Hi, I have 2 HP Storage server with 14 SAS HD. - 2 SAS HD 128GB each, RAID 0, for OS (it comes with Storage Server 2008), in the front - 12 SAS HD 450GB each, not use yet, in the back HD Layout 1__4 0000 0000 0000 9__12 1) How can i configure both server to work as one, with maximum capacity ? 2) What's the best OS/Software to use ? 3) What's the vest RAID setup to use ? 4) What's the best transport to use ? (SMBv3,iSCSI,NFS, etc...) Thank you.
1) How can i configure both server to work as one, with maximum capacity ? By putting them in a server cluster 2) What's the best OS/Software to use ? Providing you do want to do clustering, Windows Server OS. Im sure there is a way to do it with linux however I have not had any experience with that. 3) What's the vest RAID setup to use ? There is no one best raid setup. They all have their pros, cons and specific use case scenarios. it all depends on how you are going to use the drives. For example; if you arnt going to be writing data to them much and you need fast read access, raid 5 or 6. The servers you have should have an onboard raid controller with battery backup, so use that. 4) What's the best transport to use ? (SMBv3,iSCSI,NFS, etc...) Again, this depends on how and what you want to use the servers for, weather they are for storing data, providing backups, website hosting, etc.
Start by deciding whether the you are sortdatasorting important or not. If not, you can configure your servers for maximum speed and capacity. When one of your 24 disks dies or momentarily goes offline, you'll lose everything. If this does not sound acceptable, you'll need some level of redundancy. If you go this route, there are two fairly straightforward possibilities. For each, I suggest making a RAID 1 volume from the two 128 gig drives and installing the os there. Use the onboard hardware raid for this. One system will be the interface to the rest of your network, the second exists only as a storage pool. Having a dedicated network link between the two (minimum of 1 gig-e, jumbo frame link, multiple teamed links or 10 gig works better) will make your system more reliable and greatly improve performance. With this many disks, you need to be concerned about both overall reliability and bitrot. A next-gen filesystem will be your friend. For the most flexibility, install Linux and build and create one or more btrfs RAIDZ arrays. You could do one that spanned all disks on both servers or go with a 12 disk array on each machine and then combine the two into a single volume. If you want to go the Windows route, you really want to start with Server 2012 R2. Create a storage space including all 24 drives then make a parity volume formatted using REFS. That way when one of the two boxes crashes our goes offline you won't lose data. As sebus and cocachris mention you can also go the clustering route. This is a pita in Windows and only marginally easier in Linux - assuming your Linux skills are reasonably good. This approach has advantages for reliability but is non-trivial. Setting up a virtual SAN is certainly viable, but makes clustering seem easy. Which network transport to choose depends on your choice of os, whether you are accessing the share locally of over a high latency remote connection. Samba/SMB is a good overall choice. NFS is slightly slower, but often consumed less CPU. I would avoid iSCSI unless you have multiple, redundant NICs and are only sharing the volume with a single client.
I am sorry, but I must be missing something. 12 disk in each machine. What possible config would one use to make storage spaces using all 24 disks? Can Storage Spaces jump across different hardware?