Why Uptime Guarantees are Ridiculous
Last week ServInt released an updated SLA that covers all of our products, from VPS to dedicated to cloud. And like everyone else, we laid out the “uptime guarantee” for network, servers, support, etc. This guarantee, though, is simply a threshold, if your service dips below which, you may request hosting credits.
What’s ridiculous is the way some hosts – and some industry “experts” – glibly refer to uptime guarantees as if they were some sort of literal guarantee of future performance: “Wow, they’re offering five nines in their SLA,” “Did you hear about the host that guarantees 100% uptime?”
Uptime guarantees don’t promise what percentage of the time your server will remain online without network disruption, and they are not evidence of future network performance.
An uptime guarantee is – no matter which host you look at – simply a promise of what refund the host offers customers if there is a network outage.
And every network—even the most robust, redundant networks—at some point will experience an outage. Our last network outage was in 2004.
The question is not: Will my host have an outage in the future? They are: How likely is it that my host will be the next to experience an outage? and, How quickly and efficiently will they respond and fix any problem that occurs?
Uptime guarantees offer absolutely zero insight into these questions. At best, they show the hopes and wishes of the host’s marketing team.
Instead of trusting in the hopeful promises and guarantees, a better judge of network performance for a web host is historical uptime. Historical uptime is a good judge of the quality of the host’s network, the theory being: if it hasn’t gone down much, it probably won’t go down much in the future. But where can you find reliable information on historical uptime? ServInt’s historical uptime if 99.9996%, but independent verification of such claims is always warranted.
Unfortunately, reliable historical uptime data can be hard to find. Most third-party aggregators of uptime data are actually pay-for-play marketing lists.
And for the few sites that do not charge, they are based on the uptime of a few servers, not the entire network.
The only truly unbiased, reliable way to gauge the historical network uptime of a host—as well as the quality of the customer support during an outage—is to investigate hosts on a webhosting forum such as Web Hosting Talk, one of the largest of the bunch. Simply assemble your list of potential hosts and execute a few searches across the forum for terms like “outage” and “downtime” along with each host in turn.
It’ll take a while to read through the posts – and you’ll have to do a little sleuthing to determine which issues were actually due to user error (overloaded server, bad code, etc.) – but you’ll amass a wealth of information and customer opinions on the frequency and severity of network outages as well as the quality of support the hosts offers during trouble.
And in the event that you do experience network or hardware downtime with your host, your “uptime guarantee” in your SLA is your recourse for compensation. Hopefully, you’ll never need to know the details of how much uptime actually qualifies as less than 99.9% or 99.999% or whatever. But if you do, a good SLA will clearly state how uptime is defined and what you’ll receive if the “uptime promise” is not met.