Attention to detail is a must when dealing with cloud computing vendors
Cloud providers and some analysts claim that using cloud application, infrastructure or platform services (SaaS, IaaS or PaaS, respectively) is the most cost-effective choice for every IT organization. However, the success of any cloud implementation depends greatly upon the specific services used and the terms and conditions specified in underlying contract service level agreements (SLAs).
Unfortunately, too many executives sign contracts without reading them carefully. Executives must learn the definitions of cloud contract service levels as well as understand the implications of the terms and conditions, which can vary significantly by provider. Failure to do so could lead to considerable damage to a company’s bottom line and even its reputation.
Two great examples of risk exposure come to mind. One U.S. cloud vendor found itself in deep financial trouble and decided to declare bankruptcy. In so doing it told its customers that service could not be guaranteed going forward; however, the contract terms and conditions stipulated that its customers were required to pay for the cloud services, even if no services were being provided. In a more extreme case, earlier this year, U.K.-based 2e2 declared bankruptcy and was put in the hands of an administrator. Within a week of gaining control, 2e2 data center services customers were asked to collectively cough up nearly $1.5 million to keep the lights on. Apparently the administrator was running out of cash to fund the operation and 2e2 was losing $450,000 per day. Individual customers were asked to continue to fund operations by providing funds ranging from $6,000 to $60,000. Not only were these payments over and above the contract payment rates, customers received no guarantees that their service would continue or that there wouldn’t be additional rate hikes.
In addition, one can easily find news stories about cloud services outages from well-known entities such as Amazon Inc., Microsoft Corp. and Research in Motion, Ltd. (aka BlackBerry) that greatly impacted the ability of users to conduct business. The Amazon AWS (Amazon Web Services) outage lasted five days, but Amazon claimed the outage was caused by a problem outside of AWS itself so the SLA did not apply. The company later agreed to compensate for the outage but this was a P.R. move, not a legal obligation. In other instances of cloud outages, actual databases were irretrievably lost.
Given the potential for things to go wrong, it’s incumbent upon executives to study the nuances of the contract clauses. The deal you sign with a cloud vendor must be one your company can live with over the life of the contract.
The Magic of SLAs
Most SLAs range from 99.95 to 99.995 percent availability, which means that over a calendar year (24x7x365), the vendor is allowed 25-250 minutes (or up to about four hours) of downtime. Many cloud vendors set their outage SLA across the year’s service, meaning that even in the event of a four-hour outage on one day, the provider is still meeting its service objectives. Customers should narrow down these SLAs to weekly or monthly commitments (or in the very worst case, quarterly).
It’s important to understand that the scheduled downtime cloud providers desire to perform updates and conduct other housekeeping generally isn’t calculated into uptime requirements. Thus, availability isn’t as great as it’s advertised to be. In standard contracts cloud vendors usually request a minimum of 30 minutes of scheduled downtime at some time across a three-hour period, which may or may not be convenient for the customer. Some vendors allow themselves the right to interrupt service to conduct “scheduled maintenance” at any time of day should an emergency condition arise.