Re: Is WFA supports active/standby design ??

madhu_gogalla · ‎2015-09-22

My client has 2 different geographical data centers and asking us to implement WFA in active/standby mode , so that they can execute them from both if needed.

Does WFA supports active/standby config ?? did any one have implemented it ?

geringer · ‎2015-09-22

Madhu, I think you have a number of choices on an architecture. The right solution for your customer will depend on the next level of details on the requirements. I see a few options, and I suspect there may be others as well. 1) Run WFA on a VM, and setup a D/R instance in the remote site. Utilize the remote site D/R capabilities of your hyper-visor, or SMVI and SnapMirror. There are many ways to do this. This will give you one production instance of WFA to manage and logs/history (within RPO) will be available at the D/R site. You will only have the D/R instance available in the case of a true D/R scenario. 2) Setup two WFA instances. Run regular backups of the production WFA database and send to the D/R site. Again there are many ways to do this. In the case of a D/R, restore the latest backup on the D/R site and force data source acquisition. This will also give you one production instance the WFA database. You will need a change control process to keep both instances of the software at the same code release, and you will still have the same logs/history (within RPO) at the D/R site. 3) Setup two production WFA instances. You will need a change control process to ensure you release workflows and WFA configuration into both instances. You will not have history from a site that fails. You will not need to perform any D/R recovery for the second site. This is more active/active. Both sites will load the WFA cache from the same data sources. Mike

arjunan · ‎2015-09-22

Option 1 : Backup and restore : Have two instances of the WFA server on the two geographical locations. One WFA instance as active and whenever you need to switch to other instance take a backup from Active instance and restore it on the standby instance. steps: 1. Take a Backup of the Active instance 2. Bring down the Active instance 3. Bring up the Standby Instance 4. Restore with the Backup taken on the Active instance. drawback: ------------- every-time you need to switch you need to take a backup and restore it on the standby instance and make it active. Option 2 : High-Availability : WFA supports High-Availability option in the latest release. but we support this configuration only on the same geographical location because the shared LUN between High-Availability WFA servers should be on the same location as WFA instances. Option 3: Disaster recovery solution : This is also more like a backup and restore solution, but both the instances will have the same UUID as the DR backup takes the entire system backup and restore.

cbbarkin · ‎2016-04-26

@arjunan wrote:
Option 1 : Backup and restore : Have two instances of the WFA server on the two geographical locations. One WFA instance as active and whenever you need to switch to other instance take a backup from Active instance and restore it on the standby instance. steps: 1. Take a Backup of the Active instance 2. Bring down the Active instance 3. Bring up the Standby Instance 4. Restore with the Backup taken on the Active instance. drawback: ------------- every-time you need to switch you need to take a backup and restore it on the standby instance and make it active. Option 2 : High-Availability : WFA supports High-Availability option in the latest release. but we support this configuration only on the same geographical location because the shared LUN between High-Availability WFA servers should be on the same location as WFA instances. Option 3: Disaster recovery solution : This is also more like a backup and restore solution, but both the instances will have the same UUID as the DR backup takes the entire system backup and restore.

A couple of questions:

1. For the active/active model (WFA in each site polling the same data sources) - would you have OCUM instances in each site as well, with each OCUM polling each cDOT cluster (assuming a manageable number of cDOT clusters)

or

2. Passive model: perform a similar backup/restore task with the OCUM server

Thanks geringer and arjunan.

arjunan · ‎2016-04-26

hi cbbarkin,

1. For the active/active model (WFA in each site polling the same data sources) - would you have OCUM instances in each site as well, with each OCUM polling each cDOT cluster (assuming a manageable number of cDOT clusters)

Active\Active means if there is a failure of application\server in one location will be taken over by another site without downtime, what we support in WFA high-availability is Active\Passive option. The one node will be active and whenever the Active node goes down the passive will become Active.

whatever you have mentioned are two differnt instances,

- when one WFA at one site goes down while executing workflows that will not be taken care by another site as there is no link between them.

- This may help in monitoring but not for WFA application.

2. Passive model: perform a similar backup/restore task with the OCUM server

- I think OCUM supports backup\restore, i am not sure if they support comprehensive backup/restore

thanks,

Prabu

2. Passive model: perform a similar backup/restore task with the OCUM server

- are you talking about OCUM instance ? The backup restore is supported in OCUM but i am not sure if they support Comprehensive restore\backup.

cbbarkin · ‎2016-04-27

HI arjunan,

Thanks for the response - I should have been clearer - I realize that neither OCUM nor WFA support what anyone would consider active/active. I should have described it as independent instances of WFA/OCUM/OPM at each site.

In any case for my client we've decided on a single VM with WFA/OCUM/OPM, that will be replicated to their three sites either with SMVI/SnapMirror or vSpere replication.