Avatar of mattlast
Flag for United States of America asked on

Fastest way for a port switchover 6509

I'm in the process of a module switch over on one of our 6509s I will be moving critical servers over to our second 6509 I want to minimize down time to as little as possible. I know that enabling portfast and disabling trunking will shave 10 seconds off port convergence but are there any other tricks that could speed this process up.. Even if its only a few more seconds. Every little bit helps. Thanks
Network OperationsNetworking Hardware-OtherSwitches / Hubs

Avatar of undefined
Last Comment

8/22/2022 - Mon

I've been there. Portfast is standard configuration for an access port but you will still lose about 10-15 seconds of connectivity if you are changing the port a server connects to.
If they're critical servers then they are no doubt multi-homed, so you should be able to have one of the connections to the second 6509 and removing the one from the first 6509 shouldn't cause any disruption at all.

Enabling portfast and making the port an access port should be your normal configuration unless you need multiple VLAN's to a server.

Well, the best way would be to have a second NIC in each server, configure the server with NIC teaming and connect the second NIC to the second 6509.
Garry Glendown

+1 on giltjr ... though in order for "real" teaming/portchannel/etherchannel, you will need to configure VSS on the two 6509's ... that way, you should have very low/next to no loss during the switchover ... question is whether the results is worth the work & trouble, or if a scheduled 1min downtime at some impossible time of the night won't achieve the same thing with less hassle ;)
I started with Experts Exchange in 2004 and it's been a mainstay of my professional computing life since. It helped me launch a career as a programmer / Oracle data analyst
William Peck

Which we don't have vss I found that out the hard way the bad thing is not all these server or devices have 2nics and some services are mapped to specific Ips of the servers but not all out side of setting up intel or Broadcom teaming software is there a way to make a passive active environment for the critical servers. Ya we are 24 hour operations so of course the bosses up stairs say we can never have down time..... We all know how that works...

Most NIC teaming software allows either active/standby or active/active setups.

The active/active setups can operate in at least two ways.

One way is full aggregation, which requires the switch side to be configured in a way that it support this, LACP/Etherchannel.

The other way is "partial" aggregation.  The server can send traffic outbound across each NIC, but only receives inbound on one.

The difference is where the NIC's are connected and what the switches supports.

If all NIC's in the team are in the same switch, then you can do it either way.  However if the NIC's are in difference switches the switch has to support "stacking", that is multiple physical separate switches acting as one.

The 6500's don't support stacking, so you can only do partial aggregation.  This gives you increased outbound bandwidth from the server, but not inbound.  It also gives you fail over if one of the NIC's fail.

While you can't stack 6500's, using the Virtual Switching Supervisor Engine 720 in each enables two physical switches to appear as the one logical switch.

It's hard to say if you can set up an active/passive environment without knowing more details of the function the servers provide. Typically, you would have multiple servers providing the same services or an active/standby or cluster set up. If you haven't got these already then it may mean additional hardware, time and money.
Try out a week of full access for free.
Find out why thousands trust the EE community with their toughest problems.
Garry Glendown

View this solution by signing up for a free trial.
Members can start a 7-Day free trial and enjoy unlimited access to the platform.
See Pricing Options
Start Free Trial
Ask your own question & get feedback from real experts
Find out why thousands trust the EE community with their toughest problems.

Thanks guys, I will be trying all the scenarios we have 10 non critical servers that I can test on.   This has given me a starting point.