P4500 G1's in production

[Northwoods]

Who is running P4500 G1's in production and under support?

If you are running them, are you using them in a mixed generation cluster?

[andre_muc]

Hi,

we have 1 multi-site san bundle with G1 und G2 in production in one site and a G2 on a remote site, without any problems.

... ok, only the loss of space.

we use remote snapshots also without any problems.

Andre

[mars_seven]

2 P4500 G1 (SAN/iq 9)

2 P4500 G2 (SAN/iq 9)

trying to downgrade 2 P4500 G2 SAN/iq 10 to SAN/iq 9 to work w/ current cluster.

[RemyZ]

We run 30 G1 nodes in production (5 management groups). We do not run mixed clusters with G1/G2 nodes, but we do use G2 450GB and G2 600GB nodes mixed in one cluster.

[oikjn]

sorry to hyjack, but remy, why five management groups?  I could see multiple clusters or even maybe two management groups, but unless you have five sites or five isolated business groups, I don't know why you would want five managment groups

[RemyZ]

The main reason we do not create big management groups is that updates of the software are always done management group wise. Did you ever update a management group with 14+ nodes? Our official maintenance window is not big enough (4 hours) to complete the updates. And you cannot start the update and head home, since the update will stop if a disk gets faulty in the mean time (yes, we experienced that multiple times).

Another reason is that in our philosophy one should have a management group per application. So management groups for Exchange, file services, ESX, backup, etc. In our case we have a per-site management group for Exchange.

Besides all that, the recommendation of HP is not to put more than 20 systems in one management group (source: http://h20628.www2.hp.com/km-ext/kmcsdirect/emr_na-c03686039-2.pdf, page 113). In general, our management groups consist of 12 to 18 nodes. The G1 nodes are in management groups which are spread across multiple sites. 

Did I mention we have 106 production nodes (and counting) in 12 management groups (and 18 clusters)?

[oikjn]

that makes a lot of sense.  I was thinking you had 30 nodes with 5 groups which would mean ~6 nodes per group which seems a bit low and would have expected it in 2-3 groups, but I totally get your logic for your groups when you have 100+ nodes.