Advanced
Mailing List Archive

Mailing List Archive

  1. Home >
  2. nsp>
  3. foundry
MLXe slow on 10G doing inter-vlan routing
sraymond at acedatacenter
May 24, 2016, 1:42 PM
Post #1 of 7 (2652 views)
Permalink
MLXe, very lightly loaded, with only one card BR-MLX-10Gx8-X. Simplest of configuration, layer 2 link with VE interfaces. During inter-vlan routing where the ingress traffic and egress traffic are on the same port, say e1/2, the input traffic level will be fine, but the output traffic is effectively maxed out at about 50Mbps (on a 10GigE port). Shift hosts around so that the egress is a different port on the same card, bang problem gone and full expected traffic rates achieved.

Tried numerous code versions, from 05800a to 05900a to 06000. No improvement.

Of course tried rebooting, new optics, new cabling. No errors incrementing on the interface. This is the silliest dumb problem in a long time. Any ideas are appreciated.

Thanks!



_______________________________________________
foundry-nsp mailing list
foundry-nsp@puck.nether.net
http://puck.nether.net/mailman/listinfo/foundry-nsp
Re: MLXe slow on 10G doing inter-vlan routing [ In reply to ]
georgeb at gmail
May 24, 2016, 1:59 PM
Post #2 of 7 (2646 views)
Permalink
Do you notice the same behavior with both UDP and TCP streams?



On Tue, May 24, 2016 at 1:42 PM, Steven Raymond <sraymond@acedatacenter.com>
wrote:

> MLXe, very lightly loaded, with only one card BR-MLX-10Gx8-X. Simplest of
> configuration, layer 2 link with VE interfaces. During inter-vlan routing
> where the ingress traffic and egress traffic are on the same port, say
> e1/2, the input traffic level will be fine, but the output traffic is
> effectively maxed out at about 50Mbps (on a 10GigE port). Shift hosts
> around so that the egress is a different port on the same card, bang
> problem gone and full expected traffic rates achieved.
>
> Tried numerous code versions, from 05800a to 05900a to 06000. No
> improvement.
>
> Of course tried rebooting, new optics, new cabling. No errors
> incrementing on the interface. This is the silliest dumb problem in a long
> time. Any ideas are appreciated.
>
> Thanks!
>
>
>
> _______________________________________________
> foundry-nsp mailing list
> foundry-nsp@puck.nether.net
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>
Re: MLXe slow on 10G doing inter-vlan routing [ In reply to ]
mkallen at gmail
May 24, 2016, 3:07 PM
Post #3 of 7 (2640 views)
Permalink
Turn off ICMP redirects. Common problem in one arm routing configs. IMO it
should be the default setting but it's not.

Mike
On May 24, 2016 2:00 PM, "George B" <georgeb@gmail.com> wrote:

> Do you notice the same behavior with both UDP and TCP streams?
>
>
>
> On Tue, May 24, 2016 at 1:42 PM, Steven Raymond <
> sraymond@acedatacenter.com> wrote:
>
>> MLXe, very lightly loaded, with only one card BR-MLX-10Gx8-X. Simplest
>> of configuration, layer 2 link with VE interfaces. During inter-vlan
>> routing where the ingress traffic and egress traffic are on the same port,
>> say e1/2, the input traffic level will be fine, but the output traffic is
>> effectively maxed out at about 50Mbps (on a 10GigE port). Shift hosts
>> around so that the egress is a different port on the same card, bang
>> problem gone and full expected traffic rates achieved.
>>
>> Tried numerous code versions, from 05800a to 05900a to 06000. No
>> improvement.
>>
>> Of course tried rebooting, new optics, new cabling. No errors
>> incrementing on the interface. This is the silliest dumb problem in a long
>> time. Any ideas are appreciated.
>>
>> Thanks!
>>
>>
>>
>> _______________________________________________
>> foundry-nsp mailing list
>> foundry-nsp@puck.nether.net
>> http://puck.nether.net/mailman/listinfo/foundry-nsp
>>
>
>
> _______________________________________________
> foundry-nsp mailing list
> foundry-nsp@puck.nether.net
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>
Re: MLXe slow on 10G doing inter-vlan routing [ In reply to ]
foundry-nsp at puck
May 25, 2016, 2:06 AM
Post #4 of 7 (2642 views)
Permalink
i second that,

no ip icmp redirects - in global config mode or
no ip redirect - in interface config mode

and you are good to go.

Tudor
--------------------------------------------
On Wed, 5/25/16, Mike Allen <mkallen@gmail.com> wrote:

Subject: Re: [f-nsp] MLXe slow on 10G doing inter-vlan routing
To: "George B." <georgeb@gmail.com>
Cc: foundry-nsp@puck.nether.net
Date: Wednesday, May 25, 2016, 1:07 AM

Turn off
ICMP redirects. Common problem in one arm routing configs.
IMO it should be the default setting but it's not.
Mike
On May 24, 2016 2:00 PM,
"George B" <georgeb@gmail.com>
wrote:
Do you notice the same behavior with both UDP and
TCP streams?


On Tue, May 24, 2016 at
1:42 PM, Steven Raymond <sraymond@acedatacenter.com>
wrote:
MLXe,
very lightly loaded, with only one card BR-MLX-10Gx8-X. 
Simplest of configuration, layer 2 link with VE
interfaces.  During inter-vlan routing where the ingress
traffic and egress traffic are on the same port, say e1/2,
the input traffic level will be fine, but the output traffic
is effectively maxed out at about 50Mbps (on a 10GigE
port).  Shift hosts around so that the egress is a
different port on the same card, bang problem gone and full
expected traffic rates achieved.



Tried numerous code versions, from 05800a to 05900a to
06000.  No improvement.



Of course tried rebooting, new optics, new cabling.  No
errors incrementing on the interface.  This is the silliest
dumb problem in a long time.  Any ideas are appreciated.



Thanks!







_______________________________________________

foundry-nsp mailing list

foundry-nsp@puck.nether.net

http://puck.nether.net/mailman/listinfo/foundry-nsp




_______________________________________________

foundry-nsp mailing list

foundry-nsp@puck.nether.net

http://puck.nether.net/mailman/listinfo/foundry-nsp

-----Inline Attachment Follows-----

_______________________________________________
foundry-nsp mailing list
foundry-nsp@puck.nether.net
http://puck.nether.net/mailman/listinfo/foundry-nsp
_______________________________________________
foundry-nsp mailing list
foundry-nsp@puck.nether.net
http://puck.nether.net/mailman/listinfo/foundry-nsp
Re: MLXe slow on 10G doing inter-vlan routing [ In reply to ]
youssef at 720
May 25, 2016, 2:50 AM
Post #5 of 7 (2641 views)
Permalink
Hello,

Could you please explain the correlation between this commands and
performance issues observed by the OP ?

Best regards.



2016-05-25 11:06 GMT+02:00 Tudor Capatina via foundry-nsp <
foundry-nsp@puck.nether.net>:

> i second that,
>
> no ip icmp redirects - in global config mode or
> no ip redirect - in interface config mode
>
> and you are good to go.
>
> Tudor
> --------------------------------------------
> On Wed, 5/25/16, Mike Allen <mkallen@gmail.com> wrote:
>
> Subject: Re: [f-nsp] MLXe slow on 10G doing inter-vlan routing
> To: "George B." <georgeb@gmail.com>
> Cc: foundry-nsp@puck.nether.net
> Date: Wednesday, May 25, 2016, 1:07 AM
>
> Turn off
> ICMP redirects. Common problem in one arm routing configs.
> IMO it should be the default setting but it's not.
> Mike
> On May 24, 2016 2:00 PM,
> "George B" <georgeb@gmail.com>
> wrote:
> Do you notice the same behavior with both UDP and
> TCP streams?
>
>
> On Tue, May 24, 2016 at
> 1:42 PM, Steven Raymond <sraymond@acedatacenter.com>
> wrote:
> MLXe,
> very lightly loaded, with only one card BR-MLX-10Gx8-X.
> Simplest of configuration, layer 2 link with VE
> interfaces. During inter-vlan routing where the ingress
> traffic and egress traffic are on the same port, say e1/2,
> the input traffic level will be fine, but the output traffic
> is effectively maxed out at about 50Mbps (on a 10GigE
> port). Shift hosts around so that the egress is a
> different port on the same card, bang problem gone and full
> expected traffic rates achieved.
>
>
>
> Tried numerous code versions, from 05800a to 05900a to
> 06000. No improvement.
>
>
>
> Of course tried rebooting, new optics, new cabling. No
> errors incrementing on the interface. This is the silliest
> dumb problem in a long time. Any ideas are appreciated.
>
>
>
> Thanks!
>
>
>
>
>
>
>
> _______________________________________________
>
> foundry-nsp mailing list
>
> foundry-nsp@puck.nether.net
>
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>
>
>
>
> _______________________________________________
>
> foundry-nsp mailing list
>
> foundry-nsp@puck.nether.net
>
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>
> -----Inline Attachment Follows-----
>
> _______________________________________________
> foundry-nsp mailing list
> foundry-nsp@puck.nether.net
> http://puck.nether.net/mailman/listinfo/foundry-nsp
> _______________________________________________
> foundry-nsp mailing list
> foundry-nsp@puck.nether.net
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>



--
Youssef BENGELLOUN-ZAHR
Re: MLXe slow on 10G doing inter-vlan routing [ In reply to ]
foundry-nsp at puck
May 25, 2016, 3:24 AM
Post #6 of 7 (2641 views)
Permalink
Hello,

You can have a nice read here:

https://puck.nether.net/pipermail/foundry-nsp/2006-December/005382.html

and just for reference the documentation for the commands:

http://www.brocade.com/content/html/en/configuration-guide/NI_05800a_SWITCHING/GUID-04EDFD31-E5FB-4593-8434-0DF8EDB3249E.html

Have a nice day!
--------------------------------------------
On Wed, 5/25/16, Youssef Bengelloun-Zahr <youssef@720.fr> wrote:

Subject: Re: [f-nsp] MLXe slow on 10G doing inter-vlan routing
To: "Tudor Capatina" <ktudore@yahoo.com>
Cc: "foundry-nsp@puck.nether.net" <foundry-nsp@puck.nether.net>
Date: Wednesday, May 25, 2016, 12:50 PM

Hello,

Could you please explain the correlation
between this commands and performance issues observed by the
OP ?

Best regards.



2016-05-25 11:06 GMT+02:00
Tudor Capatina via foundry-nsp <foundry-nsp@puck.nether.net>:
i second
that,



no ip icmp redirects - in global config mode or

no ip redirect - in interface config mode



 and you are good to go.



Tudor

--------------------------------------------

On Wed, 5/25/16, Mike Allen <mkallen@gmail.com>
wrote:



 Subject: Re: [f-nsp] MLXe slow on 10G doing inter-vlan
routing

 To: "George B." <georgeb@gmail.com>

 Cc: foundry-nsp@puck.nether.net

 Date: Wednesday, May 25, 2016, 1:07 AM



 Turn off

 ICMP redirects. Common problem in one arm routing
configs.

 IMO it should be the default setting but it's not.

 Mike

 On May 24, 2016 2:00 PM,

 "George B" <georgeb@gmail.com>

 wrote:

 Do you notice the same behavior with both UDP and

 TCP streams?





 On Tue, May 24, 2016 at

 1:42 PM, Steven Raymond <sraymond@acedatacenter.com>

 wrote:

 MLXe,

 very lightly loaded, with only one card
BR-MLX-10Gx8-X. 

 Simplest of configuration, layer 2 link with VE

 interfaces.  During inter-vlan routing where the
ingress

 traffic and egress traffic are on the same port, say
e1/2,

 the input traffic level will be fine, but the output
traffic

 is effectively maxed out at about 50Mbps (on a 10GigE

 port).  Shift hosts around so that the egress is a

 different port on the same card, bang problem gone and
full

 expected traffic rates achieved.







 Tried numerous code versions, from 05800a to 05900a to

 06000.  No improvement.







 Of course tried rebooting, new optics, new cabling. 
No

 errors incrementing on the interface.  This is the
silliest

 dumb problem in a long time.  Any ideas are
appreciated.







 Thanks!















 _______________________________________________



 foundry-nsp mailing list



 foundry-nsp@puck.nether.net



 http://puck.nether.net/mailman/listinfo/foundry-nsp









 _______________________________________________



 foundry-nsp mailing list



 foundry-nsp@puck.nether.net



 http://puck.nether.net/mailman/listinfo/foundry-nsp



 -----Inline Attachment Follows-----



 _______________________________________________

 foundry-nsp mailing list

 foundry-nsp@puck.nether.net

 http://puck.nether.net/mailman/listinfo/foundry-nsp

_______________________________________________

foundry-nsp mailing list

foundry-nsp@puck.nether.net

http://puck.nether.net/mailman/listinfo/foundry-nsp


--
Youssef
BENGELLOUN-ZAHR

_______________________________________________
foundry-nsp mailing list
foundry-nsp@puck.nether.net
http://puck.nether.net/mailman/listinfo/foundry-nsp
Re: MLXe slow on 10G doing inter-vlan routing [ In reply to ]
youssef at 720
May 25, 2016, 6:15 AM
Post #7 of 7 (2640 views)
Permalink
Hello,

Indeed, makes a lot more sens now. Thanks for the links.

Best regards.



2016-05-25 12:24 GMT+02:00 Tudor Capatina <ktudore@yahoo.com>:

> Hello,
>
> You can have a nice read here:
>
> https://puck.nether.net/pipermail/foundry-nsp/2006-December/005382.html
>
> and just for reference the documentation for the commands:
>
>
> http://www.brocade.com/content/html/en/configuration-guide/NI_05800a_SWITCHING/GUID-04EDFD31-E5FB-4593-8434-0DF8EDB3249E.html
>
> Have a nice day!
> --------------------------------------------
> On Wed, 5/25/16, Youssef Bengelloun-Zahr <youssef@720.fr> wrote:
>
> Subject: Re: [f-nsp] MLXe slow on 10G doing inter-vlan routing
> To: "Tudor Capatina" <ktudore@yahoo.com>
> Cc: "foundry-nsp@puck.nether.net" <foundry-nsp@puck.nether.net>
> Date: Wednesday, May 25, 2016, 12:50 PM
>
> Hello,
>
> Could you please explain the correlation
> between this commands and performance issues observed by the
> OP ?
>
> Best regards.
>
>
>
> 2016-05-25 11:06 GMT+02:00
> Tudor Capatina via foundry-nsp <foundry-nsp@puck.nether.net>:
> i second
> that,
>
>
>
> no ip icmp redirects - in global config mode or
>
> no ip redirect - in interface config mode
>
>
>
> and you are good to go.
>
>
>
> Tudor
>
> --------------------------------------------
>
> On Wed, 5/25/16, Mike Allen <mkallen@gmail.com>
> wrote:
>
>
>
> Subject: Re: [f-nsp] MLXe slow on 10G doing inter-vlan
> routing
>
> To: "George B." <georgeb@gmail.com>
>
> Cc: foundry-nsp@puck.nether.net
>
> Date: Wednesday, May 25, 2016, 1:07 AM
>
>
>
> Turn off
>
> ICMP redirects. Common problem in one arm routing
> configs.
>
> IMO it should be the default setting but it's not.
>
> Mike
>
> On May 24, 2016 2:00 PM,
>
> "George B" <georgeb@gmail.com>
>
> wrote:
>
> Do you notice the same behavior with both UDP and
>
> TCP streams?
>
>
>
>
>
> On Tue, May 24, 2016 at
>
> 1:42 PM, Steven Raymond <sraymond@acedatacenter.com>
>
> wrote:
>
> MLXe,
>
> very lightly loaded, with only one card
> BR-MLX-10Gx8-X.
>
> Simplest of configuration, layer 2 link with VE
>
> interfaces. During inter-vlan routing where the
> ingress
>
> traffic and egress traffic are on the same port, say
> e1/2,
>
> the input traffic level will be fine, but the output
> traffic
>
> is effectively maxed out at about 50Mbps (on a 10GigE
>
> port). Shift hosts around so that the egress is a
>
> different port on the same card, bang problem gone and
> full
>
> expected traffic rates achieved.
>
>
>
>
>
>
>
> Tried numerous code versions, from 05800a to 05900a to
>
> 06000. No improvement.
>
>
>
>
>
>
>
> Of course tried rebooting, new optics, new cabling.
> No
>
> errors incrementing on the interface. This is the
> silliest
>
> dumb problem in a long time. Any ideas are
> appreciated.
>
>
>
>
>
>
>
> Thanks!
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> _______________________________________________
>
>
>
> foundry-nsp mailing list
>
>
>
> foundry-nsp@puck.nether.net
>
>
>
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>
>
>
>
>
>
>
>
>
> _______________________________________________
>
>
>
> foundry-nsp mailing list
>
>
>
> foundry-nsp@puck.nether.net
>
>
>
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>
>
>
> -----Inline Attachment Follows-----
>
>
>
> _______________________________________________
>
> foundry-nsp mailing list
>
> foundry-nsp@puck.nether.net
>
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>
> _______________________________________________
>
> foundry-nsp mailing list
>
> foundry-nsp@puck.nether.net
>
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>
>
> --
> Youssef
> BENGELLOUN-ZAHR
>
>


--
Youssef BENGELLOUN-ZAHR

©2024 GT.net. A Gossamer Threads company.
5th Floor, 455 Granville St., Vancouver, BC V6C 1T1, Canada | Legal