Guys,
I'd require some input for a project I'm involved in. I have two sets of 3050 clusters in two sites. The primary site is serving CIFS data to home and project directories. The secondary site is just the snapvault destination, and tape backup source.
Now, we're planning to do a DR implementation for this setup. From what I can recall, there are a few things that you need to do to make it work: 1. have snapmirror and snapvault (SM-SV combo) license at the secondary site. I can turn a snapvaulted volume snapshot into a RW volume. Is that correct? 2. I copy the cifs shares config file from the primary filer, edit it a little bit, and paste it into the secondary filer 3. I need refresh my DNS server to point to the secondary filer IP address using the primary filer hostname.
Anything else that I need to do? I heard something about changing netbios names as well, but I don't really understand windows enough to know what that will do to help with the DR.
Oh, and also, the MS active directory domain are running on win2k3 servers. I'm assuming that different AD versions will significantly in make DR implementations different?
Thanks for any feedback
Have you looked at using DFS to manage your CIFS shares? DFS makes it easier to do DR for CIFS, esp. if you use Domain based DFS and have many CIFS shares on multiple filers, and if you also use VFM (Virtual File Manager) from Netapp you can automate almost all the tasks you mentioned, except may be converting a snapvaulted volume to RW. I'd suggest using snapmirror instead. Snapmirror is for DR and snapvault is for Backups. Both have different purposes.
Here's how we have it setup - All of our CIFS shares are shared out using domain based DFS and we use VFM to manage our DFS. We have 3000+ links in 6 locations on 10 different filers around the globe, so its not easy to manage DFS using Microsoft's native tools. - We use Snapmirror to replicate our most critical data to two secondary sites on R200s. (There is a way to convert snapvaulted volumes to RW, but it looks complex and you may not be able to automate it) - We have some DR policies setup within VFM, which takes care of all the tasks like creating the cifs shares on the secondary and updating DFS with inactive links to the secondary BEFOREHAND. - During DR, one has to just activate the relevant DR policy, which will then break the SM relationship and make the secondary RW, then update DFS to promote the inactive link as the primary. Since DFS is domain based, it updates AD and all clients are redirected to the fail-over site after a brief timeout. No need for any DNS updates or netbios name change. - We reverse the steps for fail-back with an added task of resyncing the snapmirror in the other direction. Works pretty well in all the tests we did so far. Note: VFM and Snapmirror licenses are not cheap :( -G
On 1/17/07, Ammar Zolkipli ammar.zolkipli.lists@gmail.com wrote:
Guys,
I'd require some input for a project I'm involved in. I have two sets of 3050 clusters in two sites. The primary site is serving CIFS data to home and project directories. The secondary site is just the snapvault destination, and tape backup source.
Now, we're planning to do a DR implementation for this setup. From what I can recall, there are a few things that you need to do to make it work:
- have snapmirror and snapvault (SM-SV combo) license at the secondary
site. I can turn a snapvaulted volume snapshot into a RW volume. Is that correct? 2. I copy the cifs shares config file from the primary filer, edit it a little bit, and paste it into the secondary filer 3. I need refresh my DNS server to point to the secondary filer IP address using the primary filer hostname.
Anything else that I need to do? I heard something about changing netbios names as well, but I don't really understand windows enough to know what that will do to help with the DR.
Oh, and also, the MS active directory domain are running on win2k3 servers. I'm assuming that different AD versions will significantly in make DR implementations different?
Thanks for any feedback