UNSOLVED

mimmus1

updated

16 years ago

M

mimmus1

59 Posts

0

1240

November 3rd, 2010 02:00

NS480 iSCSI, VMware and high availability

Hi,

we recently purchased a NS480 and I'd like to connect it to a VMware infrastructure by iSCSI, ensuring the highest level of availability (and performance).

In the past, I was not totally satisfied from features of a Celerra: due to active-passive nature of data-mover architecture, I was not able to build a totally safe IP fabric

On the VMware side, I configured two NICs, one to a switch and one to the other. Same thing from the Celerra side: one datamover to a switch and one the other switch. No uplink between switches.

In the event of "active" NIC/link failure, NIC/link connected to passive datamover is completely useless.

Peraphs, I was missing something but now I'd like to builld an IP storage network totally safe, like a fibre-channel SAN. Could you point me to the right direction?

Thanks in advance

  • JGreen

    46 Posts

    632

    0

    Posted November 3rd, 2010 08:00

    mimmus,

    Please reference the attached document "Configuring and Manageing Celerra Network High Availability.pdf"

    I believe you will find what you need.

    1 Attachment

  • mimmus1

    59 Posts

    632

    0

    Posted November 3rd, 2010 09:00

    OK, I forgot to say that I configured FSN on the Celerra side

    But I'm still not completely satisfied: if NIC/link fails, I have not a path to the active datamover.iSCSI.png

  • JGreen

    46 Posts

    632

    0

    Posted November 3rd, 2010 10:00

    For link/port redundancy on the data mover side configure LACP "trunks" (Celerra term) as the underlying FSN devices utilizing internal LACP link redundancy on the active FSN member.

  • dynamox

    11 Legend

    20419 Posts

    87439 Points

    632

    0

    Posted November 3rd, 2010 10:00

    it does not matter what you configure on Celerra, unless Celerra detects link state change it does not do anything with failover.  I had a situation where Celerra was connected to uplink switch that lost its backend connectivity to core switch, the link never went down but Celerra was hosed. Unless you get one of the new switches that allows LACP trunking between two physical switches you are at the mercy of link state.

  • JGreen

    46 Posts

    632

    0

    Posted November 3rd, 2010 10:00

    Understood. The proposed LACP configuration on the data mover side was for data mover link redundancy on the active FSN device, nothing more.