IC SunsetThe developerWorks Connections platform will be sunset on December 31, 2019. On January 1, 2020, this forum will no longer be available. More details available on our FAQ.
Topic
  • 4 replies
  • Latest Post - ‏2019-11-06T17:16:30Z by stuart.haddock
TMaverick
TMaverick
1 Post

Pinned topic V5000 IP replication bandwidth issue

‏2017-05-13T11:05:16Z | v5000

Hi,

I have two V5030 id different DC's running native IP replication between them, using GM changed volumes on IBM i host with 72 volumes=20TB. Even though I have 1Gbps link bandwidth the replication can't sustain more than 30-40MBps. The network has been tested and FTP and other transfer are able to sustain 90-100MBps speeds, but not for the replication. The remote partnership on the V5000's are configured to use compression and set for 1000Mbps=1Gbps, 100% background copy rate. I'm using 1 Remote Copy Group with ports 2 on each node assigned to it.

IBM support identified that I have 1.5% lost packages on 15 minute interval and therefore the bad replication throughput, but network team can't see any lost packages on the ports I use for the replication. I also see a very strange behavior, when I switch off the compression of the partnership the replication bandwidth goes below 10MBps.

 

Any ideas would be helpful.

  • JuAnKmil0
    JuAnKmil0
    1 Post

    Re: V5000 IP replication bandwidth issue

    ‏2018-11-08T16:21:00Z  

    Hi

    I have the same issue. Have you got any response from IBM support?

    Recently I replaced two old FCIP routers with V7000 native IP replication. I can't get bandwidth use at 100% and some of replication relationships are lagged behind my desired RPO.

     

    Thanks.

     

    JC

     

  • Ricardo_Alias
    Ricardo_Alias
    1 Post

    Re: V5000 IP replication bandwidth issue

    ‏2019-07-02T07:53:28Z  

    Hi

    I have the same issue. Have you got any response from IBM support?

    Recently I replaced two old FCIP routers with V7000 native IP replication. I can't get bandwidth use at 100% and some of replication relationships are lagged behind my desired RPO.

     

    Thanks.

     

    JC

     

    Hello,

     

     we have the same issue with two v3700, line bandwidth is 300Mb/s but replication is using 90Mb/s, the consolidation time between Consistency Groups is high. Do you resolve this issue? Thanks in advance.

  • stuart.haddock
    stuart.haddock
    76 Posts

    Re: V5000 IP replication bandwidth issue

    ‏2019-11-06T16:45:34Z  

    Check that the switch is not throttled down (I know), Changed ports/cables/switches? Not using those port for iSCSI? Not on compressed volumes?

    4.1.2 Performance considerations The following advice has come out of field experience: 2145-CG8 with request for price quotation (RPQ) 8S1296 is advised if IP replication and IBM Real-time Compression are being used. This is advised so that processor-intensive services have adequate resources for optimal usability. If the nodes performing IP replication are 2145-CF8 or earlier, and Real-time Compression is being used, configure ports for the IP replication traffic on I/O groups not containing compressed volumes. This is advised so that processor-intensive services have adequate resources for optimal usability. If Internet Small Computer System Interface (iSCSI) is used for host connectivity, do not use the same port for host I/O and IP replication traffic. Optimal performance can be achieved by isolating the different traffic onto different ports. Restriction: IBM Storwize V7000 and IBM System Storage SAN Volume Controller (SVC) nodes with more than two I/O groups are supported. However, there is a limitation that only four IP addresses from two I/O groups can be included in a remote copy port group. Important: Decreased throughput on the source back-end controller can adversely affect the consistency of the throughput of IP replication traffic.

     

     

  • stuart.haddock
    stuart.haddock
    76 Posts

    Re: V5000 IP replication bandwidth issue

    ‏2019-11-06T17:16:30Z  

    Check that the switch is not throttled down (I know), Changed ports/cables/switches? Not using those port for iSCSI? Not on compressed volumes?

    4.1.2 Performance considerations The following advice has come out of field experience: 2145-CG8 with request for price quotation (RPQ) 8S1296 is advised if IP replication and IBM Real-time Compression are being used. This is advised so that processor-intensive services have adequate resources for optimal usability. If the nodes performing IP replication are 2145-CF8 or earlier, and Real-time Compression is being used, configure ports for the IP replication traffic on I/O groups not containing compressed volumes. This is advised so that processor-intensive services have adequate resources for optimal usability. If Internet Small Computer System Interface (iSCSI) is used for host connectivity, do not use the same port for host I/O and IP replication traffic. Optimal performance can be achieved by isolating the different traffic onto different ports. Restriction: IBM Storwize V7000 and IBM System Storage SAN Volume Controller (SVC) nodes with more than two I/O groups are supported. However, there is a limitation that only four IP addresses from two I/O groups can be included in a remote copy port group. Important: Decreased throughput on the source back-end controller can adversely affect the consistency of the throughput of IP replication traffic.

     

     

    Are you getting a lot of 1920 - 1720 errors? There is a new V8 Redbook https://www.redbooks.ibm.com/redbooks/pdfs/sg247938.pdf

    Check section 11.11 Troubleshooting remote copy page 599.