1. Home
  2. Ethernet LAN
  3. Procurve and redundancy.

Procurve and redundancy.

Hi,

We are trying to create a redundant internal network with the following configuration :

[ports b1, b2, c1, c2, d1, d2 ] [ on each switch is connected ] [to the same port on the other] /-------------\\ /-------------\\ | Procurve | | Procurve | | 5308 |--------| 5308 | core switches | | | | \\-------------/ \\-------------/ | | | | | | | | | | | | | | \\---\\ /--/ | | | | | | | | | | | | | | | | ---------------- | | | | | Procurve 2848| | | user switches | | ---------------- | | | | | | | | | \\-------\\ /---------/ | | | | | | | | | | | V | ---------------- | | | Procurve 2848| | | ---------------- | | | | | \\-----------\\ /-------------/ | | | | ---------------- | Procurve 2650| ----------------

i.e. on each 'user' switch, there will be two uplinks to our core switches. I setup a monitor port on each of the switches and could see the entire network traffic on all of the switches.

I understand this is because I have not configured each of the user switches to recognise that there are two uplinks. The core switches see all possible mac addresses in each of the user switches, because all mac-addresses appear to be connected to a switch plugged into the other uplink port.

I tried configuring through the menu system a trunk (Trk1) on the user switches, and added ports 1 and 2 to the trunk. This was setup as one time as a basic Trunk and LACP trunk.

I still see all of the network traffic on the monitoring port.

Spanning Tree is enabled to try to prevent loops or broadcast storms.

This indicates to me that I clearly have not managed to make the switch aware that the two ports are associated.

How do I achieve what is listed in the diagram, please ?

Reply to
alstamp
Loading thread data ...

I forwarded your post to someone I know in ProCurve land, and they stated there wasn't enough information in the original post to formulate a reply. He suggested making an HP ITRC forums post in:

formatting link
or placing a trouble call. The ProCurve folks hang-out in the ITRC forums.

hth,

rick jones

Reply to
Rick Jones

The problem with your setup is that you are trying to do load balancing in a bridge network which is not possible without a lot of additional engineering like, for instance, using multiple spanning trees.

If you try to treat each set of uplinks as a trunk, the problem is the links are going to 2 different systems. With LACP, the trunks will not form. I'm not sure what the behavior would be with static trunks.

In any case, spanning tree _will_ take care of loops, but it will do so by blocking one or more ports some where in the network. If the port that blocks is the one between the 5308s, then until the addresses to which traffic is being sent by the 2848s, all traffic will be forwarded on both links. Also multicast traffic will continue to be forwarded on both links forever.

Ano> Hi,

Reply to
anoop

^^^^^^^^

source address from the 2848s to be learnt by both 5300s.

Reply to
anoop

Perhaps, you seem to be looking for a feature similar to Nortel's Split Multi Link Trunking (SMLT). Does HP provide an equivalent feature? With connections like what the above diagram shows and the Spanning Tree Protocol running, load-sharing is not possible. Redundancy still exists, albeit with the Spanning Tree reconvergence delay.

Reply to
p_cricket_guy

To bad the edge switches are 2848's and 2650's. If he had some slightly more robust edge switches he could do it all at layer 3 using OSPF and ECMP. Reconvergence in my redundant network after a link failure happens quick enough that VOIP calls in progress can reroute on the fly without dropping.

(I use 3400cl's at the edge and 9304's at the core)

Reply to
snertking

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.