices for at least 6 months with the option to do it yourself, hence we tried the switch back in December. And were surprised to be able to roll back so easily. It would appear that Microsoft gives 2 days notice… we were also BPOS users until March and there was still no sign of getting an upgrade from the 2007 (BPOS) to the 2010 (Office 365) service.
When evaluating a move to cloud-based services for regular business activities, for some systems the benefits will easily outweigh the loss of control. But for others, keeping in-house may be the optimal solution for quite a while yet. Just like those mainframes that didn’t disappear over night when PCs invaded the workplace.
And a hybrid solution may be preferable for some.
One of the challenges of syncing across automated data centres is that a failure can wipe all instances of your data. We use Dropbox to sync active files and make them easily accessible when on customer sites. However we weekly sync to a server that is disconnected from the Internet in between updates. A problem with Dropbox could permeate to all devices syncing with it. In which case, we still at least have a copy stored on our offline server. We do the same with email, diary and contacts.
- How long does it take to reboot a cloud? – Jay , Gartner, May 2011
- Microsoft’s BPOS cloud customers hit by multi-day outage – Mary-Jo Foley, ZDNet, May 2011
- Sony offers free games after bringing PlayStation network back online – Matthew Lynley, Venture Beat, May 2011
- Danger’s servers crashed, and they don’t have a backup – Jason Kincaid, TechCrunch, Oct 2009
Featured image: Storm Front kindly shared on Flickr by L J Mears