Slow Network

Posted on 2006-06-01
Medium Priority
Last Modified: 2013-12-07
I am on a Network with a Windows 2003 Domain controller, and Novell 6 Server.  Ever since I have been the administratror the Network has been slower than it should be.  I have been monitoring errors on my HP ProCurve Switch.  I do get a lot of error which i keep track of.  I have replaced all the computers that get multiple errrors everyday.  Now I still have about 15 computers that get a error every day or two but the network still does not seem to be real fast.  I have started to look at some of the other Port Counters they have to monitor.  The other Port Counters I can monitor are "MCast RX", MCast TX, BCast RX, BCast TX, PKTS RX, PKTS TX, & Error RX.  Which one should I be paying the closest attention to, to figure out which devices on my network are sending out more data accross the network than they should?  I'm not sure if there is any easy answer for this but I do notice when I look at some of the stats some ports have a lot more of one particular counter than the others do.  

Any help on suggestion on this would be greatly appreciated.  Or if there is some software or network device that I can hook up or install to tell me which computers are transmitting more data than the others that will work too.

Thanks for any help I can get.

Question by:shermeta
  • 3
  • 2
  • 2
  • +2
LVL 30

Expert Comment

ID: 16809650
How are the NIC's and switches set in terms of duplex?

On the NW 6 server go into to monitor
LAN/WAN and look at statistics under the NIC
Look for things like CRC errors ,runts,collisions and such.
On a switched network there should be NO collisions.
LVL 30

Expert Comment

ID: 16809678
If things look clean,you might have a worm or trojan that is infecting your systems.
You'll need to load something like Ethereal and sniff the network for issues.

Expert Comment

ID: 16809737
if your novell server is configured for IPX/SPX running both IPX/SPX and IP produces lots of broadcast packets, and most managed switches dont monitor IPX/SPX, which i believe the Procurve series dont. try enabling Novell to use IP only.

Also in regards to network devices, there is software to measure bandwith but it is a speed test only, it doesn't monitor packets in a live environment, you need a packet sniffer/protocol analyser to monitor data on your live environment you can use the likes of http://www.ethereal.com/ or you can invest in a hardware protocol analyser like http://www.netoptics.com/ thats the best way to monitor traffic on your network to see where slowdown is introduced, i personally use ethereal

Modern healthcare requires a modern cloud. View this brief video to understand how the Concerto Cloud for Healthcare can help your organization.


Expert Comment

ID: 16809746
beat me on the ethereal suggestion :)

Author Comment

ID: 16810530
Most of the NIC's if not all are set to Auto duplex mode.  I noitice on my switch I get a lot of port (like 15) that produce an error of "High Collisions or drop rates".  All 15 of these seem to happen at the exact same time.  Not the same time everyday but the same time as the other 14 ports get the error.  

I just looked on my Novell Server and I do have it configured with both IP and IPX, but the workstation I configure with IP only.  I have lots of errors.  Under collisions I get in Monitor on the NetWare Server.  Below I listed them.
Transmit Succeeded - Single Collision - over 22,000
Transmit Succeeded - Multiple Collision - over 13,000
Transmit Succeeded - Deferred - Over 38,000
Transmit Failed - Excessive Collisions - 129

I have played with the speed and duplex setting on the NetWare box and when i make any change I get huge problems on the network where everyone complains and can barely work.  Currently the Novell box is set to 100HD on both sides.  Maybe the whole problem stems from this Server.  Any suggestion on how to get rid of the errors.  The Server has been up for only 16 days.  It has been on longer than that but it was last rebooted 16 days ago so i'm pretty sure the Server resets the error counts on a reboot, so I guess this is a lot of errors in just 16 days.

Any suggestion how to clean up my NetWare box would be great.  This is a Compaq Proliant Box running NetWare 5.1 with server pak 5.1.7 and Compaq's Service pak 7.1.



Accepted Solution

gcaley earned 1200 total points
ID: 16812146
If you want to remove errors. you should set the server, switch ports and clients to be hard coded at 100 full duplex. Definately having your Novell server at half duplex is not a good idea. Auto-negotiation fails regularly, especially with different vendor components.

Also, if none of your clients or other devices use IPX then disable it on the Novell server, it's an old protocol that adds nothing but traffic to the network.

As someone mentioned, use ethereal to find out where the bulk of the traffic is coming from, it is well possible (I've seen many instances) that you do have either a trojan (SQL servers are often the culprit), or just have a NETBIOS browser storm. Ethereal should pinpoint the culprit as you'll see lots of traffic from that host

When looking at the number of errors, the more interesting stat is the percentage or errors compared to those that aren't (if looking at packets etc)
LVL 30

Assisted Solution

pgm554 earned 800 total points
ID: 16812264
How is the switch configured?
Is it set to manual or auto?
This will dictate as how you need to set the card in the NW box.

It works like this:
Switch set to auto> cards set to auto
Switch set to manual 100 full >cards set to manual 100 full

When running half duplex it doesn't matter,but the switch starts to act like a hub and you start to get collisions.

You CANNOT mix and match when running full duplex or auto.

It's all one way or the other.

You can get away with setting the cards to half and the switch can be auto,full or half with no major ill effects(you just lose some of the speed and efficiency of the switch).

Unless you are running IPX on the WS, IPX isn't needed,but some backup software likes to see IPX loaded on the server even though it isn't being used for transport.)IPXRTR needs to be loaded.

Author Comment

ID: 16817427
Ok I thank everyone so far for all your help.  I have bumped up the point value on this so once my problem is all the way corrected or at least I am pointed in the general direction I can allocate point to everyone that helped me.

Now, Ethereal is loaded.  Wow, Sure is a lot of Data.  Any tips on what I should be looking for would be greatly appreciated.  No clue on how I should use this at all.  All i really want is for it to tell me what IP address is sending the most information out on the Network so I can issue a new computer and hopefully speed up my network.

LVL 14

Expert Comment

ID: 16822770
get rid of ICS on the WinXP pro box
get a private/public type router like these to keep your internal network separate from your guests
set it as the main router, like:  internet - priv/pub rtr (dhcp srvr)
  then connect your switches to it
  then connect your other wireless routers (using the wireless connection - turn off their dhcp service thus making them access points)
LVL 14

Expert Comment

ID: 16822772
sorry wrong thead

Featured Post

Independent Software Vendors: We Want Your Opinion

We value your feedback.

Take our survey and automatically be enter to win anyone of the following:
Yeti Cooler, Amazon eGift Card, and Movie eGift Card!

Question has a verified solution.

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

This article will show how Aten was able to supply easy management and control for Artear's video walls and wide range display configurations of their newsroom.
LinkedIn blogging is great for networking, building up an audience, and expanding your influence as well. However, if you want to achieve these results, you need to work really hard to make your post worth liking and sharing. Here are 4 tips that ca…
Get a first impression of how PRTG looks and learn how it works.   This video is a short introduction to PRTG, as an initial overview or as a quick start for new PRTG users.
Monitoring a network: why having a policy is the best policy? Michael Kulchisky, MCSE, MCSA, MCP, VTSP, VSP, CCSP outlines the enormous benefits of having a policy-based approach when monitoring medium and large networks. Software utilized in this v…

571 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