Most of the time we hear about cloud service providers offering high uptime SLA’s in accordance with customer system requirements. However in the latest survey done by Gartner, they reported two of the TOP Cloud providers Amazon and HP offering worst SLA’ s ever not meeting the base line requirement of their customers as SLA’s have been manufactured to benefits only customers with high CLOUD setup.
Gartner report for Amazon and HP Cloud Service SLA’s
“In the AWS case, a region is considered unavailable if you’re running instances in at least two AZs (availability zones) within that region, and in both of those AZs, your instances have no external network connectivity and you can’t launch instances in that AZ that do; this is metered in five-minute intervals,” Leong said. “In the HP case, a region is considered unavailable if an instance within that region can’t respond to API or network requests, you are currently running in at least two AZs, and you cannot launch a replacement instance in any AZ within that region; the downtime clock doesn’t start ticking until there’s more than 6 minutes of unavailability.”
Now here Instance is the term used for a Virtual or CLOUD server that is setup in one availability zone. SLA is not designed per instance, where as it is instance which gets down and not the AZ (Availability Zone).
Single instance down does not keep much value as another instance is also required to be not working in another zone to qualify for a downtime, well that surly sounds interesting, isn’t it to add a charm let say clock will only trigger after 5 min. So till 5 min we certainly need to try to clone a new instance in the same or different availability zone. This is what I call not really user friendly SLA.
Possible Reason “ As a service provider they are entitled to provide an Infrastructure to host your environment this IaaS is termed as availability zone and if AZ is not down then their IaaS is running fine.
Instance on the other hand considered as an application running on a server, which is configured by user. However the catch is till the time it does not get clear about why an instance was down, because of internal application or hardware error downtime can never be decided and owned by any of the parties.
Service providers are not just to offering value for money, but also be looking to have some generated for themselves. With SLA’s like above it encourages the SMB’s and Mid Tier to go with higher number of instances to be on safer side and all covered up. With Pay as you go model the value of service has increased but in comparison to non cloud hardware setup.