Marci Posted March 15, 2011 Posted March 15, 2011 TheDude from Microtik - tis free... allows full SNMP monitoring of all switches...
dhicks Posted March 15, 2011 Posted March 15, 2011 Does anybody know of any open-source software which i can use to monitor things on the switch? as the web interface on the switches i have is awful! Do your switches support port mirroring? If so, you could mirror, say, the main uplink port from the switch your servers use and monitor all traffic with Wireshark or similar running on a laptop.
teejay Posted March 15, 2011 Posted March 15, 2011 Have you checked that you haven't got multiple IP addresses for the servers in dns? If you've been playing with teaming etc, you may have some duff dns entries.
aleach2 Posted March 15, 2011 Posted March 15, 2011 I had a friend with a issue just like this. Turned out to be something to do with SMB (Server Message Block) signing. I did a quick search and found the following page: Windows Explorer and SMB Traffic - Ask the Performance Team - Site Home - TechNet Blogs Hopefully you sort it soon. Adam.
oxide54 Posted March 15, 2011 Posted March 15, 2011 Toby, I had very similar problems on our two backup servers, with all the hp nic teaming stuff installed, they were much better once I rebuilt using built in 2008 R2 drivers
glennda Posted March 15, 2011 Author Posted March 15, 2011 Have you checked that you haven't got multiple IP addresses for the servers in dns? If you've been playing with teaming etc, you may have some duff dns entries. nope checked that
glennda Posted March 15, 2011 Author Posted March 15, 2011 Toby, I had very similar problems on our two backup servers, with all the hp nic teaming stuff installed, they were much better once I rebuilt using built in 2008 R2 drivers Thats my plan come the summer - although at the current rate it might be easter which does mean i have something to do! @dhicks - no these switches are awful they don't do much apart from switch and the web interface is useless.
oxide54 Posted March 15, 2011 Posted March 15, 2011 what we had was really weird at times, it would ping some hosts and come back with times of 600ms and others <1ms, and would out and out refuse to talk to some other hosts, and talk to some no problem,
glennda Posted March 15, 2011 Author Posted March 15, 2011 what switches are they? Nortel 5530-24tfd x 2 in a managed stack with Netgear gsm7212 switches bolted on the side. I do think the issues is something switchy related but i've got to prove my theory to be correct before they will give me 10k to replace the switch.
oxide54 Posted March 15, 2011 Posted March 15, 2011 Nortel 5530-24tfd x 2 in a managed stack with Netgear gsm7212 switches bolted on the side. I do think the issues is something switchy related but i've got to prove my theory to be correct before they will give me 10k to replace the switch. how are the netgears "bolted on"
glennda Posted March 15, 2011 Author Posted March 15, 2011 how are the netgears "bolted on" Check the image attached on the previous page. It got them out of the whole two summers ago but needs replacing we just don't have the money!
oxide54 Posted March 15, 2011 Posted March 15, 2011 so you've got 12 x 1gb ports connecting to the "server" core via 1 x 1gb fibre.?? who's your boss ? person i think it is?
oxide54 Posted March 15, 2011 Posted March 15, 2011 can't you get a few more nortels and add them to the stack? wouldn't that be cheaper than replacing the whole lot? i haven't seen them, but surely they allow for more than 2 in a stack?
Abaddon Posted March 15, 2011 Posted March 15, 2011 Looking at your diagram, surely this is just a network saturation problem? Two switches with 12Gb each, connecting to your core with only a single 1Gb uplink each? What's your stats like on those connections?
Michael Posted March 15, 2011 Posted March 15, 2011 can't you get a few more nortels and add them to the stack? wouldn't that be cheaper than replacing the whole lot? i haven't seen them, but surely they allow for more than 2 in a stack? Looking at your diagram, surely this is just a network saturation problem? Two switches with 12Gb each, connecting to your core with only a single 1Gb uplink each? What's your stats like on those connections? You're both right as Nortel like others brands use fast stacking architecture which allows you to connect switches @ 40Gbps each, which cuts out any bottlenecks. The problem being however is the Nortel's may be the root of the problem, so adding to it won't make any difference. It's not impossible, but maxing out a 1Gbps to the server is pretty difficult and you'd get very high CPU and disk utilisation, but as far as I can make out Glennda hasn't mentioned this, so it appears to be a networking related issue. In theory you should be able to update the firmware on the switches themselves and reset any configuration to factory default. I've never installed any Nortel's, so I've no idea how good or bad they are.
teejay Posted March 15, 2011 Posted March 15, 2011 I think at this point as previously mention by someone else, I'd be firing up Wireshark to see what sort of traffic is bouncing around the network.
glennda Posted March 15, 2011 Author Posted March 15, 2011 In theory you should be able to update the firmware on the switches themselves and reset any configuration to factory default. I've never installed any Nortel's, so I've no idea how good or bad they are. awful - there is no updated firmware for these switches (that i can actually find as the nortel site takes you round and round in circles). They are set to default just with tree spanning enabled and IGMP snopping plus the port configs set to fiber or copper. @abaddon We don't use all 12 1gb ports (about 6) and the utilization is loads hits about 60% at most for the uplink when i've been watching it - I think it is the switching problem (with the nortels) but my main point in posting was to double check that i hadn't missed anything out in my working out what the problem was and to double check i had been through absolutly everything before making them spends thousands on a replacement which could effectively do nout! @oxide54 - it is that person but he wants to be a bit more sure that it is the problem like me doesn't want to be spending all that money if its not going to make a difference. and the Nortels are around 4 years old. The model of switch that i'm looking at (hp E4800g) is the similar to the 3com that you have there (as hp bought 3com)
oxide54 Posted March 15, 2011 Posted March 15, 2011 You're both right as Nortel like others brands use fast stacking architecture which allows you to connect switches @ 40Gbps each, which cuts out any bottlenecks. The problem being however is the Nortel's may be the root of the problem, so adding to it won't make any difference. surely the 12gb into 1gb is the problem here?
oxide54 Posted March 15, 2011 Posted March 15, 2011 awful - there is no updated firmware for these switches (that i can actually find as the nortel site takes you round and round in circles). They are set to default just with tree spanning enabled and IGMP snopping plus the port configs set to fiber or copper. @abaddon We don't use all 12 1gb ports (about 6) and the utilization is loads hits about 60% at most for the uplink when i've been watching it - I think it is the switching problem (with the nortels) but my main point in posting was to double check that i hadn't missed anything out in my working out what the problem was and to double check i had been through absolutly everything before making them spends thousands on a replacement which could effectively do nout! @oxide54 - it is that person but he wants to be a bit more sure that it is the problem like me doesn't want to be spending all that money if its not going to make a difference. and the Nortels are around 4 years old. The model of switch that i'm looking at (hp E4800g) is the similar to the 3com that you have there (as hp bought 3com) that 3com i have is a pain if you need to config via the serial port, the cli is complicated (i do it when i have to but id rather use the web browser with particular switch) our edge switches i'd rather use the cli
Michael Posted March 15, 2011 Posted March 15, 2011 surely the 12gb into 1gb is the problem here? Well according to Glennda the utilisation is 60% or 600Mbps of 1000Mbps, so no it doesn't appear to be a problem. Glennda's right to have a 1Gbps fibre connection to each cabinet. The data transfer to each cabinet is going to be 50Mbps (if all 12 fibre connections are in use) in a perfect world where all was equal Edit: Thinking about it even more, there are three servers, so the 600Mbps could in fact be 200Mbps transfer from each server with its own 1Gbps connection (again in a perfect world if all was equal)!
CyberNerd Posted March 15, 2011 Posted March 15, 2011 They are set to default just with tree spanning enabled and IGMP snopping plus the port configs set to fiber or copper. set the root bridge priority on the core ! otherwise the switch with the lowest MAC address wins any election - your probably sending data to the other side of the network....
glennda Posted March 15, 2011 Author Posted March 15, 2011 set the root bridge priority on the core ! otherwise the switch with the lowest MAC address wins any election - your probably sending data to the other side of the network.... The top one in the stack is set, I just described the basic config of things enabled
glennda Posted March 15, 2011 Author Posted March 15, 2011 I've just got 95% confirmation that i can order the new one!!!! aslong as it is under 12k pretty much
CyberNerd Posted March 15, 2011 Posted March 15, 2011 good. Call Mike Curry at switchshop. He's very good. 01438 831870 Switchshop Ltd for your networking needs
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now