Want to protect your cyber security and still get fast solutions? Ask a secure question today.Go Premium

x
?
Solved

slow LAN speed

Posted on 2007-07-22
21
Medium Priority
?
494 Views
Last Modified: 2010-03-18
hi,

   This morning i found that my outlook pst file stored in samba server isn't accessible. After i restart samba file server, i found that the LAN speed is really slow. Transfer a 20 MB file take about 5 min.  So i just power cycle the switches and the speed is bring up to about 1 min.  

  The path between samba file server is just two switch away. I test transfer speed on the machine directly connected to first switch of file server, the speed is all right (about 5 sec).  So there isn't anything wrong on the software side but more on the switches? i change the switches as well but it is still slow... Can anyone help me on that?


[samba file server] -> switch1 -> switch2 -> switch 3

**switch 3 actually present a group of switches. All pc behind switch group 3 are slow down alot.
** pc behind switch1&2 are about the same speed.

ps: the switch we have just some soho type 8 port home switch with no management interface.

0
Comment
Question by:stock99
  • 10
  • 7
  • 3
  • +1
21 Comments
 

Author Comment

by:stock99
ID: 19544351
" The path between samba file server is just two switch away".  I

what i mean to say is there are 3 switch in between file server and my pc.
0
 

Expert Comment

by:EDP_NIAdmin
ID: 19544468
If you've got slow LAN speed then a remote PST file (especially a large one) will have lots of problems. From your comments, if you're saying that the PC's behind Sw 1 & 2 are OK but all PC behind sw 3 are not good, then your likely problem is either : 1) faulty sw 3; 2) Faulty cabling: 3) Faulty port on sw 2 or sw 3.
If you haven't got a spare sw to confirm that it's not just sw3 that is the problem, then the first thing I'd do is take sw 3 to where sw 2 is and try & make sure it's not faulty cabling or a faulty port before I go out & buy a new sw.
0
 

Author Comment

by:stock99
ID: 19544488
after i reset pc, swith and my NFS server (on a tested linux box near me). The speed went back to normal. I would like the cause of it. Can someone give me some hints on finding the source of the problem?  Besides, we have about 4 8ports switch on 'group 3'.  Is it making any difference if i change them into single 48 ports switch (maybe cisco?)  
0
Technology Partners: 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!

 

Expert Comment

by:EDP_NIAdmin
ID: 19544494
hmmm, re-reading your issue I missed something. Looks like your prob is more at sw 2. Have you tried swapping that? as per comments above, don't rule out cable or port problem between sw1 & sw 2.
0
 

Author Comment

by:stock99
ID: 19544507
EDP_NIAdmin:

I did what you said before my problem miraculously gone away. Also i did a continuity test on the cable from group2 to group3 switch (use the cheap cable tester from ebay) and there seems to be no problem.

The problem was switch after cascade 3 time (ie group 3 switches) the speed drop. Is it normal for those soho cheap 8port switch?

The only thing i can think about to trigger this is probably the large amount of file copying going through switch in group 1 ->2 ->3. There are about 150G data copying as weekly process and both smbfs and nfs are used.
0
 

Author Comment

by:stock99
ID: 19544514
but pc behind switch 2 are fine. Also, i try to swap switch 2 to spare switch and it make no different.
0
 

Expert Comment

by:EDP_NIAdmin
ID: 19544527
good to see it's OK, but... Always hard to track down if it's an unmanaged sw network (hard if it's managed as well). The layout shouldn't be a issue as long as it's a branching net & has no loops. However the question is why? If you're doing it just to get capacity then yes, go for larger switches. If you have just a basic network then I'd simply look at larger unmanaged switches, eg 24 / 32. Bang for buck in a basic network is better that multiple small switches, especially as you are inly able to use 28 of your 32 ports currently
0
 

Expert Comment

by:EDP_NIAdmin
ID: 19544535
can you confirm your topology is : sw1 (single) < --- > sw2 (single) <--->> sw3 (group). And also confirm location of server(s) , all above sw 1??
0
 

Expert Comment

by:EDP_NIAdmin
ID: 19544576
We use a lot of soho 5 & 8 port units across a very wide LAN (country area with fibre interlinks) & have very little trouble with them. I'm not ruling them out, but you say you've swapped a sw2 & sw 3 without a change (sw 2 pc still fast) so it tends to rule them out. By any chance, are there processes/services/systems that are totally isolated to PC's at your sw 3 layer. eg anything happenning there which would cause intense network use at this layer but wouldn't communicate with the server at sw1?
0
 

Author Comment

by:stock99
ID: 19544601
server ->(switch1) ->(switch2)->(switch3 in group)

switch 1 has a few pc attached
switch 2 has one pc attached
switch 3 group are ones most pc connected to.
switch 4 for some room with network printer or maybe sharing single cat5 uplink to switch3 group

ps: i have a another file-server and NFS server sit near me, which connected after switch 4 (which then go to wall outlet to switch3 group). The 150G files are coming to the two servers near me.
0
 
LVL 27

Assisted Solution

by:Nopius
Nopius earned 498 total points
ID: 19545653
stock99, hi.
I completely agree with EDP_NIAdmin comments regarding topology (that should not be a problem unless you have loops) and about single 48-port switch instead of many 5 and 8 ports (it's better, but probably requires new cabling).

What about a reason of your problem it may be soho 5 firmware, even switches managed by firmware, that causes them to slowdown. Also you say this problem occurs after large 150Gb transfer, I guess that your switches might be overheated and this causes some instability. Try to touch your switch just after 150Gb transfer. A had the same problem with D-Link 8 port switch, it have no fans and it was standing below other such switch.
0
 
LVL 27

Expert Comment

by:pseudocyber
ID: 19546482
You could have had a duplex mismatch.  This typically occurs when one connection is set to full duplex and the other side is autonegotiate.  The Full side turns off collision detection.  The auto side tries to negotiate with the full side, but the full side ignores the autonegotiation.  The Auto side assumes the full side is old and slow and DOWNGRADES the connection to HALF DUPLEX in order to "match" the old and slow side.  One side becomes half duplex, the other side is auto.  What results is collisions, giants, runts, CRC errors, Alignment errors.  All these errors result in TCP retransmits and the effect is that you get about 10% throughput and it usually shows up on large file transfers.

To combat duplex mismatch - make SURE both sides of an ethernet link are configured the same.  Either both auto, or both full duplex.  I have seen many instances when a server was NOT operating at the duplex the server admin had configured it for.  Configured does not always equal operating.  Keep an eye on logs, ethernet error statistics, strange behavior.

Hope this helps.
0
 
LVL 27

Expert Comment

by:pseudocyber
ID: 19546488
Sorry, meant to say, ... "One side becomes half duplex, the other side is full"
0
 

Author Comment

by:stock99
ID: 19546821
hmm... it is not managed switch.... how do i ensure the duplex thing?

or do you mean the NIC on each linux box?

By the way, i notice the speed is really slow when i taring file over NFS. Usually when i tar over smbfs to a windows box without going through the switch3 group, the 100G finish in 8 hours. If i passing through the switch3 group , my speed is low as 3G per hour..

lastly, i am not going to use VLAN and i don't think there is a looping environment. Any tips on configuring cisco switch? I probably pick 2950 series are they are cheaper.
0
 
LVL 27

Assisted Solution

by:pseudocyber
pseudocyber earned 498 total points
ID: 19546841
If you're using an unmanaged switch, just make sure the PC/Server is autonegotiate.
0
 

Author Comment

by:stock99
ID: 19546988
i run a mii-tool -v for both server. They seems to be both auto-negotiate..

--- debian server (taring to ubuntu)

---debian side-----


eth1: negotiated 100baseTx-FD flow-control, link ok
  product info: Intel 82555 rev 4
  basic mode:   autonegotiation enabled
  basic status: autonegotiation complete, link ok
  capabilities: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD
  advertising:  100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD flow-control
  link partner: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD flow-control



-----------ubuntu (receiving tar from debian server)--------
eth1: negotiated 100baseTx-FD, link ok
  product info: vendor 00:00:00, model 0 rev 0
  basic mode:   autonegotiation enabled
  basic status: autonegotiation complete, link ok
  capabilities: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD
  advertising:  100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD
  link partner: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD flow-control

0
 

Expert Comment

by:EDP_NIAdmin
ID: 19552262
Not sure this issue is link layer as whilst this could effect the throughput between the two systems doing the transfer, arp tables in the switches should have this issue isolated to the path only. This is effecting all PC at the sw3 layer.

Based on a couple of previous comments, can you confirm that your sw1 & sw2 switches  are same as your sw3 group. These two comments strike me as interesting :

"but pc behind switch 2 are fine. Also, i try to swap switch 2 to spare switch and it make no different. "

"By the way, i notice the speed is really slow when i taring file over NFS. Usually when i tar over smbfs to a windows box without going through the switch3 group, the 100G finish in 8 hours. If i passing through the switch3 group , my speed is low as 3G per hour.."

One thing I'd try is connecting your NFS box to sw2, just out of interest.

0
 

Author Comment

by:stock99
ID: 19552392
[One thing I'd try is connecting your NFS box to sw2, just out of interest.]

i would need to get a 30 meter long cable....will probably try that later. But my nfs box is on the same switch  with a tested rhel box. The speed has no problem between the nfs and rhel test box.

I just did the transfer last night. 12 hours it only tar about 70G. I tried the `dd` too and it is about the same speed.

Will the old switch with (uplink and downlink port) mix with auto detect new switch mix together cause any issue? we have a few such in the network.
0
 

Author Comment

by:stock99
ID: 19552408
Based on a couple of previous comments, can you confirm that your sw1 & sw2 switches  are same as your sw3 group. These two comments strike me as interesting :

switch 1 and 2 are different brand of switch, if that is what you asking. Switch 1 is wise and 2 is netgear. There is a netgear switch among the swtich 3 group.
0
 

Accepted Solution

by:
EDP_NIAdmin earned 504 total points
ID: 19552550
>>Will the old switch with (uplink and downlink port) mix with auto detect new switch mix together cause any issue? we have a few such in the network.

Should be fine, this is basically to do with crossover or straight cables & the devices at the ends (DTE/DCE). New switches autodetect DTE/DCE & change accordingly. Older switches don't & therefore need the physical ports (or manual switch) or crossover cables as required. If you get LINK all is OK.

As to the switch types. I'm just wondering if one of your brands is actually dodgy. Try making wise & netgear only brands in the large transfer path & see how that goes.

0
 

Author Comment

by:stock99
ID: 19766692
i have replace the 4 oldest switches with a cisco 3548 and trying to change to most other soho switches to be netgear and wise only.  If i encounter the same problem again , i will open another topic and probably refer back to this thread.

Thanks for all the suggestions & advices.  I equally divided the points.

0

Featured Post

Who's Defending Your Organization from Threats?

Protecting against advanced threats requires an IT dream team – a well-oiled machine of people and solutions working together to defend your organization. Download our resource kit today to learn more about the tools you need to build you IT Dream Team!

Question has a verified solution.

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

Every server (virtual or physical) needs a console: and the console can be provided through hardware directly connected, software for remote connections, local connections, through a KVM, etc. This document explains the different types of consol…
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 …
If you're a developer or IT admin, you’re probably tasked with managing multiple websites, servers, applications, and levels of security on a daily basis. While this can be extremely time consuming, it can also be frustrating when systems aren't wor…
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…
Suggested Courses
Course of the Month10 days, 14 hours left to enroll

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