network measurement

[cma]

Hi,

I´m getting some complains about the network being to slow some times. My IT park is in one building (A), i have a 4510 switch that connects by fiber at 1 GB to another building (B) with a 2960 switch. People who work in B connects via RDP to servers in A. Some times, and i experience the same, it gets to slow. I measure my network with iperf3 and always get excellent results. I´m out of ideias. Help please.

Thanks

[Flavio Miranda]

Hi,

  Start by the simpler test, how does a extended ping with big MTU looks like?  Do you have rate limit somewhere?  How fast your DNS revolve names on the network?  What about trace route? 

 What about Server CPU and Memory when you get slow?

-If I helped you somehow, please, rate it as useful.-

[cma]

Hello,

First of all sorry for the late reply. Extending MTU to 10000 works fine, DNS resolves names very fast and traceroute works like is expected. The only thing that we dont measure is the server, but like i posted before, people in the main building, where the servers are, doesn't complaint, only people outside of the main building.

Thanks

[chrihussey]

Sorry, but have you checked the interfaces connecting the switches for errors, excessive broadcasting or utilization spikes?

Is there perhaps a user or users at the remote building doing excessively large transfers or back-ups in the middle of the day?

Just some thoughts.

[cma]

I think i checked everything, network side, but sometimes it get´s slow accessing a document management application server in the main building. Follows some relevant information about the connections between the main building and other building:

Main Building switch:
interface GigabitEthernet8/4
description BuildingA - [] Gi1/1
switchport trunk allowed vlan xxx,xxxx,xxxx,xxxx
switchport mode trunk

switchport nonegotiate

Building A switch:

interface GigabitEthernet0/51
description MainBuilding - [] Gi8/4
switchport trunk encapsulation dot1q
switchport mode trunk

I attached the reports from Nagios.

[cma · Accepted solution]

We detected the problem, it was a DNS servers configuration. Thanks to all for the help