2 Replies Latest reply on Apr 17, 2016 2:20 AM by Gilganik

    ISCSI bottleneck LAG(LACP) only 1 Gbe

    Gilganik Lurker

      Hi all!

       

      I'm trying to explore the performance world about ISCSI.

       

      I'm trying to reach theorically double of the speed using a configuration in LACP but something is going wrong. I always obtain the speed as i used only 1 Gb.

      This in my configuration:

       

      - Storage: Qnap 453 Pro. 2 Gbe Ports in trunk with 802.3ad protocol. 9000 MTU. 2 SSD Samsung 850 EVO in RAID 0. LUN configured in ISCSI mode (default setup)

       

      - Phisical Switch: Netgear GS108Tv2. all ports have MTU 9216. 2 Ports are configured in LAG (LACP) access vlan 10 (ESX Trunk). Another 2 ports with the same configurations (NAS trunk)

       

      - Esx 6.0: I've configured VDS with 2 phisical adapters in LAG mode. LAG is the only active uplinks for the Porgroup i've created. I've setup the VDS with 9000 MTU

      as "Load Balancing teaming" i've tried all the configuration the LAG proposes to you, but nothing changed.

       

      - VM: I've installed a Windows 2012 R2 VM With these configs:

      -- DISK TYPE: Eager Zeroed

      -- Disk Mode: Dependent

      -- Controller: LSI Logic SAS

      --vNIC: VMXNET3

       

      For the test i'm using "Crystal Disk" to see how fast are Read/Write performance of the VM disk.

       

      I cannot understand why i obtain always 117 MB/S R/W speed (sometimes 90 in write)

       

      What i'm doing wrong? Any ideas?

       

      Thanks in advance