Alcatel Unleashed

Our company has seven OS6850 switches ( 6.3.1.894.R01) connected in series via 10GB interfaces as our network backbone.
They have a basic configuration with no routing, just 10 VLANs.
On each of these switches there are a number of other switches connected (like HP Procurve 2840 and 2810).

The problem we are experiencing is in slow/erratic response time from the http management gui as well as ping and SNMPv3.
The cpu and network load on the switches are low and there are no errors as far as I can tell.

When looking at health in the web GUI on one switch in the middle of the chain, it takes about a minute for that page to load but then shows 88% free cpu and 31% free memory.
Performing a simple ping -t to the management ip of the switch produces times from 2ms up to occational 3000ms.
I have a Cacti installation performing traffic monitoring using SNMPv3. The graphs produced have gaps in them indicating reoccurring periods of no response.

In the GUI, looking at Physical -> Ethernet -> Statistics -> Counters Errors everything seems ok. No errors on the 10gbit links.
The really weird part is that all HP switches connected to the backbone are performing flawlessly. Consistently low ping even across the entire network, and SNMP monitoring works perfectly.
All of their traffic is sent through the 10Gbit backbone, so a bad link should have affected them as well. All switches are using the same management VLAN and IP subnet.

Any ideas what' could be causing this behavior?

innovatum wrote:Our company has seven OS6850 switches ( 6.3.1.894.R01) connected in series via 10GB interfaces as our network backbone.
They have a basic configuration with no routing, just 10 VLANs.
On each of these switches there are a number of other switches connected (like HP Procurve 2840 and 2810).

The problem we are experiencing is in slow/erratic response time from the http management gui as well as ping and SNMPv3.
The cpu and network load on the switches are low and there are no errors as far as I can tell.

When looking at health in the web GUI on one switch in the middle of the chain, it takes about a minute for that page to load but then shows 88% free cpu and 31% free memory.
Performing a simple ping -t to the management ip of the switch produces times from 2ms up to occational 3000ms.
I have a Cacti installation performing traffic monitoring using SNMPv3. The graphs produced have gaps in them indicating reoccurring periods of no response.

In the GUI, looking at Physical -> Ethernet -> Statistics -> Counters Errors everything seems ok. No errors on the 10gbit links.
The really weird part is that all HP switches connected to the backbone are performing flawlessly. Consistently low ping even across the entire network, and SNMP monitoring works perfectly.
All of their traffic is sent through the 10Gbit backbone, so a bad link should have affected them as well. All switches are using the same management VLAN and IP subnet.

Any ideas what' could be causing this behavior?

At first it would be interesting where your Cacti-Installation is located and where you did the ping-tests to your Omniswitches. I don't think that they have load problems, but did you graph this with Cacti too?
- Did you check the availability of the nearest switch from Cacti?
- Do all Omniswitches show the gaps at the same time and the HP Switches serve SNMP-results at this time?
- Do you consider to update the Omniswitches to actually OS?