Forum Discussion

fraguet_53463's avatar
Icon for Nimbostratus rankNimbostratus
May 12, 2011

Cluster with heterogen platform




Not sure that it's the right topic.... but...



My company plans a big reorganization of it's several datacenter and concentrate them in one.



The datacenter wich will concentrate all the others runs actually with a cluster of BigIP 3900 LTM in version 10.1.



To upgrade the power, we will replace the 3900 by viprion. This operation has to have the minimum impact for the production.



My first idea is to pull off the cluster the passive 3900 and replace it by a viprion, synchronize, force to stanby the active 3900 (then the virpion takes the flow) and replace the last 3900 by the second viprion.



But this solution supposes that a cluster made of one 3900 and one viprion is available. Is it possible ?



Thanks for your answer





2 Replies

  • Frederic,

    I am planning something similar, moving from one platform to another. My approach is to grab a block of IPs (or a specific VLAN) for virtual servers that can be bound to the new BigIP Platform. This allows the new system to be built up and and the virtual servers to be migrated and tested without interrupting production and then when you are ready, you can switch the DNS entries to the new VS address on the new platform.


    While trying to make disparate platforms work as an HA pair may be possible, I think there are too many variables to account for and it leaves you without a viable test environment before the "cutover". A consideration for me is I will be going from a legacy 9.4.3 config to a 10.2 system and attempting to cure the configuration sins of previous admins.




    Hope This Helps,




    Carl B


  • Carl



    Thanks for your answer



    I also thought as you with a second block of vs. The first inconvenient is that I have to reconfigure all the vs actually running (by a copy/paste in the config file and changing ip for vs) so manual action which mean time to pass and possibility of errors. The second inconvenient is that it causes a reconfiguration of my supervision plateform which has view on the actuals vs.



    Working with cluster like I told means no configuration error because it's made by synchronisation and no impact in my supervision plateform. But based on compatibility of 3900 and viprion in a same cluster...... !