PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 10/19/2017 2:25 PM by  JoeD
Disaster Recovery
 4 Replies
Sort:
You are not authorized to post a reply.
Author Messages
JoeD
Project leaded / Developer
US Bank
Basic Member
(12 points)
Basic Member
Posts:8


Send Message:

--
10/19/2017 1:18 PM
    We have recently upgraded to LSF 10 and LMK 11. We used to be able to copy our production data to our DR data, make a few adjustments and bring up the systems.
    Infor is telling us that it is impossible now. Their solution would take 48 hours and to test we would need to shutdown our production system.
    Seems extreme. I am interested in what other do for their disaster recovery process.
    0
    Leonard Courchaine
    Private
    Private
    Veteran Member
    (128 points)
    Veteran Member
    Posts:50


    Send Message:

    --
    10/19/2017 1:33 PM
    We've been on LSF 10 and LM 10.1 for 2+ years now and are doing both Failover and DR successfully (we think?!) We're AIX and so rely on IBM HA for failover and DR. Our configuration is an Oracle box, an LSF box and a Landmark box (all aix). For installation, hostnames became critically important for both, especially Landmark. So the installation had to be done, planning for the Failover and DR. When we do the Failover or the DR failover, HA calls a script I wrote that brings everything down on LSF and Landmark. Then HA brings up LSF first, again calling a script I wrote, and then Landmark, ditto. The only snags that we plan for are:
    1. We're on TDS and it's really fussy about node names (hostnames) - so I have processing in my script that watches the hostname and makes an appropriate update to a TDS file
    2. Landmark hostname - Before things are brought up on Landmark we need to run a couple simple AIX commands to fool with the hostname as Landmark is extra fussy about that.

    Otherwise, we've done successful user testing/verification of both Failover and DR failover.
    Holler with questions,
    Lenny
    P.S. It wasn't simple, to be clear!!
    0
    Leonard Courchaine
    Private
    Private
    Veteran Member
    (128 points)
    Veteran Member
    Posts:50


    Send Message:

    --
    10/19/2017 1:34 PM
    P.S. AVAAP did our installation and they were awesome at understanding the funky hostname tweaks that were needed and, especially, around Landmark.
    0
    Kwane McNeal
    Private
    Private
    Veteran Member
    (1197 points)
    Veteran Member
    Posts:399


    Send Message:

    --
    10/19/2017 1:53 PM
    Ok, this happens to be an area I have a depth of knowledge, so let me see if I can help.
    First, let's clarify terms.
    1) DR is usually a host level clone. I'll refer here as system-failure ("hard") DR
    2) DR where the hardware and OS ARE not cloned or synchronized is referred as component-failure ("soft"). This typically involves cloning individual file-systems, multi node clusters, or realtime file copies
    3) DR where the data is copied, is just a data refresh

    As far as a data refresh of an existing DR environment (as referred to by JoeD) this is possible, but is quite tricky. I have a process that works, and is certainly not 48 hours, but is messy and not well tested as yet. Connect with me offline if you're interested in discussing.

    Next up is component level DR. I have successfully made this work, but when looking at the work I've done, there are cases where I'm inclined to say it's easier to reinstall portions that to reconfigure them.

    Full system DR is extremely straight-forward, and if installed correctly, is rather painless to do. As Lenny alludes to, it's all about the host name.


    0
    JoeD
    Project leaded / Developer
    US Bank
    Basic Member
    (12 points)
    Basic Member
    Posts:8


    Send Message:

    --
    10/19/2017 2:25 PM
    We run Windows and DB2.
    We have have an environment setup for LSF and LMK. Our database is mirrored to the DR database. For an exercise or a disaster the file system is added and the database is ready.
    So we have host name issues that we need to clean up on the Landmark side of things.
    Not sure if you would call that a soft DR or a component DR.

    My idea is to extract the LMK Gen database to the server, then after we do the Prod split to DR I would import the LMK Gen data back.
    Infor is coming back and going to reinstall LMK around the beginning of November.
    I appreciate the posts. I will touch base when we get the system running as a separate environment.

    Thanks
    0
    You are not authorized to post a reply.