Find Part 1 here, where I discussed some basics on uptime/downtime, establishing that cloud vendors generally provide a certain “guarantee” about the uptime with which their offering operates. I am now going to discuss the “so what” – what happens if that uptime guarantee is not met, i.e., the remedy.
The short answer: Credits. These offerings are generally priced on an annual basis. That fee can be divided into months, and the vendor will issue a credit against that monthly fee based on excessive downtime suffered. The uptime service level agreement (SLA) will have bands of uptime – the less uptime (more downtime), the greater the credit.
Two issues that I will speak to a little deeper. The first is credit vs. refund. Refunds are pretty straight forward. Credits can get a little trickier. I have seen some vendors issues credits which can only be redeemed on renewal. This begs the question – do you want to renew a contract with a vendor who just suffered significant downtime? Probably not.
The second issue is the “completeness” of an uptime SLA. I have seen SLAs that have maybe three bands of uptime, and the band with the lowest downtime (e.g., <90%), results in a 25% credit. The implication with this? The vendor keeps 75% of their fee even if uptime is well below the threshold of that band. Imagine having a critical event, and the application is effectively down for during all business hours for two weeks, effectively half of a month. Despite 50% downtime, the customer could only get 25% of the refund.
This isn’t true for all cloud vendors. I have done surveys based on publicly available SLAs, and there are many companies who offer “complete” SLAs, where the lowest band of uptime results in a 100% refund. These companies include large operators, who you might think otherwise have the leverage to force an incomplete SLA.
Whether a customer decides to accept an “incomplete” SLA is, of course, a business decision, but I can’t help but think that a vendor’s SLA can be extrapolated as to how they manage their business.
I think I have at least one more part to this series on uptime. If you got this far, thank you. If you have any thing you’d like to know more about, please drop a comment or email me.
Leave a comment