Visit Open-E website
Results 1 to 8 of 8

Thread: Replication bandwidth

  1. #1

    Default Replication bandwidth

    Hi!

    Here's two DSS running volume replication on 15 iSCSI volumes.
    The replication network is Gbit (crossed p-to-p wire).

    The bandwidth of any replication task can be defined separately.
    Is it useful to set different values, depending on size and write traffic
    of each volume?

    Any chance to modify the bandwidth of a certain replications task, when
    the traffic constraints of the volume have changed?

    Regards,

    Robert

  2. #2

    Default

    Hi Nipper

    Once the bandwidth is set for the replication task, you cannot change the Bandwidth.

  3. #3

    Default

    However, I believe you can create a new replication task without clearing the metadata (so resync only takes a few seconds) that uses a different bandwidth.

  4. #4
    Join Date
    Nov 2008
    Location
    Hamburg, Germany
    Posts
    102

    Default

    Yes, I do think that as well. Removing a replication task does not remove the metadata. I do also think, that I have read this somewhere else on the forum already.

    Use one replication task to initially sync local and remote and then use another to keep them in sync.

    Cheers,
    budy
    There's no OS like OS X!

  5. #5

    Default And what about different values for bandwidth

    Thank you all for the hints! I'll test the issue of kept metadata.


    And what do you know about my other issue: It is possible to define an own bandwidth value for each volume. Thus, is it recommended to choose the maximum for all, or is it better to split/grade these?

    Best,

    Robert

  6. #6

    Default

    Quote Originally Posted by budy
    Yes, I do think that as well. Removing a replication task does not remove the metadata. I do also think, that I have read this somewhere else on the forum already.

    Use one replication task to initially sync local and remote and then use another to keep them in sync.

    Cheers,
    budy
    I tested this and, indeed, it works:

    - have a replication task using bandwith X
    - volumes are consistent
    - stop replication task
    - remove replication task
    - remove reverse(!) replication task on the mirror
    - create new replication task for same volume with different bandwidth Y
    - start new replication task
    - volumes are consistent _immediately_, no new initial sync necessary!

    This is fine.

    Robert

  7. #7

    Default Update on bandwidth param

    Good afternoon!

    We also have had poor performance on replication over a WAN/slow T1 connection. After perusing several postings relating to replication performance, I still don't have a clear picture on what the bandwidth parameter does in setting up a replication job. According to everything I've seen, the whole system, end-to-end, is limited by the slowest link.

    So what does this bandwidth parameter do?

    Gregg

  8. #8

    Default

    Quote Originally Posted by gregg_hughes
    So what does this bandwidth parameter do?
    My experience:

    Suppose that the volume replication is running over 100 Mb/ network connection.
    Usually DSS can provide the transfer speed much higher than 100 Mb/s.
    That means that volume replication will take all available bandwidth of the network
    connection, thus considerably slowing down any other possible traffic.

    So, in this case setting bandwith to 5 MB/s will be a good solution. In this case replication
    will only take 50% of the available bandwidth, leaving enough space for other network activity.

    A note: It is only important during initial replication (or in other similar situations when a big amount of changed data must be replicated).

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •