Got bored... future AoV server upgrade

Red Squirrel · Post by **Red Squirrel** » Sun Jun 21, 2009 9:33 pm

Once the revamp is over and I start working on ServerWeave, I will then need a bigger server so I can do web hosting, which will bring in money.

I found that liquidweb colocation costs 300/mo and allows 3 tower servers, right now I'm paying 149/mo (that's USD so it comes up to around 160-180).

With a server this powerful, renting costs aren't realistic and colocation is better. Renting a server of this power would be like 600/mo if more. The beauty with collocation is I could later on send them another server.

For kicks I decided to put together a box to get an idea of pricing if I was to do it now:

In the end this is the specs:

* 8x 1TB hard drives (some web hosts offer 999999TB for $4/mo, I need to compete somehow

)

* 4GB of ram, with capacity for 24GB

* 2GB lan

* cheapo video card (no point in getting a huge high heat generating vid card for a server)

* 8 cores of processing power at 2.66GHz each!

I would probably raid 2 drives using the built on hardware raid if it's any good just so my OS is redundant. Then I'd make either a raid 10 or 5 volume with the other 6 drives. All 8 drives are hot swappable which means no downtime should one fail.

Later on if I was to add more servers I could add a gigabit switch in the mix and use the 2nd port to have a private network between the servers for very fast backups between servers.

Just imagine though, 8GB of ram! Right now we have 2GB and to go 4 would cost like 50 bucks extra per month!

Of course I'm just dreaming here. The one time cost to build this is not THAT bad as I could just save it up, but the 300/mo is still steep for our needs. Later on though, maybe. It's fun to dream.

Putting this together was better then watching porn tbh.

Archived topic from AOV, old topic ID:4714, old post ID:30650

jrhather · Post by **jrhather** » Tue Jun 23, 2009 8:52 am

Red Squirrel wrote:Putting this together was better then watching porn tbh.

Wow.

Archived topic from AOV, old topic ID:4714, old post ID:30653

dprantl · Post by **dprantl** » Tue Jun 23, 2009 11:39 am

You have a long way to go padawan.

Archived topic from AOV, old topic ID:4714, old post ID:30654

Red Squirrel · Post by **Red Squirrel** » Tue Jun 23, 2009 12:28 pm

Nice ESX farm. Ours (at work, not AoV lol) is about 4 ESX hosts and approx 70 VMs. We are running on the edge though. There's a total of 16 cpus and 84GB of ram. One of these days we will add more servers. They are installed, but there's lot of politics and BS slowing things down.

At home on my current server (quad core, 8GB of ram) I have about 7 VMs running. It only slows down if I power on my vista vm. Other then that it's solid. Of course, there arn't 100's of users hitting up the mail server every second.

Oh and I just realized our uptime on our ESX are all 129. It's a rule that we need to have some huge major unexpected outage at least every 100 days. Maybe the 2nd ESM card in our san will fail. it's 26k for a card like that so they did not want to spend that much to replace the other one (2 redundant per shelf).

Archived topic from AOV, old topic ID:4714, old post ID:30655

dprantl · Post by **dprantl** » Tue Jun 23, 2009 12:35 pm

Our bladeserver chassis is only half full and so is the SAN. It's also sitting pretty idle right now since we only have ~30 VMs running when we could have over 100 without running out of resources. The only reason those uptime numbers aren't years is because we had a huge power failure in the city and coincidentally we had a problem with our backup generator at the same time. We only had a couple of hours of battery backup so everything was shut down. You don't want to wait for the power to go out on a bladeserver/SAN.

Archived topic from AOV, old topic ID:4714, old post ID:30656

Red Squirrel · Post by **Red Squirrel** » Tue Jun 23, 2009 1:38 pm

dprantl wrote:Our bladeserver chassis is only half full and so is the SAN. It's also sitting pretty idle right now since we only have ~30 VMs running when we could have over 100 without running out of resources. The only reason those uptime numbers aren't years is because we had a huge power failure in the city and coincidentally we had a problem with our backup generator at the same time. We only had a couple of hours of battery backup so everything was shut down. You don't want to wait for the power to go out on a bladeserver/SAN.

Been there done that lol. We have a crappy IBM san and whenever it goes down, it takes days for it to come back up.

Archived topic from AOV, old topic ID:4714, old post ID:30657