Goodbye Media Temple, Hello WebHostingBuzz

Media Temple

I've been hosting with Media Temple (MT) for about a year or so. The idea that my instance is on a cluster was fancy and awesome. Static pages being served from Media Temple are blazing fast. However, using the default MySQL container proved to be mediocre, if not sub-par. Couldn't take this. So I upgraded to a dedicated MySQL Grid Container, essentially a dedicated instance of MySQL also on the cluster. Performance dramatically increased compared to the old set up. I was happy for a few months.

Then one day on my trip to Mammoth (Jan 28th, 2011), things stopped working. Support wasn't sure why my container essentially "took a dump". They told me this was the first time they'd seen something like this happen. Restarting the container allowed it to come back long enough to back up my files. Their reasoning was that I was hitting resource limits on my MySQL container.

Ok. Fine. But that doesn't make sense!

I hadn't done any active development that would have caused things to change/spike. My sites were essentially functionally static for about two months so I'm not sure what really changed. I wasn't going to accept the answer that the problem "built up over time" either since restarting the MySQL container should've resolved this issue for at least another month. But it was within hours of restarting that the problem resurfaced and things went down. Even the initial restart had issues and required supports intervention.

So I started taking down all of my database driven sites and left three of the most critical ones up. Still failed within hours, if not sooner than previous failures. After fidgeting with a few settings to no avail, MT support decided to temporarily upgrade my MySQL container to the next package up. The upgrade was free for the rest of the billing cycle.

At this point, I was already backing my stuff up and trying to figure out which hosting company would fit my needs. I wasn't particularly happy that they took so long to respond to email. The phone support I received really wasn't that great either. They were definitely nice people though, so I have no complaints there! But really, I couldn't stomach the new costs. With the upgraded container, my hosting costs would've ballooned to $95/month, up from $40.


I had been hosting some smaller sites with a developer's package with WebHostingBuzz (WHB) for awhile and hadn't had any issues. They're a relative newcomer and came with SitePoint's recommendation. So I shopped the idea of shared hosting/reseller's packages on Hostgator and WHB, cloud computing with Rackspace and Amazon, and VPS with Rackspace, MT and WHB.

I decided against MT due to the price as well as delayed and unsatisfactory email responses.

I decided against Rackspace because of their limited selection for what I was looking for in terms of price range. Though, I have no qualms about their tech support. I've dealt with them in the past and they've always been great, living up to "fanatical". But it was always my employer footing the bill.

I decided against shared hosting/reseller packages in general because I needed more control of my own server without having to pay for a dedicated physical server.

Ultimately, I chose WHB's oVZ VPS Silver package because it was cheaper than everything else and also offered Cpanel. After the second day, I noticed some quirkiness with latency and unresponsiveness. Did I just gun too quickly and chose oVZ over Xen purely because of price?

I did some additional reading about the differences between oVZ and Xen. Although things cleared up and support was able to verify my claims (a mild disappointment), I had had enough crap dealing with my sites being down that I wasn't going to take a chance. oVZ's memory management was not ideal for me. Xen I went.

What really sold me with WHB was how well they handled my concerns and my upgrade from oVZ to Xen. They answered all of my questions before, during, and after the transition. I was definitely happy with the timeliness of their support and definitely happier with the performance boost my database driven sites had gained.

After the fact, I did some load time tests between WHB and MT. Note that MT (Culver City) is down the street from me, with ping latencies between 15ms and 25ms. WHB was across the country (ATL, Georgia), with ping latencies ranging from 70ms to 90ms. Although WHB's servers expectedly took longer to load pages, the database driven sites definitely felt snappier. Onload times however, were pretty much neck and neck, MT averaging a better time by about ~.5s.

~.5s "slower" for what felt faster and is $38 cheaper?


Proudly hosting on WHB now.