My comment is simple. Don't.
The extra availability you might get from teaming is countered by teaming
the heartbeat NIC. Quick, lightweight, and fast is what that path needs.
The latency limits are pretty stringent and will end up forcing a node
offline if you miss. Then again, clustering on a blade chassis is not
optimal to begin with.
--
Geoff N. Hiten
Senior SQL Infrastructure Consultant
Microsoft SQL Server MVP
[quoted text, click to view] "salsafreak" <salsafreak@discussions.microsoft.com> wrote in message
news:CC5D2BCB-011A-4FF7-AB13-0633A3BC396A@microsoft.com...
> Thanks for your reply. Actually, given our hardware constraints (a blade
> with two physical NICs), we're wondering what are the concerns of teaming
> up
> both physical NICs to create two VLANs - one for the heartbeat and one for
> the public network. Given this setup and our goal of achieving
> redundancy,
> what comments would you have on this?
>
> "Geoff N. Hiten" wrote:
>
>> NIC teaming is definitely NOT recommended for the heartbeat network. I
>> have
>> found it useful on the public network(s). Disabling NIC teaming is a
>> first
>> step in troubleshooting a cluster because it simplifies the setup, not
>> because it is inherently unstable.
>>
>> --
>> Geoff N. Hiten
>> Senior SQL Infrastructure Consultant
>> Microsoft SQL Server MVP
>>
>>
>>
>>
>> "salsafreak" <salsafreak@discussions.microsoft.com> wrote in message
>> news:35BAB044-E540-4E70-AEB2-4424C3EA7C4B@microsoft.com...
>> > Our failover SQL cluster is a two-node single instance environment -
>> > SQL
>> > 2000
>> > Enterprise SP4 on Windows Server 2003 Enterprise. We're currently
>> > using
>> > NIC
>> > teaming to give us some hardware redundancy since the blade has two
>> > Broadcom
>> > NICs on it: one for the hearbeat private network and the other that
>> > connects
>> > to the public network. Our understanding is that Microsoft discourages
>> > this
>> > set up and will ask that you disable teaming when troubleshooting
>> > cluster
>> > issues.
>> >
>> > My questions:
>> > 1. What issues are associated with NIC teaming and, if not this, what
>> > other
>> > alternatives are there to minimize a single-point-of-failure (if the
>> > one
>> > NIC
>> > that connects to the public network goes down, your cluster is down)?
>> > 2. Why does Microsoft not want you to use NIC teaming?
>> >
>> > Any ideas will be appreciated.
>> > Thanks,
>> > Raul
>>
>>