UNSOLVED

ciscen

updated

13 years ago

C

ciscen

27 Posts

0

253704

March 3rd, 2014 12:00

windows 2012 R2 NIC Teaming and MPIO

Hi there,

I have a test server with 4x1Gb onboard and 1x Dual Port 10Gb card and installed win 2012 R2 on it. I've created 2 xNic Teaming with pairing 10Gb (active) and 1 Gb (standby) together. On top of this config, I've installed the Equallogic MPIO and seems all working.

As I'm curious, I'd like to know if Equallogic MPIO is supported on teamed adapters (lets assume 2x10Gb teamed adapter) ? And how about my test config (10Gb&1Gb nic team with 1Gb standby), is it also supported config from the Dell Equallogic (The theory behind this config is in case if the 10Gb card becomes malfunctioning)?

Thanks,

Cem

  • 2459

    0

    Posted March 4th, 2014 00:00

    Microsoft does support teaming and MPIO if you use their native teaming solution (so no Intel or Broadcom software, just what comes with Windows), as long as you create a virtual switch connected to the team and virtual interfaces attached to the virtual switch.

    The first link Donald posted (which also has good comments, BTW) explains that:

    "

    · The Technet statement that basically quotes “iSCSI + NIC Teaming not supported” is still true for all teaming solutions, with the EXCEPTION of the Windows Server 2012 inbox NIC Teaming solution we provide.

    · If iSCSI Initiator is used with dedicated NICs such as in a stand-alone and/or Failover Clustering environment, then NIC Teaming should not be used (because it adds no benefit over MPIO for dedicated NICs).

    · If iSCSI Initiator is used in a shared NIC scenario (see figure below) such as in a Hyper-V 2012 environment, then iSCSI Initiator used over the Hyper-V switch (and over NIC Teaming) is supported.

    "

    That doesn't mean you have to run VMs on your host, just have the Hyper-V role enabled to be able to create virtual switches and vNics for the parent partition.

  • ciscen

    27 Posts

    2459

    0

    Posted March 5th, 2014 22:00

    Thanks  Mirko,

    With our test server (1xdual 10Gb + 4x1Gb onboard), I've created very similar config as you mentioned. Probably I should have provided a bit more details.

    I've enabled NPAR and created 4 partition per 10Gb port so I've ended up 8 partition (nics). Enabled iscsi offload for the first partition on both 10Gb port and used the default settings for the global bw etc.

    As I have only 1x 10GbE card, I've created Nic team by using 1st partition(iscsi offload) from the 10Gb port1 + Nic1 (1Gb). Also switch independent mode is selected as lbf and 1GbE is specified as standby mode in order to utilize just the 10GbE. I've created the other Nic team by using 1st partition from 10Gb port2 + Nic2 with the same settings. HIT kit is install and running MPIO on top of this config.

    I've created hyperv switch and vnics for other purposes but iscsi is basically running on it is own nic/partition(or team).

    In case, 10GbE card fails then this config should maintain the connection via 1GbE ports so it should be redundant. On the other hand, if the 10Gb cable or switch fails then 1GbE will be active at least for one of the path which is part that concerns me.

    This may not be recommended configuration but I'm curios if it is supported from Dell EQL team.

    Thanks,

    Cem