The usefulness of an SLA is not the refund to the customer. It is the cost to the provider, which if painful enough, assures the customer that the provider has incentive to keep the service up.
This blog post says, in so many words, they would rather not be bound by such incentives. (and they are seeking customers who have no such requirements.)
Sort of... the point was more that the promises are pretty hollow in the first place. We don't need financial penalties being applied to motivate us to keep our service up, at least beyond the natural loss of customers if we were to be down for an extended period of time. I think offering anything beyond that, for a service as low-cost as ours, would be lying to our customers.
Agreed, and lying is never a great idea. But you also specifically noted that you do not invest in the kind of solutions that help ensure better uptime stats. And other sites that have failover and other HA techniques are not making empty promises. They are showing that they intend to try harder, and your blog post so much as admits they really do try harder (with more expensive architecture, and the willingness to be penalized on failure.)
So, regardless of the icing on the cake, you seek customers who do not rely on uptime. That's a valid choice, and I mean no disrespect. I'm just calling it what it is.
I suppose you're right. We felt that it was more worthwhile to price our services lower and avoid going overboard with HA, because our application is arguably not "mission-critical". If it's down for awhile, it's bad, but no one dies and money is probably not lost (at least not a lot).
The investments we do make in infrastructure are geared more towards disaster recovery, because if we went down and lost our customers' data, it would be a much more critical failure on our part.
5 comments
[ 4.3 ms ] story [ 22.5 ms ] threadThis blog post says, in so many words, they would rather not be bound by such incentives. (and they are seeking customers who have no such requirements.)
So, regardless of the icing on the cake, you seek customers who do not rely on uptime. That's a valid choice, and I mean no disrespect. I'm just calling it what it is.
The investments we do make in infrastructure are geared more towards disaster recovery, because if we went down and lost our customers' data, it would be a much more critical failure on our part.