Jump to content

iscsi, really?


Kung Fu Jesus

Recommended Posts

While I understand the performance benefits from removing the overhead of something like NFS, there are huge advantages to running ESXi over NFS.

http://blogs.netapp.com/dave/2007/09/why-run-vmware-.html

File management is easier, you can snapshot VM's, etc etc. Plus with iSCSI it's much harder to manage virtual space. For example, the way I'd do it (and I have a ZFS zpool with FreeBSD), I'd have to basically carve a chunk out of my zpool and allocate it as an iSCSI target (that is when iSCSI target is finished in FreeBSD). This actually subtracts from the overall storage in the zpool. Even though I'm carving out 20 or 30G, I may only us 15GB in the end, effectively wasting about 15G of space. A vmdk is dynamic in size and knows to not use the empty blocks. No space is wasted this way. Don't get me wrong, when I finally run out of ports to expand on my 4.7 TB NAS and utilize most of the bandwidth for the infiniband out that will come with the cheap SaS controller I'll purchase, I'm probably going to purchase a very large chassis and build a computer with as much storage as possible with openfiler, activate the iSCSI target functionality, and then use FreeBSD's iSCSI initiator over the spare gigabit port. The reason I'd do such a bizarre thing is simply so I can have at least virtual block devices to build more vdevs to add to my zpool. But for my virtual server environment, I'd probably just use NFS and spare myself the loads of extra wiring.

My current NAS is packed to the brim, it's running FreeBSD with 4x750GB drives in one RAIDZ1, and 4x1TB drives in another, allowing for a fairly large and redundant pool. It's all packed into a tiny micro-atx case, with a cheap Sempron powering all of it. Since this server largely is used for NFS/CIFS shares, many of the CPU cycles are unused anyway. It has 4GB of memory, which provides a sufficient adaptive rebuild cache for now.

Link to comment
Share on other sites

the way I'd do it (and I have a ZFS zpool with FreeBSD), I'd have to basically carve a chunk out of my zpool and allocate it as an iSCSI target (that is when iSCSI target is finished in FreeBSD). This actually subtracts from the overall storage in the zpool. Even though I'm carving out 20 or 30G, I may only us 15GB in the end, effectively wasting about 15G of space.

think it depends on the situation, in your case the iscsi would be sort of redundant since the drives are already part of a zpool

Link to comment
Share on other sites

Yes, but the wonderful features of the zpool include things like redundancy and checksumming. There are reasons to use the zpool as a level of storage above the iscsi volume, just as there are reasons to do the same with LVMs (as I've seen done in the HP-UX, Solaris, RHEL and AIX worlds). The main reason being that you have the ability to carve out different volumes from a much larger one.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...