Point-to-Point T1 Speed Issue

[RMCS]

I am an IT consultant with limited Cisco knowledge. I have a client who has an existing P2P T1 that seems to be really slooow!

Can anyone give me some guidance on how to troubleshoot the issue?

Basically the config is:

Site #1 (192.168.0.x)
|
|
Cisco 2500 Series (IOS 12.0(27))
|
| (Serial Connection)
Adtran Device
|
|
P2P T1 Bridge (Not routed)
|
|
Adtran Device
| (Serial Connection)
|
Cisco 2500 Series (IOS 12.0(27))
|
|
Site #2

I spoke with the phone vendor and 8 of the channels are dedicated to the phone system with the remaining 16 channels available for data. That should give a 1024 link - right? This seems consistent with the 950kbps throughput. However, during the day the throughput drops dramatically.

The P2P link is configured as a bridge and so both sites are using the 192.168.0.x subnet. Is this the best way to configure this link?

Would it be better to setup IP routing between sites instead?

If so, what routing protocol should I use?

Thanks for the help!

 

[BuckWeet]

You could potentially be utilizing quite a bit of the bandwidth from broadcasts.. It would be best to setup a routed network.

As for a routing protocol, I would just use static routes if thats your full network.


BuckWeet

 

[vipergg]

Sounds like they may be using QOS to also run the phone system across this if so this will limit the thruput if you get a lot of users on that line as a T1 is not much bandwidth , most home users have at leastr that much at home now . Though I am not sure how much QOS you can do on a old 2500 series router as those were in use long before VOIP became prevalent . They may be doing the QOS stuff on the data switches themselves depending on what they have . 1 person doing a download is going to pretty much chew up that T1 along with any voice on the line . I would look at the configs on the routers and the switches if the switches are manageable.

 

[wabob]

If that Adtran device is a mux (like the TSU600), it gives 64 Kbps by default for each DS0 channel, 24 total for a full T1. If you have a V.35 serial module in the mux for WAN connectivity, you should console in to the mux and make sure all available unallocated channel bandwidth is logically mapped to that serial module. This does not happen by default, it must be done manually.

You can visit Adtran's website to obtain more information, or you can call them. They seem to be very helpful, even if you do not have an agreement with them.

"We know now we can't beat their machines. We've got to beat THEM."
Dr. Clayton Forrester

 

[RMCS]

Thanks for all of your input! My thought was the same as BuckWeet: that broadcasts are eating away at the bandwidth. My thought was to create a routed connection and put the two sites on different subnets.

I can get into both 2500 routers and there is NO QoS being performed, nor are their switches capable of QoS. That would be a possible upgrade, but I think separating the sites into two subnets would help a lot.

As for the Adtran being a mux device, I believe that is the case. Thanks for the advise of calling them for help. I definately will. I don't think that the channels are being allocated properly. (I don't think that site #2 needs 8 channels dedicated to voice)

Is there any way to determine what the "link" speed is between the two 2500 routers? Obviously I "think" it is about 1024, but I am only assuming based upon some throughput tests from one workstation to another.

 

[BuckWeet]

you should be able to do a 'show interface s0' and it should show theclock rate in bps that is being offered by the DCE (being the adran)


BuckWeet

 

[plshlpme]

with a serial connection between the adtran and the router i wouldn't think the clock rate would be meaningful on the router.. since there is no csu in the 2500 it would default to 1536 kbps unless it has been manually set with the bandwidth statement.

i think you will need to log into the adtran and check the provisioning of that channel group.

be careful before messing with your voice alocation though..
ide get in touch with the voice guys before making changes there

 

[BuckWeet]

bandwidth statement has nothing to do with the serial clock speed. the bandwidth statement is just for routing protocol metrics..

the router would see the clock rate (aka baud rate) on the serial interface because it has to know what speed teh serial link is operating at.. since the adtran is the DCE, it would receive the serial speed information from it.


BuckWeet

 

[RMCS]

Here is what I get when I "show interface s0":

Serial0 is up, line protocol is up
Hardware is HD64570
Description: College to Mulberry Serial Interface
MTU 1500 bytes, BW 1544 Kbit, DLY 20000 usec, rely 255/255, load 1/255
Encapsulation HDLC, loopback not set, keepalive set (10 sec)
Last input 00:00:00, output 00:00:00, output hang never
Last clearing of "show interface" counters never
Input queue: 0/75/57/0 (size/max/drops/flushes); Total output drops: 56987
Queueing strategy: weighted fair
Output queue: 0/1000/64/56981 (size/max total/threshold/drops)
Conversations 0/23/256 (active/max active/max total)
Reserved Conversations 0/0 (allocated/max allocated)
5 minute input rate 3000 bits/sec, 4 packets/sec
5 minute output rate 3000 bits/sec, 3 packets/sec
233503282 packets input, 320575594 bytes, 55 no buffer
Received 33229979 broadcasts, 0 runts, 0 giants, 0 throttles
3829 input errors, 3781 CRC, 936 frame, 24 overrun, 24 ignored, 586 abort
236150495 packets output, 570340166 bytes, 0 underruns
0 output errors, 0 collisions, 2 interface resets
0 output buffer failures, 0 output buffers swapped out
2 carrier transitions
DCD=up DSR=up DTR=up RTS=up CTS=up

It appears as if it is showing a 1544 Kbit link, which can't be correct!

 

[wabob]

That is the default bandwidth setting on a serial interface. It doesn't actually control the speed, it is used by routing protocols like EIGRP and OSPF to calculate a metric for that link.

I still think you need to verify how much bandwidth is mapped to the serial interface module on the Adtran units.

"Never in the field of human confict was so much, owed by so many, to so few..."
Sir Winston Churchill

 

[robertjo24]

I would assume from this discussion that the Adtran device is a T1 mux, with no voice actually going over the network?
No IP phones, correct?

If the Adtran is a T1 mux, the voice traffic should have no efect on the network speed.

I would agree with previous posts that the two locations need to be set up on different LANs to break up the broadcast domains.
Also, what traffic is passed between locations during normal hours?
If the daytime traffic were large file transfers, or chatty applications (like Citrix), there would be a noticable congestion as users got busy.

 

[RMCS]

In response to robertjo24:
The Adtran device is a T1 mux, so no voice traffic is on the network (no IP phones!).

I will definately breakup the two locations into two separate broadcast domains. The normal daytime traffic is periodic file transfer to multiple servers and heavy Terminal Server use, but no Citrix.

 

[wabob]

My thought is that the serial module in the Adtran mux does not have all the available DS0 channels mapped to it, so during periods of high traffic, it could be causing a "bottleneck" across the WAN.

"Never in the field of human confict was so much, owed by so many, to so few..."
Sir Winston Churchill

 

[RMCS]

Thanks for all of the help. I have reconfigured the T1 for routing instead of bridging. However, I am still having some speed issues:

Internet access seems to be much faster, but now people are complaining that their print jobs (that travel over the T1) take a lot longer than they did before the reconfiguration!

Could someone please look at the config files and see if there are any glaring issues? Thanks!

Site 1: Main DNS/DHCP server is 192.168.10.10
Site 2: Secondary DNS/DHCP server is 192.168.0.10

Site #1 Config File
!
service timestamps debug uptime
service timestamps log uptime
service password-encryption
no service tcp-small-servers
no service udp-small-servers
!
hostname CORP-RTR
!
enable password 1234
!
no ip name-server
!
ip subnet-zero
no ip domain-lookup
ip routing
!
interface Ethernet 0
no shutdown
description connected to 192.168.10.x
ip address 192.168.10.254 255.255.255.0
no keepalive
!
interface Serial 0
no shutdown
description connected to MFG-RTR
ip unnumbered Ethernet 0
encapsulation hdlc
!
interface Serial 1
no description
no ip address
shutdown
!
router rip
version 2
network 192.168.10.0
network 192.168.0.0
no auto-summary
!
!
ip classless
ip route 0.0.0.0 0.0.0.0 192.168.10.1
no ip http server
snmp-server community public RO
snmp-server location College Ave
!
line console 0
exec-timeout 0 0
password 1234
login
!
line vty 0 4
password 1234
login
!

Site #2 Config File
!
service timestamps debug uptime
service timestamps log uptime
service password-encryption
no service tcp-small-servers
no service udp-small-servers
!
hostname MFG-RTR
!
enable password 1234
!
no ip name-server
!
ip subnet-zero
no ip domain-lookup
ip routing
!
interface Ethernet 0
no shutdown
description connected to 192.168.0.x
ip address 192.168.0.1 255.255.255.0
no keepalive
full-duplex
!
interface Serial 0
no shutdown
description connected to CORP-RTR
ip unnumbered Ethernet 0
encapsulation hdlc
!
interface Serial 1
no description
no ip address
shutdown
!
router rip
version 2
network 192.168.0.0
network 192.168.10.0
no auto-summary
!
!
ip classless
ip route 0.0.0.0 0.0.0.0 192.168.10.1
no ip http server
snmp-server community public RO
snmp-server location Mulberry Ave
!
line console 0
exec-timeout 0 0
password 1234
login
!
line vty 0 4
password 1234
login
!

 

[burtsbees]

You said that this speed issue happens during the day, but otherwise you have 950kbps throughput? I would perhaps think the broadcasts in one big broadcast domain would be an issue, but why would they stop at the end of the day?
Also, I would throw a sniffer on the line to see what is eating the bandwidth. If you have good throughput at night but not during the day, then chances are it is not a configuration issue---more like someone running Limewire or something...just a thought.

Burt

 

[burtsbees]

Oh man---sorry---different problem...printer, huh? Well, how are the printers set up? What kind? Some printers allow you to configure the pc whom is doing the printing to rip the printjob rather than the printer itself. What part of the print jobs are slow? Have you done an extended ping from a pc to the printer?

Burt