what do you use for srm disaster recovery site for networking solution

[vmwareluverz]

i'm new to vmware srm and we have two sites and trying to understand how srm and networking works between sites so when dr kicked in the networking scheme would not be affect and its for new jersey->new york offices. what do you use and suggestion please.

[azn2kew]

What is your current ESX version and how large is your environment? What types of storage and replication do you have between sites? What applications and dependencies involve and what is your expected RTO? If you're new to SRM I would suggest to contact VMware VAC or VMware PSO to consult and deploy SRM more effectively and just because its your core disaster recovery strategies it would be consider and architect appropriately.

If you're using ESX 3.5 right now then you can use SRM 1.0.x version but if you're planning to upgrade or use vSphere 4.0 than you would have wait until SRM 1.5 beta is available for GA then you can implement them. From the design and architecture perspective, I would focus on these:

1. Should have enough ESX hosts and licenses on both sites.

2. Should have datastores on both sites and choose the types of replication "asynchronous or synchronous" depends on bandwidth and budget.

3. Make sure you have vCenter server and databases on both sites as far as SRM components too.

4. Consider what types of applications and dependencies you need to sort the process more effectively.

5. How do you handle networking scheme and most people uses RE-IP.exe tool or DNS batch scripting tools you can download. Most like if you have large networking team, they would implement stretched VLAN to your DR site and you don't have to worry on that part.

I would look at reading some DR guide on www.vmware.com for details.

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!

Regards,

Stefan Nguyen

VMware vExpert 2009

iGeek Systems Inc.

VMware, Citrix, Microsoft Consultant

[dburgess]

So as below SRM's first job is to recover the machine from the storage and register in the second VC instance. Because it is using a copy of the VM in both test and 'real' failover mode if you do nothing the network config will be the same initially. For safety reasons then, SRM will create isolated test bubblesand connect the VM's to these so unless you do something else you're safe, but poorly connected! You have options to change this and make SRM hook the VM's up to different switches etc when running tests/failovers.

Next question is then do you want to/have to change the IP address of servers? If the answer is yes then you can use the inbuilt customisation rules of SRM or a script that runs at power on etc.

If you don't need/want to change the IP address then you will need to make sure you have a VLAN that spans both sites. This is frequently done but equally is frowned apon in many shops as you are stretching a failure domain also.

You could also consider keeping the IP addresses and have the routers change over the 'postion' of the VLAN. This way you have kept the IP addresses, which is good not only because you avoid the complexity of changing them but also you have avoided the issues that the clients etc often have if you suddenly change the IP address of a service, DNS can be problematic here. Secondly you have also avoided spanning your VLAN's. Down side is you have to allow for the routers to converge etc so will require wor, you could consider Route Health Injection techniques which would allow you to have all the routes setup prior and you just essentially change the cost of the route, possibly programatically, in the event of a failure.

Remember you have two considerations though - what to do in a test and what to do in a real scenario.

Cheers,

dB

[AsherN]

Weighing heavily on how much to automate is both the technical expertise at both sites and how long can you be down.

If routers need to be reprogrammed, can you afford the 15-30 minutes to do so? is there somebody available at the remote site to do it?

It may be as simple as having an alternate config to upload to the router and reboot, if you don't mind cutting your connection for a few minutes.

[vmwareluverz]

I would like to keep the IP solution and so the best solution for SRM is to have redundant on the router to configure to switch over when needed? You you explain more details how this part works if we're not using REIP and Stretched VLAN, can you give me some logics so I can explain to networking team? Thanks.

[azn2kew · Accepted solution]

If you have time read these threads had good discussion about SRM and Networking failover options as mentioned. Hopefully that helps you to plan it more effectively.

http://communities.vmware.com/message/1113568#1113568

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!

Regards,

Stefan Nguyen

VMware vExpert 2009

iGeek Systems Inc.

VMware, Citrix, Microsoft Consultant

[vmwareluverz]

thanks for the links there are very good discussion regarding srm network designs and looks like i have to depend on my networking guru to accomplish the goal. anyone actually have done a complete srm implementation yet with which methods?