Ok. I have 2 PC's that at one location called site A are running very slow but when I bring them to site b they speed up considerably. Site A and B are connected via Gig Fiber. I also changed the connection type to wireless but it made no diffrence. I replaced the switch at site a. Still no change. Disabled every other device on the switch at site A except one PC and the fiber connection to site B. No change. I am running out of things to replace. No spyware or virues on PC's either. Any suggestions would be welcomed.

Ok. I have 2 PC's that at one location called site A are running very slow but when I bring them to site b they speed up considerably. Site A and B are connected via Gig Fiber. I also changed the connection type to wireless but it made no diffrence. I replaced the switch at site a. Still no change. Disabled every other device on the switch at site A except one pc and the fiber connection to site B. No change. I am running out of things to replace. No spyware or virues on PC's either. Any suggestions would be welcomed.
You're segmenting with a switch? Is it a layer 3 switch or a router?
Are they are on the same network?
Same Subnetmask?
What are the ips of router/switch a and b
Copy and paste ipconfig/all from one machine on the A network and one on the B network.

Thank!

They are on the Same network, IP of Client is 10.31.8.5 Mask 255.255.255.0 Gate 10.31.8.1. Server is 10.31.4.5 Mask 255.255.255.0 THe only router is our core router at the servers location. Took the client PC to another location and it works fine from there. Tried 2 different routes, one fiber one wireless but both have same result. Replaced switch at client location still no change. Bypassed swith altogether at client location, no change.

They are on the Same network, IP of Client is 10.31.8.5 Mask 255.255.255.0 Gate 10.31.8.1. Server is 10.31.4.5 Mask 255.255.255.0 THe only router is our core router at the servers location. Took the client pc to another location and it works fine from there. Tried 2 different routes, one fiber one wireless but both have same result. Replaced switch at client location still no change. Bypassed swith altogether at client location, no change.

When you login to the switch and do a tracert, and ping, what are the results?

Pings & tracert are good and with quick responses from both the switch and PC.

What is "running slow"

Running applications from the server. One in particular opens in 10 sec at every other location but at that site it takes 3 min. Browsing and opening files on multiple servers is not as fast as at other locations.

Copy and paste the tracert to me from the 3minute machine to the Server

I will have to get out to that location so it will take some time. But every tracert I ran went like this. 2 hops. First 10.31.1.1 ( our core router) Second was the server. Responses were 5-8 ms.

I will have to get out to that location so it will take some time. But every tracert I ran went like this. 2 hops. First 10.31.1.1 ( our core router) Second was the server. Responses were 5-8 ms.

It never hit the switch?

Passes through the switch. Switched network not routed network.

They are on the Same network, IP of Client is 10.31.8.5 Mask 255.255.255.0 Gate 10.31.8.1. Server is 10.31.4.5 Mask 255.255.255.0 THe only router is our core router at the servers location. Took the client pc to another location and it works fine from there. Tried 2 different routes, one fiber one wireless but both have same result. Replaced switch at client location still no change. Bypassed swith altogether at client location, no change.

I have a few questions for you as well. You said that the locations are connected via fiber. How far apart are they? Is the fiber multi-mode or single mode? If it is multi-mode, then it will support GB connections however 50/125 fiber is preferred over 62.5/125 for longer distances (2000 meters vs. 700 meters). Is it owned buy you or is it leased?

Where are your routers (10.31.8.1 and 10.31.4.5) located at? You stated that they are on the same network but they are not. Both networks are Class C's so they have different network ID's.

The Client is 10.31.4.0 and the server is 10.31.8.0.

As to the switches, the previous posts have some good questions for you. What type are they? Are they managed switches? Are you using VLAN's?

Passes thru the switch. Switched network not routed network.

Every network I have ever Configured that was segmented had a switch or a router
http://www.newcomcs.com/34.jpg

I have a few questions for you as well. You said that the locations are connected via fiber. How far apart are they? Is the fiber multi-mode or single mode? If it is multi-mode, then it will support GB connections however 50/125 fiber is preferred over 62.5/125 for longer distances (2000 meters vs. 700 meters). Is it owned buy you or is it leased?

Where are your routers (10.31.8.1 and 10.31.4.5) located at? You stated that they are on the same network but they are not. Both networks are Class A's so they have different network ID's.

The Client is 10.31.4.0 and the server is 10.31.8.0.

As to the switches, the previous posts have some good questions for you. What type are they? Are they managed switches? Are you using VLAN's?
:) but using a Class C subnet

Every network I have ever Configured that was segmented had a switch or a router

[INTERNET]
Site A | Site B
CORE
PC 1 PC 2 Server 1 ROUTER PC 1 PC 2 SERVER1 SERVER2
8.5 8.4 8.10 / | 4.5 4.6 4.10 4.11
| | | / | | | | |
------------------------ / \ ---------------------------------
| / \ |
SWITCH A ------------/ \ SWITCH B
10.1.8.1 \------10.1.4.1


Exactly!
Either you are routing via a router or through VLAN's in your switch. You have configured your server and your client on two separate network ID's. Your "Core" router would have to be aware of both networks with a route table and it would have an IP address visible on both networks.

:) but using a Class C subnet

Yes, typo on my part. I am so used to seeing the 10. Networks as a Class A.

Hehe :)

Can you capture and post a Tracert from your client PC to your server?

Have you checked the Event Logs on your switches for errors? Can you check the Port statistics to see if there are any drops or errors? Is Spanning Tree Protocol enabled in your switches and do you see any STP errors?

Can you capture and post a Tracert from your client PC to your server?

Have you checked the Event Logs on your switches for errors? Can you check the Port statistics to see if there are any drops or errors? Is Spanning Tree Protocol enabled in your switches and do you see any STP errors?


I asked him to do that a while back but he said he'd have to go onsite so I am guessing he's doing so... or maybe on another service call?

Let me try to explain by using your picture. Site A PC1(10.31.8.69) Switch 10.31.8.11
From there fiber to Core Switch/Router Which has Route tables
And IP 10.31.8.1, 10.31.7.1 etc.
Site B has switch 10.31.1.11 and server 10.31.4.38

Let me try to explain by using your picture. Site A PC1(10.31.8.69) Switch 10.31.8.11
from there fiber to Core Switch/Router Which has Route tables
and IP 10.31.8.1, 10.31.7.1 etc.
Site B has switch 10.31.1.11 and server 10.31.4.38
I am confused now, can you draw it out?

Also, still no answer as to type of fiber and what is the distance? And the switch event logs?

I have the distinct pleasure of having some old single mode fiber that does not support GB connections and it is PAINFULLY slow or does not work (depending on length) if you try it. And even if it is Multi-mode, what spec is it and what is the distance?

Fiber is new. Single mode. Working on a diagram but I am having some trouble. I will post when I get it done.

Also, what are the models of the Switchs/ router This may help us out a lot.

Fiber is new. Single mode. Working on a diagram but I am having some trouble. I will post when I get it done.
Good deal :)

Fiber is new. Single mode. Working on a diagram but I am having some trouble. I will post when I get it done.

Check with your vendor on your Single Mode fiber as I have not seen one yet that supports GB. You will need to degrade your switch connections down to 100MB.

That is why Multi-mode costs more.

Check with your vendor on your Single Mode fiber as I have not seen one yet that supports GB. You will need to degrade your switch connections down to 100MB.

That is why Multi-mode costs more.
Hey Chuck, running 1gb over single mode fiber would show up in the ping, wouldn't it?

Core is a HP 9300. Site A and B are HP 2848.

* Four 10/100/1000 ports or mini-GBIC slots for optional fiber connectivity such as Gigabit-SX, -LX, or -LH

Now the question would be is is Single Mode or Multi?
And get that diagram for me :)

Single mode from Site A to core (about 2 miles or so). Multimode from Core to Site B (located in same Room).

Increasing the cache size of the ping using the /l switch would provide some meaningful information. A 32byte default ping size is not that big.

I just did some updating of my Fiber Optics knowledge. It looks like the newer Single Mode fibers are the way to go.

good deal, hey look at your PM chuck

Did. Back at you.

Hey, JKF, I am out of here in 15 but will be back on Sunday PM or Monday AM. Good luck buddy and look forward to your updates. This is a good one. :confused: ;)

JKF,
How are you coming along on your diagram?
Do you have Visio?

Having trouble getting it sized correctly to post.

I have attached a .zip file that should have a jpg image of what I am trying to explain.

Do you have a link or something to the zip?

Attached .pdf. Did not look like the zip file went.

I see a backup connection to the Internet. How is that connected. Do you have two ISP connections? Do you have two routers? Is this a single ISP and router with a connection to each switch?

If this is the same ISP and router, then do you have Spanning Tree Protocol enabled on your switches? Did you assign a cost metric to each leg with the backup connection receiving the higher cost metric?

Have you logged on to your switches and looked at the event logs? Be sure and go to the last page for the latest information. Any STP errors?

No backup connection to ISP. It is backup to our core switch. Normally Wireless connection is left powered down and olny used (powered up) if the fiber connection goes down. So we did not set up spanning tree. From our core we link to firewall and then ISP. Event logs on switches don't show any errors, collisions, nothing that looks out of the norm. No high packet counts on any specific ports.

To take routing out of the picture, try changing a machine in Site A to a Class B subnet mask and a machine in Site B to a Class B subnet mask. This will put them both on the same Network ID. Then ping the machine in Site A from the machine in Site B and vice-versa. Change your Ping to use the /L switch as follows:

Note: Before changing the subnet masks, run the following pings to get a baseline performance indicator. Then make the changes.

Ping /l 64 PC_IN_SITE_A
Ping /l 640 PC_IN_SITE_A
Ping /l 6400 PC_IN_SITE_A

Repeat in the opposite direction. You should see about a 10% increase in time from 64 to 640. Then about double the time from 640 to 6400.

Note your findings. Did you see any changes in speed with these two machines. If so, there is a problem with your route tables or router configuration.

If it is not a route issue, then my next question is when the fibers were laid, were they certified by the installer for transmission quality? Do you have the correct modules in the switches for the fiber types? The modules for Multi-mode fiber are not strong enough to push light through the much smaller diameter Single-mode fiber.

No change in speed. Site A to B replys were <1ms. Site B to A <10ms. No changes when Subnet mask was changed or packet size was increased. Neither way would reply to 6400 size. Fiber test good per vendor but like I stated previously even when we bypass the fiber and go with our wireless backup link the slowness still exists.

When you replaced the fiber with wireless, this was on the short link. There would be no way for you to replace the fiber on the long 2 mile link unless you have a microwave system. This has to be where the problem lies. There are two fibers that handle a single data connection. One is for transmit and one is for receive.

When you ping from A to B, your time is <1ms. When you ping from B to A, your time is <10ms. This means that the Transmit fiber from B to A is your likely culprit.

A way to verify: Have someone stationed at both ends. Remove your Fibers from the Transceiver and swap the connections. Then rerun your Ping tests. The results should be exactly opposite.

This would verify that one of your fibers is degraded in quality and there is significant light loss. The ends are almost always the culprit. Replace the connectors.

As far as the 6400 size goes. This is approaching the size of a Jumbo Frame at 9000 bytes. Your network gear should be able to handle Jumbo Frames but Flow Control must be turned OFF on most switches. There are only a few Cisco switches that can handle Jumbo Frames AND Flow Control. Your HP switches can not even though their IOS allows it to be turned on when Jumbos are enabled.

Bump.

I will try switching the fiber pairs but I do not think that is the problem because of the same issue when we switch to the wireless link instead of using the fiber.

You diagram shows that you are running MM fiber from A to your switch/router and SM fiber from B to your switch/router. And since The sites are interconnected and routed via two fiber connections, it does not matter if you replace the MM with a wireless connection, you still have the other 2 mile SM connection to contend with. This is where I believe the problem lies. Two miles is ceartainly within the limits of the fiber capabilities but at that distance, a suspect quality fiber will degrade your performance substantially. I have had it happen to me at only 2,000 feet.

The fiber connections run in pairs at each end of the connection. One end transmits and the other receives. The other fiber in the pair runs in the other direction. By switching the pairs at both ends, you should see the ping performance also change directions if one of the two fibers is degraded in quality. Often, the fibers are installed in multipair trunks. If you have additional fibers run to your facillity, you can swap the line with another fiber to see if performance changes. If not, you will need to have an end to end tester check the line quality.

To Clarify: Site A -> Core 2 miles SM Fiber OR Wireless. Site B-> Core MM Fiber about 10 Feet ( same Room).

OK. This going to be a real needle in the haystack for you but here is a great utility.

Wireshark is a protocol analyzer that you can install onto a laptop and plug straignt into your switch. In the switch configuration, setup the port that you plug it into as a Monitoring Port.

Wireshark: Go deep. (http://www.wireshark.org/)

It is a free utility with excellent support on the web site. Look for patterns. See if a device is storming the network with broadcasts. You can select which network adapter to listen on if you want to use it to troubleshoot wired or wireless networks.

Try to chenge the split speed... try to dive in 6:4... then check what happen`s