LAE

 View Only
  • 1.  Disaster recovery test on Lavastorm instances

    Employee
    Posted 01-31-2015 11:05

    Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

    Originally posted by: vijayramachandran

    Hello everyone,

    I don't know it is the right topic to discuss in here, but I believe some of you would have an idea for it.

    I want to perform DR test for my Lavastorm servers, which has individual analytic engines instances running on each server.

    I have two PROD and two DR, and two servers are awaiting for PROD activation before that I want to perform DR for those servers as well.

    Tell me one thing, how to perform DR for Lavastorm servers. In the past, it was done alike, configuring a sample graph, which is run on one server via Control-M, and checking it in DR also by swapping DNS with a CNAME. I really don't understand that test, and what he witnessed out of it. Let's not worry about it.

    So, in order to kickoff this process in my case, should I setup a sample graph on PROD via Control-M, and see it is running on DR, while the time I bring the Lavastorm service down on PROD to check the DR is working fine? Does it make sense?

    If it's an URL based application, which comes up on a webpage, we can easily test, but it in this case, we can do only by scheduling a test graph as a script to fire at the right time on Lavastorm servers using Control-M, and what are the possibilities to proceed with this DR test?

    Any ideas??


    Thanks,

    Vijay


  • 2.  RE: Disaster recovery test on Lavastorm instances

    Employee
    Posted 01-31-2015 11:52

    Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

    Originally posted by: stonysmith

    You are welcome to bring the question up here, but it might be best for you to have a chat with your Technical Account Manager to explore what solutions best fit your needs.

    What level of Disaster Recovery are you looking for?

    1) running node continuation on 2nd server
    2) running graph (controller) continuation on 2nd server
    3) FARM partial failure
    4) failed graph restart in middle
    5) failed graph complete rerun from start

    Situation 1 is usually only available on machines with special hardware and operating systems.
    LAE does not at this time have a facility to handle situations 2 or 3.
    Situations 4 and 5 can be implemented, but both are dependent upon Control-M, not LAE.

    An important part of any DR consideration is whether a second server has access to the temporary disk files.
    Common configurations of hardware involve SAN/NAS disk drives - and that alone represents a possible point of failure.
    The connection to the disk space must be available on the 'recovery' server or you won't be able to implement a restart facility.

    ====
    When you execute the LAE Controller, an option you have is to produce an output BRS using the -dump filename.brs command line option.
    That BRS can then be used to re-start a failed graph, assuming that the referenced temp files are available to the second server.
    Note: the temp files must be available on exactly the same filenames, so any disk mounts must be available with the same names on both machines.


  • 3.  RE: Disaster recovery test on Lavastorm instances

    Employee
    Posted 02-28-2022 05:28

    What would be the special condition of Hardware and Software to meet requirement number 1? I have this need right now