MediaTemple GS - Problem Solved
Originally published at ArticleSnatch Blog. You can comment here or there.
When I wrote the first post, my biggest concerns were that this sort of thing was happening regularly and that a lot of customers were affected. After the site was recovered and fully functional again, I had the pleasure of speaking with someone at MediaTemple who was willing to listen to my concerns as well as answer a few questions. What I found out from that conversation made me feel 100 times better about the MediaTemple GS service than I did before.
My main concern was that this downtime was widespread, it turns out that while most customers experienced intermittent periods of 30 minutes to 60 minutes of downtime, myself and maybe 2 others were affected for a long time. That's not too bad considering the 125,000 domains that are hosted on the Grid Service. Sure it stinks that our article directory was down for so long, but hopefully thats a one off occurrence and won't happen again.
My other concern was that the times when I tried to reach customer service, it took 15 to 30 minutes. Fortunately the phone system at MT will tell you how long the expected hold time. I found that these hold times are not normal and for most of 2007 they had an average hold time of about 3 minutes. Which is a lot more what I was expecting. Since the incident I have not been on hold at all any time I called in. I spoke to them about this and was told that MediaTemple has been growing rapidly and they are always hiring and training new customer service employees. Also since I called when things were down, that was the most likely time for other customers to be calling as well.
My last major concern was that something like this would happen again. I was told that they have been working on the cause for the downtime to correct it so that it won't cause the problem again. Additionally he said that they are constantly improving the way that the Grid works, so as to minimize problems in the future.
Media Temple made things right for the downtime - they gave me a discount on a DV (Dedicated Virtual) server so now our other site is no longer hosted on the Grid. It was a lot easier than I thought to move from one to the other. I just had to rsync the files and mysqldump the databases, update the connection string and that was about it. Pretty simple out of the box.
The DV servers use plesk control panel to help configure everything - I was surprised at how simple it really was, as this was the first time I used plesk. I realize now why resellers use Plesk to setup domains - it makes a lot more sense and there are tons of controls.
In the end what started out looking like a very negative situation turned into a positive experience. I guess sometimes its how a company handles the problems that matters more than the problems themselves.
As for the other things in the previous post - Mosso has also dropped their request based billing to go to a billing based on CPU and disk I/O usage. We will have to wait and see how it goes - during June.
