Adtran Netvanta 3448 - throughput

[Anonymous]

I have a Netvanta 3448.  We just recently upgraded to Fiber with guarantee speed of 20 M Up and Down.  When we ran the speed test with just a laptop connected straight to the isp, we are getting 20 M up and Down.  When we run a speed test with Adtran Netvanta 3448, we are getting download speed of only 10 Meg.  There is no one else on the network when we are running this test.  We have already added the command ip ffe on all ethernet interface.  Does any one else have a suggestion on what we can do to make this work?  I really need to get my download speed up to 20 Meg.  Also, we are also in the process of upgrading the 2 bonded point-to-point T1 as well to a 20 Meg point-to-point as well.  Will the Adtran Netvanta 3448 handle two 20 Meg connections?  I am running Qos on these router as well. 

[Anonymous]

touristsis:

Thank you for asking this question in the support community. 

There are several things to check when "speed testing" websites, found on the Internet, show unexpected results.  The most important step is to try a different speed testing website.  Many of these websites use different variables to determine the upload and download speeds.  Since, they are all different, often you will have drastically exaggerated download or upload speeds between different testers.  So, it is recommended to try using multiple sites.

On the ADTRAN unit, you should verify the interfaces do not have any errors, they are negotiated to the proper speed/duplex, and often the RapidRoute "Fast Forwarding Engine" (ip ffe) will need to be enabled on all interfaces.

To view the status of the interfaces to check for errors and proper speed/duplex settings, issue the show interface command.

ROUTER-NAME# show interfaces

Displaying interfaces . . .

eth 0/1 is UP, line protocol is UP

  Hardware address is 00:A0:C8:2D:57:06

  Ip address is 10.1.1.30, netmask is 255.255.255.0

  MTU is 1500 bytes,  BW is 100000 Kbit

  100Mb/s, negotiated full-duplex, configured full-duplex

  ARP type: ARPA; ARP timeout is 20 minutes

  5 minute input rate 88 bits/sec, 0 packets/sec

  5 minute output rate 64 bits/sec, 0 packets/sec

    Queueing method: fifo

    Output queue: 0/256/0 (size/max total/drops)

    Interface Shaper: NOT ENABLED

    31184 packets input, 5071001 bytes

    1127 unicasts, 15519 broadcasts, 14538 multicasts input

    361 unknown protocol, 0 symbol errors, 0 discards

    0 input errors, 0 runts, 0 giants

    0 no buffer, 0 overruns, 0 internal receive errors

    0 alignment errors, 0 crc errors

    16035 packets output, 3650729 bytes

    1476 unicasts, 20 broadcasts, 14539 multicasts output

    0 output errors, 0 deferred, 0 discards

    0 single, 0 multiple, 0 late collisions

    0 excessive collisions, 0 underruns

    0 internal transmit errors, 0 carrier sense errors

    0 resets, 0 throttles

FFE is enabled on a per-interface basis with the ip ffe command:

ROUTER-NAME(config-eth 0/1)# ip ffe

I hope that makes sense, but please do not hesitate to reply to this post with any additional questions.  I will be happy to help in any way I can.

Levi

[Anonymous]

Hi Levi,

Below is my results.  How can I fixed the errors?  By the way, the speed test runs much better at a different sites.

eth 0/1 is UP, line protocol is UP

  Description: INTERNET

  Hardware address is 00:A0:C8:78:36:3B

  Ip address is xx.xx.xx.xx, netmask is 255.255.255.248

  Secondary IP:

   xx.xx.xx.xx, 255.255.255.248

  xx.xx.xx.xx, 255.255.255.248

  MTU is 1500 bytes,  BW is 100000 Kbit

  100Mb/s, full-duplex

  ARP type: ARPA; ARP timeout is 20 minutes

  5 minute input rate 134104 bits/sec, 23 packets/sec

  5 minute output rate 39320 bits/sec, 23 packets/sec

    Queueing method

        Configured Queueing Method: fifo

        Effective  Queueing Method: weighted fair

    Output queue: 0/1/684/64/0 (size/highest/max total/threshold/drops)

      Conversations  0/1/256 (active/max active/max total)

      Available Bandwidth 75000 kilobits/sec

    Interface Shaper: NOT ENABLED

    12089436 packets input, 704821182 bytes

    12084915 unicasts, 4521 broadcasts, 0 multicasts input

    0 unknown protocol, 0 symbol errors, 0 discards

    277 input errors, 0 runts, 0 giants

    0 no buffer, 0 overruns, 0 internal receive errors

    0 alignment errors, 277 crc errors

    10074652 packets output, 2110254211 bytes

    10068424 unicasts, 184 broadcasts, 6044 multicasts output

    0 output errors, 0 deferred, 0 discards

    0 single, 0 multiple, 0 late collisions

    0 excessive collisions, 0 underruns

[Anonymous]

Looks like eth 0/1 is hard-set to 100/full duplex.  Is the carrier's device also hard-set to 100/full duplex?  As a rule of thumb, both should be set -or- both should auto-negotiate.  If you're not sure or can't verify, it's helpful to just try both settings and compare results.

Best,

--

Chris

[Anonymous]

touristsis:

I agree with .  If there is a duplex mismatch, more than likely you will see collisions and/or input errors on either side of the link, as you do in the output above.  Since, the ADTRAN is hardset to 100/Full duplex, then the other side is probably set to auto-negotiate.  Therefore, it will negotiate to 100/Half duplex.  I would either change the ADTRAN to auto-negotiate, or hardset the other side.  It doesn't matter which, they simply both have to match.

To set the ADTRAN to auto-negotiate do the following:

ROUTER-NAME(config-eth 0/1)# speed auto

Please, let me know if that does not resolve the errors.

Levi

[Anonymous]

MTU is 1500 bytes,  BW is 100000 Kbit

  100Mb/s, negotiated half-duplex, configured full-duplex

I've changed that, above is what it change to.  Is this normal?  It negotiated half-duplex?

I've change the speed to auto

How to I change full-duplex, or half-duplex to auto negotiated?

[Anonymous]

I wonder if the provider's device you're linking to is setup as 100M half-duplex. While the 3448 is auto-neg 100/half, how's the performance? Similar to how your computer performs while bypassing the 3448? Do collisions and/or CRC errors continue to increment?

[Anonymous]

eth 0/1 is UP, line protocol is UP

  Description: INTERNET

  Hardware address is 00:A0:C8:78:36:3B

  Ip address is 50.84.224.34, netmask is 255.255.255.248

  Secondary IP:

    44.44.44.44, 255.255.255.248

    44.44.44.45, 255.255.255.248

  MTU is 1500 bytes,  BW is 100000 Kbit

  100Mb/s, negotiated half-duplex, configured full-duplex

  ARP type: ARPA; ARP timeout is 20 minutes

  5 minute input rate 9672648 bits/sec, 824 packets/sec

  5 minute output rate 318096 bits/sec, 475 packets/sec

    Queueing method

        Configured Queueing Method: fifo

        Effective  Queueing Method: weighted fair

    Output queue: 0/1/684/64/0 (size/highest/max total/threshold/drops)

      Conversations  0/1/256 (active/max active/max total)

      Available Bandwidth 75000 kilobits/sec

    Interface Shaper: NOT ENABLED

    1208731 packets input, 1782092281 bytes

    1208712 unicasts, 19 broadcasts, 0 multicasts input

    0 unknown protocol, 0 symbol errors, 0 discards

    0 input errors, 0 runts, 0 giants

    0 no buffer, 0 overruns, 0 internal receive errors

    0 alignment errors, 0 crc errors

    617646 packets output, 46427995 bytes

    617602 unicasts, 1 broadcasts, 43 multicasts output

    904 output errors, 0 deferred, 0 discards

    0 single, 0 multiple, 904 late collisions

    0 excessive collisions, 0 underruns

    0 internal transmit errors, 0 carrier sense errors

    0 resets, 0 throttles

Since I've change it, looks like the errors went to from the input side to the output side.

[Anonymous]

Hopefully Levi will chime in with a silver bullet of some kind.    Auto-neg isn't 100% reliable, although it's pretty good.  Do you have any way to verify the settings in the other device?  If you want to be certain, I recommend hard-setting (or asking the provider to hard-set) 100/full duplex, and you set eth 0/1 the same.  Sorry to hear it's been misbehaving; that must be frustrating.

[Anonymous]

touristsis:

Since changing the ADTRAN to auto-negotiate resulted in the port negotiating to 100/half duplex, I think the other side of the connection is probably hardset to 100/Full duplex.  So, I would recommend configuring the ADTRAN back to 100/Full (with the command speed 100).  However, the symptoms still indicate a problem with auto-negotiation.  I would verify the speed duplex configuration with the owner of the directly connected device if you can. 

Next, I recommend you disable Link-layer Discovery Protocol (LLDP) on the ADTRAN's interface, because some devices have problems with the open standard LLDP.  To disable LLDP on the interface do the following:

ROUTER-NAME(config-eth 0/1)# no lldp send-and-receive

I would also recommend to simply replace the Ethernet cable.  A faulty cable could also result in the errors you are experiencing.

Please, let me know the result of these recommendations.

Levi

[Anonymous]

touristsis:

I have marked this post as "assumed answered," but do not hesitate to reply to this thread if you have further questions on this topic.  I will be happy to help.

Levi

[Anonymous]

I agree. When you are set to half-duplex (non-negotiated), and you see late collisions, that's usually an indication the other end is hard coded to full-duplex, and doesn't care if you are sending before he sends (If he cared, he'd listen first and not send. a late collision means he had enough time (from the begining of your outbound packet) to listen and not send while you are sending).

Attached is a doc on duplex mismatches. (I welcome improvements to it)

Keep in mind, the issue could also be a mismatch on the customer side of the 3448. Every ethernet hop matters.

Glen G.

[Anonymous]

Informative. 

[troy_parish]

Just ran across this thread, it stopped me from losing even more hair on a new Fiber cutover.

Thanks to all of you good folks!