Old core switch query

[BJG]

We're currently running an old HP 4108GL with a bunch of modules as our core switch. I've been thinking I should probably replace it, but the miser in me is tempted to keep it going. Additional/spare modules go for peanuts on eBay. So give me counsel...do I need to kick this thing out, or spend a large amount of cash to update something that still basically works fine...? If I was to replace it, what kind of thing should I be looking at?

[3s-gtech]

What doesn't it do that you need it to do?

Does it struggle with current network loads?

Can it handle any future upgrades?

 

You need to evaluate it first - don't just upgrade it because it's old - being a Procurve that's not an issue. We replaced our old core switch because it couldn't handle the quantity of fibre connections that we were having installed and the number of teamed connections from servers so it made sense.

[BJG]

What doesn't it do that you need it to do...don't just upgrade it because it's old

 

This is what I'm thinking, though I was looking at a replacement module on eBay for £15 (list price $1000) to change some of the connections and wondering if I was being sensible.

[localzuk]

I'd ask the following questions:

 

If it breaks, how quickly would it be repaired/replaced? Is it in warranty still?

Does it fulfill your network's needs and any future needs you have planned?

 

If it is still under warranty and could be replaced next day (or sooner), then I'd move on to the second question. If it does everything you need, and can also do the things you wish to do over the next few years (per your network plan), then I'd leave it be.

[6Foot2]

We have recently found a limitation with our core switch. It can only broadcast a WOL message across some of our VLans. This is annoying as it limits the number of stations we can wake up, update and close down again overnight.

 

Core switch: HP 5406 zl.

 

I would update this if I could but as the switch is performing well on other fronts I can't justify the cost.

[3s-gtech]

I'd ask the following questions:

 

If it breaks, how quickly would it be repaired/replaced? Is it in warranty still?

Does it fulfill your network's needs and any future needs you have planned?

 

If it is still under warranty and could be replaced next day (or sooner), then I'd move on to the second question. If it does everything you need, and can also do the things you wish to do over the next few years (per your network plan), then I'd leave it be.

 

It's Procurve. Lifetime warranty (100 years practically).

[BJG]

Thanks for the comments - just needed a reality check. I just found another 4108GL crammed with similar modules on eBay for £45, so I've grabbed that as a backup. I guess there's life in the old dog yet...

[3s-gtech]

The Quickspecs are always a handy reference when looking at future needs too:

 

http://www8.hp.com/h20195/v2/getpdf.aspx/c04283170.pdf?ver=5

 

The fabric and maximum port numbers aren't bad for the 4108GL, depending on configuration. Would probably be fine for mid sized secondary or below (depending on VLAN configuration too). No 10GB, but that may be a way off for you yet. They confirm the lifetime warranty.

 

Can't go wrong for £45. I bought some modules for our old DES-6000 years back, they were £90 each for two GB ports per card!

[mrbios]

How big is your site and router setup in terms of complexity?

 

I can't remember what it was about ours, but we actually went over capacity on our 4108GL and were forced to upgrade. It would randomly crash of its own will due to the capacity issue (It was about 7 years ago so i can't remember in what aspect we went over its limits, i wasn't really the one dealing with it then. I just remember it was a technical reason rather than physical one, we still had room for modules etc.) we've moved up to an 8206zl now.

 

Wish i could remember the details so i could be more specific....saying that i remember us being stuck on a really old firmware due to an issue with something we were using being taken out of later firmwares, but again can't remember what that was, and could have been the reason for the capacity issue.