[Last Call] Learn how to a build a cloud-first strategyRegister Now

x
?
Solved

Network Performance Testing...

Posted on 2004-03-30
4
Medium Priority
?
153 Views
Last Modified: 2010-03-18
Situation

HP ML370 Server (1000Mbps NIC, 3.06MHz Xeon Processor)
HP ProCurve 2650 (Switch With 2 GBps Interfaces and 48 100Mbps)
HP Evo D330 on a 100Mbps slot (CPU 2.8 multitrading PIV)
Cabling CAT5e

Transfering a 500MB file from Server to Evo

Problem:
Switch Stats show the transfer only takes 40% af the cap. of the 100Mbps port and a 1% load on the 1000Mbps port.
So I thought the server or the warkstations is being overloaded. Server CPU 2%, Server 1Gbps NIC 4%. RAID 5 HDD load: 4%
OK so not the server. Only one left: WKS. Evo CPU load= 4%, NIC interface load 40%.

Question
This behaviour is not normal is it, The switch schould report or allow port loads up to 99% no?
What should I look for in pinpointing the bottleneck?
0
Comment
Question by:ezjurgen
3 Comments
 
LVL 7

Expert Comment

by:Isigow
ID: 10713503
Couple things I would check

1) have you tried different NIC cable to the workstation/server?
2) have you verified that there are no collisions on the ports of the switch that are involved (the most common cause of network slowdown)
3) have you tried using a ping grapher between the two machines to see if there is a suddent jump that is slowing you down.
4) is the switch being saturated with other traffic during the tests? (the more the switch has to do, the more it will cause other things to slow down)
5) have you tried a bandwidth grapher? there are freeware graphers out there that you connect 2 machines together and run a series of tests over like 30 minutes that in turn can create a more accurate test

Isi
0
 
LVL 1

Author Comment

by:ezjurgen
ID: 10814402
OK guys found the problem

It seems HP procurve switches have a neat feature anabled by default called LACP, a dynamic trunking manager. This thing blocks ports and decreases performance on non-trunked connections.

I called HP wo informed me of this (LACP is widely used oin the US and that's the place they design the ROM configs) and showed me what it does (eventlog trough telnet) and disabled it.

Since this point on I get speeds up to 90% of the Procurves 100Mbps ports Yihaaaa!!


thanks anyway guys

asking refund
0
 

Accepted Solution

by:
CetusMOD earned 0 total points
ID: 11171184
PAQed, with points refunded (125)

CetusMOD
Community Support Moderator
0

Featured Post

Transaction-level recovery for Oracle database

Veeam Explore for Oracle delivers low RTOs and RPOs with agentless transaction log backup and transaction-level recovery of Oracle databases. You can restore the database to a precise point in time, even to a specific transaction.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

This article is in response to a question (http://www.experts-exchange.com/Networking/Network_Management/Network_Analysis/Q_28230497.html) here at Experts Exchange. The Original Poster (OP) requires a utility that will accept a list of IP addresses …
I'm a big fan of Windows' offline folder caching and have used it on my laptops for over a decade.  One thing I don't like about it, however, is how difficult Microsoft has made it for the cache to be moved out of the Windows folder.  Here's how to …
Michael from AdRem Software explains how to view the most utilized and worst performing nodes in your network, by accessing the Top Charts view in NetCrunch network monitor (https://www.adremsoft.com/). Top Charts is a view in which you can set seve…
Is your data getting by on basic protection measures? In today’s climate of debilitating malware and ransomware—like WannaCry—that may not be enough. You need to establish more than basics, like a recovery plan that protects both data and endpoints.…

829 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question