DR Plan suggestion

[vmk2014]

HI All,

    We have a small site, where 84 VM's running on HP Enclosure Blade server BL 460c Gen 8.

Configurations are :-

4 ESXi host 4.1.0, Build 800380.

Cluster :- Single

Storage - HP SAN.

There was a power outage in DC and UPS failure. Due to abrupt power failure the storage and ESXi host were down. After the power restoration, all the 4 ESXi host were up but the devices were not booted in sequential order, which caused the Chassis mid plane failure (FC switch) and all the VM's were unable to see the storage. Few of VM's were critical which caused plant service disruption and we tried to restore the VM's to local data store from the Avamar backup, but unfortunately avamar backup (VM) was also running on same cluster.

Now we have to plan DR so that in future we can prevent such kind of failure. Looking for DR plan advise.  Thanks in advance.

thanks

vmk2014

[impranayk · Accepted solution]

Hi,

Few things to be noted here before planning to DR.

Are you looking for Site Resilience in WAN network?

Are you looking for DR in same datacenter?

Are you looking for DR in LAN/MAN Network?

Once decided above, you have to define RPO/RTO for recovery during disaster. How much data should be maximum loss, and how much time it should take maximum to bring site up.

Once decided RPO/RTO, go for which kind of data recovery you are looking for? Asynchronous Replication or backup through Tape?

Once all answers in hand, here are few options you can plan DR.

Plan#1

SRM (Site Recovery Manager) - Since you have small infrastructure, it will be bit costly. However as best practices, you can go with this if cost is not constraint.

              Short Plan:-

                                   - Install VC in two different sites.

                                   - Install SRM in both sites.

                                   - Ask storage team to provide replicated LUNs to be used for this VMs.

                                   - Configure SRM as Protected and Recovery Site. Refer to SRM guide for more depth details.

Plan#2

Use 2 sites and restore backup from tape in case of any disaster.

Plan#3

Use replicated LUN for VMs.

Use 2 sites in different location.

Make visible these LUNs to other site and register VM to that site, in case of disaster.

Regards,

Pranay(VCAP5-DCA, VCAP5-DCD)

[vmk2014]

Hi Impranayk,

        We have a global support for 2000 + server, so what will be cheapest solution for DR ? They dont want to spend money on replicated LUNS/Storage.

         They need cost effective solutions.

Thanks

vmk2014

[Paltelkalpesh]

SRM is the best option to minimize downtime, but in that scenario you  have to setup same VMware infrastructure in both site which may little bit costly.

As you have budget problem, first you have to decide which applications are more critical for your business and what RPO and RTO you are looking for this.

if you are using MS SQL or MY SQL then you can enabled real time DB replication or log shipping for push data from one site to another site in shot time and for the rest of the data you can scheduled through FTP or any backup tool for transfer data from one site to another site.

Keep redundancy for all critical devices and all should be configured in active-active or active-passive mode.