synaesthesia Posted December 5, 2013 Posted December 5, 2013 (edited) I'm not 100% convinced my link aggregation setup for ESXi to our (HP) switches is spot on. Everything works, there's no apparent issues flagged on the switches but every now and then it seems clear the network is bottlenecking. Would anyone be so kind as to provide information/screenshots on their exact setup? (feel free to mask bits for security/anonymity!) What I would like to check is: Vswitch setup with multiple physical NICs link aggregated to a HP switch. Better still if VLAN'd. And the relevant HP switch setup for it to (text config or screengrab of the gui if you're masochistic!) Many thanks in advance (Rough idea of what I have currently which seems to drop a few more packets than I'd like) HP setup (Procurve 5406ZL) - 3 ports (trunk group TRK1) LACP disabled, trunk type "Trunk". Flow control disabled. Vlan 100 (servers) tagged, untagged to default ESXi (Vsphere) setup - management network goes to a separate port in switch (vmnic 0) "vswitch2" contains the guest VMs connected to 3 physical adapters (vmnic1 to 3). Vswitch on MTU1500, load balancing route based on IP hash. All 3 adapters active, full gigabit. Servers on Vlan 100, load blaancing via IP Hash Edited December 5, 2013 by synaesthesia
PotNoodleTech Posted December 5, 2013 Posted December 5, 2013 Can you get us a print out of the port statistics for each of the three trunk ports? You may find only one of the three ports in the tunk is transferring most of the data. Link aggregation is often not the magic bullet for doubling bandwidth that people think it is!
synaesthesia Posted December 5, 2013 Author Posted December 5, 2013 (edited) True indeed, for us it's more about resilience and taking the edge of things with our minimal budget Edited December 5, 2013 by synaesthesia
synaesthesia Posted December 6, 2013 Author Posted December 6, 2013 (edited) Also finding it odd that packets would get dropped to one guest but not another on the same host. Same OS (2008R2), same vnic on the same vswitch. Not many - talking a packet (well, a missed ping) an hour. VM to VM networking seems "suspect". Iperf gives us a max of 15Mb/s between two guests on the same host. 800 from my machine (client) to either machine though. Edited December 6, 2013 by synaesthesia
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now