SLA in the Cloud

Service Level Agreements (SLA) are nothing new and have existed in information technology for many, many years. Now that the cloud has opened the information utility to everyone, SLAs are out in the open. But what does it really mean? It is not like you have a meter or a gas gauge telling you how much you have used in real time. Also, what actually constitutes an outage? A period of time? As a client, how would you prove it? If there was an outage and the client could prove it, how does the client receive compensation? These questions can be answered in as many ways as there are cloud or cloud application providers. Most have very different answers to how they get to numbers that equate 99.9% or better.

Years ago, a client of mine was a service provider that offered enterprise services hosting. They touted 99.999% uptime — excluding 4-hour bimonthly maintenance windows. Those windows were on Saturdays from 12 AM to 4 AM. This is a period of time that would not be an inconvenience for almost any enterprise, even globally. In actual minutes, 99.999% is just over 5 minutes per year which does not leave much time for anything except a reboot. If the maintenance windows were taken and used, this service providers actual uptime would be closer to 98.9% The addition of virtualization has made 99.999% SLAs a potential reality — for a price.

Having sold “Cloud” based services, I can tell you firsthand that SLA is critical to the selling proposition. The questions on uptime, outages and compensation do not have easy answers. It would be nice to see an industry standard on what 99.999% actually means. Once that is set, the other questions would fall into place.