HostPC rocks.

[dbmasters]

Once again, it's time for a server move. Server moves tick me off every time, I get short tempered and cranky, combined with a holiday season I despise this year as it's the first one without my mom...

But Todd really stepped up and helped me with the move when my patience and attitude were at all time lows.

I give Todd lots of crap in IRC at times in good fun, but I gotta say, him, Sean, Ryan, Joe and the rest of the HostPC gang really do a bang up job. Even if it's an hassle of a time or whatever, they do the right thing, and do it in a timely manner. I am always glad with server moves after they are completed, but get real pissed during them.

Hats off to you guys for a Christmas present of help when I needed it. It hasn't been a stellar holiday thus far, but Todd, you rocked it for me.

:love

Only more cowbell would help make HostPC better. :D

[admin]

Thanks Dan...

Server moves are done rarely, and after great thought and planning. We only schedule a move for the following reasons:

Change of provider (hasn't happened in 4+ years).
Age of server - servers are retired after 2 yrs due to the amount of downtime that would be necessary to upgrade hardware in older machines, replace drives that have spun constantly for 700-1000 days. I dont take chances on drives. Those of you with us from the very beginning will remember the Ev1 servers (then Rackshack) days when drives provided by them died religiously every 30-60 days. I swear they were on a timer. Now we use NEW drives, but I dont trust my home computer to 2yrs of drive time, I can't trust servers to do much better. Hardware dies, it's a fact of life - cycling older machines (in this case the machines are 3+ yrs old) has become a good working practice.
Drive failure. This is rare, but it CAN occur any time. See #2. Normally, we pop a new drive in, restore data and go forward. Sometimes customers find it faster to restore to a new server - rather than wait for a reformat of the old - resulting in a technical "server change"
Just so we can make life miserable. If you believe this one, I've got some beachfront property in Utah I'd like to talk with you about.
In the instance that's affecting you Dan, it's a combination of issues. First, the server you're physically on is in the old datacenter. Still the same company, but different physical location - we're trying to put all the machines in the new datacenter for ease of management. That server you're speaking about (www5 I believe) is also one of our original servers in Boca Raton - it's actually almost FOUR years old. That's a LOT of spinning of the hard drive. I'd rather not risk your sites to drives that old.

We don't take server moves lightly. We do try to schedule them when it's least inconvenient for ALL users on that box, and when it's likely that the reseller or user will be available to make the changes. Normally we have several weeks/months notice before we do routine maintenance.

But Joe, why not raid every server? RAID, in theory is fantastic, in fact, we do use it on many servers. BUT, it's not a perfect world. When I/We were doing support work for another major hosting company, their RAID arrays were constantly broken - resulting in backups 4-6-10 months old - and of course it was never noticed until the master drive died. Linux software RAID stinks for shared servers. It's too much data to toss across the array constantly - and it pukes out. When it is working, it creates a noticable load on the servers - something else we try desperately to avoid. Our backup servers, file servers and dedicated application servers all use RAID - some are now on RAID10 and it works great. In this shared environment, it's not a reliable option. Joe knows RAID, but in this case, I dont see it as a reliable alternative to what we do routinely now.

Thanks for the comments Dan, it's much appreciated.

Joe