VirtualCenter Across Datacenters

[virtech]

Wondering how many people out there are using VirtuaCenter to manage remote data centers. I'm looking to manage two data centres over a 100MB link. Say 10 hosts per site, multiple HA clusters.

Interested to hear of any issues faced and performance issues etc...

[Jae_Ellers]

We manage 4 regional sites in Europe and Asia over 10Mb from the US, but we only have one host per site.

I don't think you'd have any problems with 100Mb link, but it would be great to hear from someone doing more hosts.

[jslarouche]

Three weeks ago I implemented an ESX server in our Calgary Datacenter which we have a 100mbit link to our Ontario Office.. We also have one ESX server at our hot site which is setup with a 1gig pipe for DR replication.

I had only twice that the host went to a disconnected state all i had to do was right click the node to re connect it.. No VM's or clients we're impacted. I think they have corrected the latency issues with Virtual Center 2.0.2. We are still at 2.0.1.

I hope this helps..

[MR-T]

This is what I'd always opt for.

A central point of management.

You've got a good link between site, a single VC will be perfecto.

Next thing you need to think about is your DR strategy for the VC & Database.

[virtech]

DR strategy for VC will be to simply to start the VC VM which is powered off on the other data centre (both data centres will be sync'd using SAN based replication). The LUNS should be indentical, thats what Im hoping for anyway.....

[MR-T]

This should be good enough.

I love having a single VC for DR purposes.

Having just completed 2 installations like this. The first had 2 physical VC's in seperate datacenters with a SQL log shipping setup to keep the database updated. This wasn't straight forward to setup and required a number of scripts to get workiing. When you've got 2 seperate VC's, you'll need to disconnect and reconnect all the hosts when you fail over as they are being managed by the dead VC.

My preffered option was my 2nd installation. I've built a single VC as a VM and also created a SQL VM which the VC uses. These are replicated to the secondary site using the SAN software and failover is straight forward.

[virtech]

Thanks MR-T, my configuration should be fine

[shaneyoder]

Do you do any queicing of the VMs before the SAN replication? What SAN are you using?

[Paul_B1]

We manage remote locations from our corporate datacenter. Currently we have hosts in NA and in EU (our connection to EU is a 2mb connection). The one thing I found is I had to change the timeouts a bit but it works fine. Just takes a bit longer for tasks to initiate.

[PetervandenBosc]

Just finished the architecture of two datacenters. Containing 6 HP DL585 32G Ram and 2 Quad core processors, a NetApp Metrocluster and 1 Gig fiber several fiber pairs available.

Two site's on each site 3 servers. One virtual Virtual Center Server, Two Clusters one contains 1 ESX hosts on each side wo wil host all virtual Oracle server, this choice i made to keep processor licenses for oracle acceptable and create high available VM's. Other Cluster hosts all other VM's. Rules in DRS wil keep VM's on one site. So no dead iron everything is used and fantastic failover capabilities!

[zbuckholz]

Petervanden thanks for sharing your details. Currently we have three data centers within 5 miles of each other, our layout is production, qa, dev. I tried several times to get details on the Metrocluster for HA and DRS for a new data center outside 50 miles. Please share some experiences you have had with the metrocluster.

Have you done DRS between data centers? If so how do you handle routing ips?

What about snapshots and backups? Do you use software or the NetApp features?

Thanks

Zach