Starting from scratch

[happyhacker]

I am buying a server and fitting it with 2 1Tb drives to be run as RAID 1. The server comes with a single 160Gb drive. I need to fit the two main drives.

HP ProLiant ML115 G5 Server
2 x Samsung Spinpoint F3 1TB 7200rpm SATA 3Gb/s 32MB
Plus fitting 8Gbyte RAM

Do I just fit the two drives for the data and use the supplied drive for the OS or what arrangement is best?

Thanks for your time.

 

[ShackDaddy]

Yes, just use the supplied drive for the OS. Once you've set things up, you can easily use the wizard to move the User folders, Exchange and WSUS data to live on your large mirror.

Dave Shackelford MVP
ThirdTier.net
TrainSignal.com

 

[happyhacker]

Thanks ShackDaddy. I assume then when I configure the two new drive for software RAID (included in this product I think) the OS drive (160Gb one) is separate from that configuration?

Thanks for your time.

 

[ShackDaddy]

Yes, that's right.

If you want the best protection, you might want to get another matching 160gb drive to mirror with the existing one, so that you can have redundancy for your OS. If you aren't going to do that, at least do a great job on backups so that you can successfully swap in a new OS drive if that one fails.

Dave Shackelford MVP
ThirdTier.net
TrainSignal.com

 

[happyhacker]

Hm, that's an interesting thought. If I mirror the OS I assume that the RAID (point about that below) 0 + 1 (parallel standby?) doing that will cut the good one in seamlessly or will I get an error message and have to decifer what's happened?

RAID: assuming one can configure 2 lots of software RAID for the pairs of drives, won't that be quite a load on the OS or BIOS (whatever is running the RAID software)?

Thanks for your time.

 

[ShackDaddy]

It will take over seamlessly, but depending on which drive goes down, you may have an error\issue when you next restart your server. The only error you'd probably see when one drive fails would be in the System logs.

I don't think software RAID (particularly a mirror) will be much of a load on the OS. People have been using software RAID for 15 years and doing it with way less CPU and memory available. On a modern server, software RAID isn't going to be a resource hog. There are other arguments against using software RAID, but resource usage isn't a significant one in most scenarios.

Dave Shackelford MVP
ThirdTier.net
TrainSignal.com

 

[happyhacker]

Well it's a good intro. to RAID so I'll go with it to avoid the cost of the h/w! From what you say the fault could be a dormant one until someone looks at system reports? I would expect the ability to email an error to say a drive has gone down especially if the next backup is going to write an errored drive over a good backup!

Thanks for your time.

 

[ShackDaddy]

You'll want to research what the likely error would look like. It will be a Disk event in the System log. I typically use some sort of 3rd-party monitoring service like GFI Max that allows me to choose alerts on an a la carte basis, and I usually configure one to check for that specific error and trigger an email to me if it comes up.

Dave Shackelford MVP
ThirdTier.net
TrainSignal.com