I’ve been using 1and1 shared hosting for about five years. Until last month it was great; fast, cheap, reliable and when I did have a problem it was fixed quickly, usually in less than an hour.
January thorough August this year I averaged 2 hours/month downtime as measured by host-tracker.com. That works out to 99.7% uptime which I consider pretty good for shared hosting. Then things started to go downhill. September saw nine hours of downtime and in the first six days this month WapReview.com has been down 22 hours.
Yeswap.com has fared even worse. On Sept 20th both sites, which were on the same shared hosting package, started returning HTTP 500 errors approximately every other request.
1and1’s explanation was that I was hitting their limit of 12 simultaneous processes. This didn’t sound right because my traffic today is half what it was in March when Chinese users discovered the HTML to WML transcoder on Yeswap.com and started to hit it with 50,000 requests a day, presumably to get around the “Great Firewall”. After a month of that, with the site handling the traffic with no signs of stress, Chinese traffic dropped overnight to about 50 requests per day. I guessing that yeswap.com got blocked in China too. Why am I hitting the limit now when I wasn’t with 2X the traffic, I don’t understand. The only logical explanation is that the 12 process limit is new.
I tried to solve the problem by purchasing a second shared hosting package from 1and1 and moving yeswap.com to it. I figured that by isolating the WapReview blog traffic from yeswap.com, both sites would have more headroom and perform better. At first, it seemed to work. I checked several times a day for a couple of days and all seemed well. Then I got sick and was unable to monitor the site for a while. Last week I took a look at yeswap.com again and discovered that the dreaded error 500 was back and worse than ever. The site varies from being completely down for hours on end to working approximately every other request. 1and1 has been “investigating” for 72 hours with no resolution or explanation.
I tend to believe that yeswap.com really is attempting to use more than 12 processes. 1and1’s PHP implementation spawns a new process for every request so the limit translates to 12 simultaneous requests. Yeswap gets a lot of requests, about 40,000 daily pageviews, but the pages are small, under 3KB, and most are pulled from a cache with no database access needed.
At this point I’m looking for recommendations for a new hosting service. 1and1 has become unreliable and the tech support structure seems to have collapsed under the load. Issues which used to be resolved in a hour or less now go unanswered for days.
My requirements are PHP 4 or 5 and MySQL support and a monthly cost of $20 or less. The sites to be hosted are this WordPress blog which averages about 5500 daily PV’s plus yeswap.com, a high volume (40,000 daily PV’s), but low bandwidth (no images, page size under 3KB) mobile web site.
I’m leaning toward Dreamhost’s shared hosting based on the generally good reports I see about DH on the web and the fact that they claim not to limit the number processes and to have “flexible” limits on CPU utilization. I’m also intrigued by Media Temple’s Grid Server which sounds good on paper; a cluster of servers connected to a SAN with traffic continually load balanced to send requests to the the least loaded server. It sounds like a smart architecture but I’m a little leery as I’ve seem a number of complaints on web forums about excessive latency and database issues with the service.
So I’m looking to the lazy web here. If you have any experience, good or bad with Dreamhost or Media Temple I love to hear from you. Or if there’s another other shared hosting provider you really like, please tell me about it.