Hello,
Thank you in advance to those who can help me.
First, in order not to waste your time, let me say that I am not administrator of the system (well, not all), I'm just looking for ideas to refer the person occurs (with admin rights) on the network.
The architecture is as follows:
Server 1 (files), Server 2 (Citrix) (runs on W. server 2003)
|
2 x 1 GB
|
Cisco 2950
|
Etherchannel 400 Mbps
|
Cisco 2960
|
1 GB
|
Cisco 2960G (I am the administrator of this one, and therefore have access to logs of traffic)
|
1 GB
|
Client (with Giga NIC)
I use a very basic app (LAN speedtest) to test the flow. I know it's a very rough test, but it is enough, given the results I will discuss later:
From client to client connected directly to the 2960G: about 400 Mbps Read / Write
From a client connected to the 2960G to client connected to 2960 while a file is tranfered (from another client) to the server: approximately 150 Mbps in the connection between the 2960G and 2960 -> OK because > 100 Mbps .
From a client connected to the 2960G to client connected to 2950 while a file is tranfered (from another client) to the server: approximately 150 Mbps in the connection between the Giga 2960G and 2960 -> OK because Etherchannel > 100 Mbps.
Test write simultaneously on the server from a client connected to 2950 and a client connected to the 2960G: 2x70 Mbps = 140 Mbps> Link 2950 to server OK because > 100 Mbps ...
So far, nothing unusual except that when I do a write test on the server from a client at the chain end (see diagram) that is connected to the 2960G, the flow is only 70 Mbps (100 Mbps theoretical, I think). Since the connections between switches are correct, the output should be limited by the connection with the lowest bandwidth (ie etherchannel 400 Mbps, 300 Mbps really, I would say).
So my questions ;-):
The etherchannel was performed using "Cisco Network Assistant. Is there a trunk that has to be done after that or is it automatically done when you create the etherchannel?
Is it possible that the manipulation was incomplete and that the switch considers the EtherChannel as one port and that, therefore, does not make them a link to the server has up to 100 Mbps (like all other ports)?
The 2960G has remained on the default VLAN (1) instead of the VLAN (20). The reason is that the manipulation would be more tedious to perform (paradoxically) to make a 2960 with a 2950. So, the trunk with the 2960 and the 2960G is not performed. I could in this case, understand that the flow is limited or nonexistent. but I doubt that will affect just the connection to the server ... What do you think?
From what I understand, Smartports are macros that can be used for connections to switches switches. Are they equivalent to a change of VLAN + a trunk?
Thank you for reading ... If after your aspirin, you still have the courage to answer me, I'd be delighted!
See you soon.
Dj.
