HP Procurve 4000M

[Peter_480]

What should my utilization rates be at before I should consider upgrading my switches? My maximum usage is currently 45-75%, but most of the time I am running in the 5-15% range. Any advice anyone can give me would be greatly appreciated.

[Kell van Daal]

As long as the switches aren't a bottleneck (now or in the near future), there is no need to upgrade them for performance reasons.
Other considerations for upgrading them however are things like:
- Do you need extra performance with more Gbit links, or even 10GbE links?
- Do you need more functionality then the 4000m can offer? (routing, higher port density etc)
- Are spare sparts easily obtainable for you (the 4000m is end-of-sale)
- Does the 4000m have all the security options I need? (no 802.1x, no ssh etc)

So in short. If the 4000m would be a bottleneck (or would become in the near future), it would nee upgrading. If it isn't a bottleneck, there could still be valid reasons to upgrade. Performance isn't often the only reason to upgrade.

[Shiraz Malik]

Hi Peter,
you can check the port statistics to have an idea about the future performance of your switch. 4000m has fabric speed of 3.8 Gbps and you can calculate the blocking by multiplying the number of ports with the speed and then multiple it by 2 for full deuplex. If the sum is greater than the fabric speed then its high time to think of a change. You don't really need to worry about the warranty as this device is entitled for life time warranty as most of the ProCurve switches.
Feel free for any other recommendations that you may require.

regards,

[Antoniov.]

I'm with Kelly,
performance is not reason to update switch.

Antonio Vigliotti

[Shiraz Malik]

If you see the symptoms of bottleneck and can find the packet drops on interfaces .. It is then a performance related issue and you should upgrade the switch. Otherwise, if there are no such problems then theres no need to worry.

[Frank Benke_1]

like stated before there basically are many different reasons to update switches which depend not on usage.

the packet drops are an strong indicator and I consider them important on the uplink interface. maximum usage is a relative value and there are still occasions were you consider a usage of 30% to be a switch/ethernet full.

for the final decission I would mirror the uplink port to a analyser like PC based freeware ethereal were you get an analysis of the "disrupted" communication.

personally I would move on from the 4000 to get a more sophisticated management interface.

as well the limited backplane bandwidth makes packet drops more likely and you can not surely distinguish between the egress and the ingress port (icoming / outcoming).

regards