-1

We have attempted to implement NIC teaming on our server that has 4 total NICs. There are 2 connections to each switch. I have a diagram to help explain: Server NIC Layout

When we attempt to create a team with all four NICs, we have connection issues and constantly drop RDP sessions. Right now we have a team of two NICs, that connect to each switch. This provide redundancy but I would like to provide some bandwidth aggregation as well. The switches are a Cisco SG300-28 and a Cisco Catalyst 2960. We have tried the 4 NIC team in switch independent mode. Is there a way to get all four NICs into a stable team?

jchadburn
  • 1
  • 2

1 Answers1

0

I would create 2 LAGG/port channel groups, one on each switch. Then configure windows to use one port channel group as primary, then fail over for the other. Since you don't have the ports configured in a VPC, I don't think it will work. I believe windows is expecting all ports in the switch to be in the same LAGG/port channel.

Linuxx
  • 1,189
  • 8
  • 7
  • After configuring the switches for LACP the server connection was still unreliable. The best I could do was configure one team for the host and one team for the couple hosted VMs. – jchadburn Jan 13 '17 at 20:30
  • @jchadburn Since your switches don't support VPC (nexus), then the only solution to give you the redundancy you are looking for is stacking. I don't think your models will stack, so your solution is the only way. I don't trust MS when they say it's switch independent. There are way too many factors that come into play, model, firmware, etc. More than likely the disconnects you were seeing was spanning tree flapping on you. – Linuxx Jan 13 '17 at 20:56
  • Sounds like something I could use to justify a second Cisco 2960(Stack-able) to replace the SG-300 that is the backup switch. I'll keep that in mind for the next round of budget proposals. I had set spanning tree to port fast but it did not resolve it. – jchadburn Jan 13 '17 at 21:32