I’ve been kind of taken aback by the vehemence of some of the comments from customers about the recent troubles that Dreamhost has been having.  To hear some of these people, it sounds like they’re losing thousands of dollars per minute when their sites are down.  But if that’s the case, I’m really curious as to why they’re betting their business on a shared hosting plan. 

Even at the rate I’m paying (I’m on the “Code Monster” plan, and was migrated there from my original plan, so I’m not paying the full rate, but it’s still about $20/month), I’m not going to get too torqued about occasional downtime.  But perhaps I’ve not had as bad an experience as some of the other customers.  I’ve been with them since April, 2000, and I just haven’t had the same kinds of problems that others are chronically complaining about.

In my professional life I’ve worked on “industrial strength” websites.  These sites have hosting bills in the range of $100K to $200K PER MONTH, depending on complexity and transaction load.  And those kinds of charges still only get an SLA with 99.5% uptime (a little over 40 hours per year downtime, due to scheduled maintenance windows).  But you do get professional admins who you can page at 3:00am if the site is down, DBA’s who know what the hell they’re doing (usually), backups, enterprise class hardware with techs who are available to come onsite 24/7/365, load balancing, clustering, and so forth.

If you’re going to bet your business on a web hosting system, you’d better be prepared to spend real money.  If you expect 99.999% uptime (FYI, that’s five minutes per year downtime),  you need clusters of clusters across multiple datacenters with redundant databases and hot-failover (just to name a few buzzwords).  That $15/month you’re spending for shared space on a single PC running Debian Linux just isn’t going to cut it.