Questions[iscsi](server)

This is a Canonical Question about iSCSI we can use as a reference.

iSCSI is a protocol that puts SCSI commands as payload into TCP network packets. As such, it is subject to a different set of problems than, say, Fibre Channel. For example, if a link gets congested and the switch's buffers are full, Ethernet will, by default, drop frames instead of telling the host to slow down. This leads to retransmissions which leads to high latency for a very small portion of storage traffic.

There are solutions for this problem, depending on the client operating system, including modifying network settings. For the following list of OSs, what would an optimal iSCSI client configuration look like? Would it involve changing settings on the switches? What about the storage?

• VMWare 4 and 5
• Windows Hyper-V 2008 & 2008r2
• Windows 2003 and 2008 on bare metal
• Linux on bare metal
• AIX VIO
• Any other OS you happen to think would be relevant

Is there a command to test whether jumbo frames are actually working? i.e. some sort of "ping" that reports whether or not the packet was broken up along the way.

I've an ESXi host with an Ubuntu VM which mounts a Dell MD3000i via iSCSI. I suspect jumbo frames are not enabled on the switch, and can't easily get admin access to it. I have the option to connect the disk array directly to the ESXi host, but would like some way of confirming that jumbo frames are a problem first.

I connect an iSCSI target, create a Physical Volume and Volume Group on it.

Then I create an LV, and

mkfs.ext3 /dev/vg00/vm

and all that works great.

Then I disconnect the target

iscsiadm -m node -T iqn.2004-04.com.qnap:ts-509:iscsi.linux01.ba4731 -p 192.168.0.4 -u

login to another Linux server, and connect the target there

iscsiadm -m node -T iqn.2004-04.com.qnap:ts-509:iscsi.linux01.ba4731 -p 192.168.0.4 -l

and I get:

linux01:~ # lvdisplay 
  --- Logical volume ---
  LV Name                /dev/vg00/vm
  VG Name                vg00
  LV UUID                NBNRGV-FkSR-ZNZ9-9AVk-chLQ-j5nc-RazeBw
  LV Write Access        read/write
  LV Status              NOT available
      LV Size                17.00 GB
  Current LE             4352
  Segments               1
  Allocation             inherit
  Read ahead sectors     0

I can see that /dev/vg00/vm doesn't exist, as I would have expected.

What am I doing wrong?

Are there any production-quality iSCSI SANs suitable for use with Windows Server 2008/SQL Server for failover clustering?

So far, I've only seen Dell's MD3000i, and HP's MSA 2000 (2012i), which both are around $6K with a minimal disk configuration. Buffalo (yea, I know), has a $1000 device with iSCSI support, but they say it will not work for 2008 failover clustering.

I'm interested in seeing something suitable for failover in a production environment, but with very low IO requirements. (Clustering, say, a 30GB DB.)

As for using software: On Windows, StarWind seems to have a great solution. But it's actually more money than buying a hardware SAN. (As I understand, only the enterprise edition supports having replicas, and that's $3000 a license.)

I was thinking I could use Linux, something like DRBD + an iSCSI target would be fine. However, I haven't seen any free or low-cost iSCSI software that supports SCSI-3 persistent reservations, which Windows 2008 needs for failover clustering.

I know $6K isn't much at all, just curious to see if there are practical cheaper solutions out there. And finally, yes, the software is expensive, but many small business get MS BizSpark, so the Windows 2008 Enterprise / SQL 2008 licenses are completely free.

Its standard practice to separate log and data files to separate disks away from the OS (tempdb, backups and swap file also) Does this logic still make sense when your drives are all SAN based and your LUNS are not carved of specific disk or raid sets -they are just part of the x number of drives on the SAN and the LUN is just space allocation