Skip to main content

Your browser may not be compatible with all the features on this site. Consider upgrading to a modern browser for an improved experience.

View Post [edit]

Poster: KirbyMeist Date: Aug 3, 2004 11:10am
Forum: petabox Subject: Possible solution: Disk Array Rackmounts

I was checking out servercase.com, and I saw some interesting Disk Array Rackmounts. I saw an interesting 29-bay one, for either SATA or PATA configs, but pretty large at 5Us. There's also a 12-bay 2U one as well, although they only have SATA. (you really should be using SATA drives, if you arent, you should, since the thoroughput is much better and the plugs are smaller)

Reply [edit]

Poster: brewster Date: Oct 24, 2004 2:20am
Forum: petabox Subject: Re: Possible solution: Disk Array Rackmounts

Have you tried it out?

With 24 x 400GB drives that would be almost 10TB. At that much disk space, we would have to be careful to make sure it could be written to quickly (over 100Mb/sec line running flat out, it is a bit under 1TB/day)

also, cooling tends to be a big design issue.

We tend to have 1 CPU per 4 drives, but it is interesting to look at other configurations.

Anyone have experience with these boxes?

-brewster

Reply [edit]

Poster: foundation Date: Nov 10, 2004 11:47pm
Forum: petabox Subject: Re: Possible solution: Disk Array Rackmounts

I've been looking at these boxes as well. Certainly they provide high density, but my main problem with them is the cost. It's cheaper to have regular cases with few drives, and more motherboards/cpu/memory etc (ignoring space as an issue).

Two supermicro SATA boards (8 ports each) would go well with one of the 3U 16 bay cases (part of the reason these cases are expensive are the hotswap trays)

Reply [edit]

Poster: HDRJapan Date: Oct 3, 2007 9:32pm
Forum: petabox Subject: Re: Possible solution: Disk Array Rackmounts

The cost was also one of our drawbacks and we went with 4 boxes, each doing a different action. Two would write and 2 would read.

This way we did not have to use one box which was overloaded in the first place.

Jasper - http://www.hdrjapan.com