Large network optimization/monitoring

Question

Hello, I think this might be beyond the scope of support so I thought I'd ask here. We have a large network, counting servers (all dell), switches (35 total), mix of older Dell PowerConnect and force10, dracs, and VMs we have around 2300 IPs in use, plus a very large amount of containers, and we move hundreds of TB's per day.  There are 21 VLANs.  All the VLAN routing is done by a pair of switches (S5232F-ON running 10.5.4.4) in VLT bonds.   For the most part everything is on their default settings.  What I'm wondering is what should/could I enable/disable to make things more efficient.  Cut down on unnecessary traffic.  Anything to optimize?  We only use jumbo frames where the application calls for it. These are the areas I'm wondering about: Arp settings Mac settings mld, igmp snooping spanning tree between the force10 and PowerConnect switches We have found a few things on the servers that have helped, expanding the receive buffers in the OS has helped in some situations.  I've enabled the three types of storm control, broadcast, multicast, and unknown multicast. When there were some issues updating the switches in the Vlt bonds, both switches in the pair would go down for around a minute when the second switch was updated and rebooted, i had support help with the updates.  They did say there were a high number of mac address moves.  Our developers say its expected because of the large number of containers that are being constantly created and destroyed, each having a virtual interface. I have multiple tools for monitoring the network and the equipment.  One thing that none are really good at is pinpointing what might be causing an issue.  Exact reason for drops, discards, etc...  It seems like without a packet capture and wireshark, tracking potential issues aren't easy.  I have tools that will show the number of discards or errors, but give no reason behind them.  None of the discards are caused by a mis-tagged vlan. I realize this is a lot of information and not the easiest thing to help diagnose without being there to dig into it.  That's why I'm hoping for any help with any settings for a relatively large network with a lot of IPs.   Any help/advice would be greatly appreciated

DELL-Young E · Answer

Hello, thanks for choosing Dell.Is your system still in warranty? We would like to ask you to collect switch logs from your S5200 to know of your network topology and maybe give you some idea for best practice.Wish you a good one.

Alanon2 · Answer

Yes, our two cores and 16 of our tor switches (s4148t-on) are under warranty until 2025.   What would you like me to gather and how should I get them to you?   Thank you very much.

DELL-Charles R · Answer

Hello Alanon2,

You can log the output of : #show tech-support

Then upload that and a topology to this url under the service tag: https://upload.dell.com/

Next send Young a Private Message with the service tag so the uploads can be retrieved.

You can click on name: DELL-Young E

Look in the Right hand column:

Contact Me

Send this user a private message with service tag

Alanon2 · Answer

Thank you.  I'm getting 'service unavailable' when I try to verify the service tag on the upload page.  I'll try later tonight or in the morning.

DELL-Charles R · Answer

Hello Alanon2,   Thank you for the update. Please allow us some time to collect the log, review and follow up with you.

Alanon2 · Answer

I've uploaded the out put and sent Dell-Young E the private message.   Thank you

DELL-Young E · Answer

Hello. As per both logs, there is no abnormal alerts found on the switch. Configuration looks good on VLT switches. That's what we could tell you as techsupport. Since there's no break/fix we need to tend to, now this should be address by deployment or consulting team for best practices. I would say the online forum isn't the best option for you right now.

Networking General

Was this post helpful?