There are several possible issues going on here.

A few pointers:

- The PowerConnect 2700 series isn't really a decent switch for iSCSI traffic, so they may be part of your issues

- Don't upgrade to vSphere 5 as this hasn't been validated with the MD3000-series (only validated with the newer MD3200 and 3600 series)

- which vmkernel are you using for vMotion? Is it one of the iSCSI vmkernels or another vmkernel that is leveraging the iSCSI vSwitch(es) or the iSCSI physical switches?

If possible, can you SSH into the server and post the output of:

- esxcfg-vswitch -l

- esxcfg-vmknic -l

and then provide the 4 IP addresses that are being used for the iSCSI ports on your MD3000i?

E.g.:

Controller 0 iSCSI port 0: 192.168.130.101

Controller 0 iSCSI port 1: 192.168.131.101

Controller 1 iSCSI port 0: 192.168.130.102

Controller 1 iSCSI port 1: 192.168.131.102

Dev Mgr, thanks for the advice on vSphere 5.

I not sure if there would be funds to replace the power connect switches, but if we were to look at doing this what would you recommend?

Vmotion is handled by the management network which is on a different subnet and which is connected to seperate network switch.

Here are the results of the two commands you asked me to run.

~ # esxcfg-vswitch -l

Switch Name      Num Ports   Used Ports  Configured Ports  MTU     Uplinks

vSwitch0         64          8           64                1500    vmnic3,vmnic2

 PortGroup Name        VLAN ID  Used Ports  Uplinks

 VM Network            0        4           vmnic3,vmnic2

 Management Network    0        1           vmnic3,vmnic2

Switch Name      Num Ports   Used Ports  Configured Ports  MTU     Uplinks

vSwitch1         64          3           64                1500    vmnic0

 PortGroup Name        VLAN ID  Used Ports  Uplinks

 iSCSI_0               0        1           vmnic0

Switch Name      Num Ports   Used Ports  Configured Ports  MTU     Uplinks

vSwitch2         64          3           64                1500    vmnic1

 PortGroup Name        VLAN ID  Used Ports  Uplinks

 iSCSI_1               0        1           vmnic1

~ #

~ # esxcfg-vmknic -l

Interface  Port Group/DVPort   IP Family IP Address                              Netmask         Broadcast       MAC Address       MTU     TSO MSS   Enabled Type

vmk0       Management Network  IPv4      192.168.196.xx                          255.255.252.0   192.168.199.255 00:1d:09:73:b2:be 1500    65535     true    STATIC

vmk1       iSCSI_0             IPv4      10.10.0.x                             255.255.255.0   10.10.0.255     xx:xx:xx:xx:xx:xx 1500    65535     true    STATIC

vmk2       iSCSI_1             IPv4      10.10.1.x                              255.255.255.0   10.10.1.255     xx:xx:xx:xx:xx:xx 1500    65535     true    STATIC

Here are the four IP addresses which are being used on the MD3000i

Controller 0 iSCSI port 0: 10.10.0.1

Controller 0 iSCSI port 1: 10.10.1.1

Controller 1 iSCSI port 0: 10.10.0.2

Controller 1 iSCSI port 1: 10.10.1.2

Replacing SATA with SAS drives does mean getting 10k or 15k SAS drives. Dell also offers 'near-line SAS', which is still 7200rpm and wouldn't make a difference in performance from SATA drives.

I also logged a call with DELL and they took a look at the MD3000i logs. They said that their was some I/O queuing on the LUNS which were running SATA drives. They advised that the SATA drives cannot handle the I/O and that we should look to replace them with SAS or try to move our more data intensive VM's onto the SAS LUNS.

Apart from replacing the drives with faster SAS drives, is there anything else that i should check that could be impacting the performance?