Unsupported content

 

This version of the documentation is no longer supported. However, the documentation is available for your convenience. You will not be able to leave comments.

Disaster recovery for multiple servers at multiple locations

As shown in the following figure, this distributed topology is composed of two CDPs, with the primary CDP in one data center and the HA-CDP in the other. The grid is intentionally distributed geographically across intermediate physical networks. Under normal operating conditions, CDP and HA-CDP are routed jobs, as both CDPs are in an active-active state.

When deploying adapters in this topology, you must deploy all adapters to both CDPs.  In the event that one of the CDPs becomes unavailable, the other HA-CDP will continue to process jobs and still have access to all its adapters locally.

To deploy this configuration, complete the following activities:

  1. Using the following completed installation worksheets, install the CDP components in network A:
    Repository installation worksheet
    CDP installation worksheet
    BMC Atrium Orchestrator Development Studio installation worksheet
  2. Using the following completed installation worksheets, install the HA-CDP components in network B.
    HA-CDP installation worksheet
    BMC Atrium Orchestrator Development Studio installation worksheet

Two Peers (CDP and AP)

 

Related topics

Installation worksheets

Working with HA-CDPs

Installing BMC Atrium Orchestrator Platform with BMC Atrium Single Sign-On high availability clusters

System requirements

BSM environment recommendations

Installing BMC Atrium Orchestrator content

Adding adapters to a grid

Configuring adapters

Enabling and disabling adapters

Was this page helpful? Yes No Submitting... Thank you

Comments

  1. Joshua Skirde

    Please advise on how the Repository server is covered by the DR architecture. In the event of a complete loss of "Data center A", I believe that "Data center B" would continue until a restart at which point it would fail as no repository server is found.

    Also, what is preventing workflow from executing on the remote data center? This would result in a significant performance hit in my environment.

    Regards,

    Joshua

    Sep 29, 2013 08:58
    1. Umesh Sehgal
      1. Having two live repository servers in a single environment is not supported. You would need to have COLD repository server in the DR site (but pointing to the same database as used by Primary), which would require manual intervention in the event of primary site failure.
      2. About you second question: I'm not entirely sure if I totally follow the question, but workflow execution would happen in both the sites, the HACDP server isn't a COLD standby.

      -Umesh

       

      Sep 30, 2013 12:24
      1. Joshua Skirde

        Hi Umesh,

        1. Shouldn't the COLD standby Repository be part of this diagram?
        2. With regard to my second point, as you say, workflow execution would happen in both sites across a WAN link. Is there any way to keep workflow execution local? Obviously we can do this for adapters (naming them with a site name for example) but the actual workflow itself is free to be executed on any peer. This could result in significant lag if the WAN link is congested. We currently have a HA-CDP and CDP at each site but this requires manual synchronisation to the COLD site. I would like to move towards this BMC recommended architecture but only when we can be sure we are not executing across the WAN.

        Kind regards,

        Joshua

         

        Mar 31, 2015 12:17
        1. Dorothy Poole

          Hi Joshua, Thanks for your comments. Umesh is going to research and respond.

          Mar 31, 2015 02:57