I got an interesting perspective on cloud computing service levels from a member of CHASE recently. I said that I anticipated that SLAs would increasingly become a headline issue for cloud computing as more corporates were looking at moving key services to the cloud.
His response: SLAs are simply not a practical way forward. Cloud suppliers are unlikely to offer meaningful compensation for loss of key business systems. The best route is to have two different cloud providers, and some form of fail-over between them. Apparently this is technically challenging, but can be achieved in a lot of cases.
Another attendee at the event had found that SLAs were not always all they are cracked up to be. He had a 99.9% SLA on power availability, but found when his servers were power-cycled twice in a week, causing chaos for the web site they were hosting, the compensation was miniscule because the actual downtime was only brief.
The moral - treat SLAs with caution! Do they really achieve your business objectives?
Comments