By Oscar Arean, Technical Operations Manager, Databarracks
Service level agreements (SLAs) are meant to ensure transparency and set standards between a customer and its service provider.
Too often though, contracts are full of legal statements that have little, if anything, to do with the actual service a customer receives. SLAs are the most important part of a contract because they outline exactly what the service provider is prepared to commit to. There is no room for sales or marketing-“fluff” because if SLAs are not met, the customer can seek compensation.
So for SLAs to have real value, it’s important that they are realistic. This is never truer than when buying cloud services.
Cloud adoption in the UK is growing rapidly. The latest figures from the Cloud Industry Forum (CIF) show 78% of UK organisations have now adopted at least one cloud based service. As adoption continues to rise in 2015, organisations need assurances from their cloud service provider (CSP) regarding the standard of service and level of uptime they can expect.
Cloud SLAs should be based on specific business needs, with performance metrics that actually matter to the end-user. This can include the uptime of a service or how long it will take to resolve common issues. Bear in mind that some SLAs, like data centre availability or uptime, may not be negotiable and will likely be standard for all customers of that CSP. It may be possible, however, for service providers to improve on their standard SLAs, though this usually comes at an additional cost.
For cloud contracts to work, providers have to set realistic expectations around the level of service they are able to provide. By failing to establish standardised SLAs within the cloud industry, a lot of CSPs have been using vague and ambiguous terminology to offer unachievable levels of service in a bid to gain leverage in a crowded market place. They offer 100% SLAs — knowing that they won’t achieve them, but on the basis that it will be cheaper to pay out to customers for missing the SLA on the occasions that they do.
All this does is leaves customers ill-equipped to deal with unexpected outages. If CSPs are transparent about the risks, customers can put the appropriate precautions in place.
As a customer, you might take the view that a service provider prepared to commit to a 100% SLA must have a lot of confidence in the service — more confidence than a comparable service provider with a lower but realistic SLA. It is likely, however, that the 100% SLA comes with a very small financial penalty. A service provider with a realistic SLA is more likely to commit to a harsher penalty in the event of a missed performance metric.
Over recent years we’ve seen a number of examples of CSPs offering 100% uptime and then, rather publically, experiencing outages. For these organisations, such incidents have left their reputation in tatters. Ultimately, end-users want honesty and transparency from their CSPs.
100% uptime usually translates as 99.99% or 99.9999% — this could be the difference between 52 minutes of downtime per year against five minutes. On the surface this difference in uptime may seem minimal, but if SLAs are being advertised at 100%, that is what you expect to receive. Any less than that and customers are being miss-sold the service they are paying for.
Cloud users need to ensure that they are very clear about what they require from the service, but equally CSPs need to ensure they are transparent with what it can and cannot offer. In doing so, clear SLAs can be defined, that ultimately contribute to a stronger, more coherent working relationship between both parties.