[j-nsp] Cant load share in Agregated sonet
Nilesh Khambal
nkhambal at juniper.net
Wed Feb 18 23:12:11 EST 2009
Activate "indexed-next-hop" and try this as hash key
hash-key {
family mpls {
label-1;
payload {
ip {
port-data;
}
}
}
}
Thanks,
Nilesh.
As ad Arafat wrote:
> Hello Nilesh,
>
> Thank you very much for your suggestion.
> I had try It out but the bandwidth same as before :(
>
> load-balance {
> no-next-hop-cloning;
> inactive: equal-weighted-mode;
> inactive: indexed-next-hop;
> }
> hash-key {
> family inet {
> layer-3;
> layer-4;
> }
> family mpls {
> label-1;
> label-2;
> payload {
> ip;
> }
> }
> }
>
> Logical interface as1.0 (Index 72) (SNMP ifIndex 133) (Generation 248)
> Flags: Point-To-Point SNMP-Traps 0x4000 Encapsulation: PPP
> Statistics Packets pps Bytes bps
> Bundle:
> Input : 1574238316 27467 549195500508 70996288
> Output: 1585499563 27551 822963970038 121087264
> Link:
> so-0/1/1.0
> Input : 1542528979 26723 531046561900 67867536
> Output: 593617855 9115 130919023851 15468328
> so-0/1/0.0
> Input : 31709337 744 18148938608 3128752
> Output: 991881708 18436 692044946187 105618936
>
> On Thu, Feb 19, 2009 at 10:24 AM, Nilesh Khambal <nkhambal at juniper.net> wrote:
>> try, these things in the same order one at time and see if one of them makes
>> any difference.
>>
>> 1. Remove "equal-weighted-mode". Check load-balancing
>>
>> 2. Remove "indexed-next-hop" and add "no-next-hop-cloning" Check
>> load-balancing.
>>
>> load-balance {
>> no-next-hop-cloning;
>> }
>>
>>
>> Thanks,
>> Nilesh.
>>
>> As ad Arafat wrote:
>>> Nilesh,
>>>
>>> Sorry to bothering you again, as JTAC suggestion, i had apply the
>>> forwarding-option configuration at the both side of PE:
>>>
>>> load-balance {
>>> equal-weighted-mode;
>>> indexed-next-hop;
>>> }
>>> hash-key {
>>> family inet {
>>> layer-3;
>>> layer-4;
>>> }
>>> family mpls {
>>> label-1;
>>> label-2;
>>> payload {
>>> ip;
>>> }
>>> }
>>> }
>>>
>>> But it still cant load-share in the bandwith, :(
>>> PE1.batam "sh int as1 detail"
>>>
>>> Logical interface as1.0 (Index 72) (SNMP ifIndex 133) (Generation 248)
>>> Flags: Point-To-Point SNMP-Traps 0x4000 Encapsulation: PPP
>>> Statistics Packets pps Bytes bps
>>> Bundle:
>>> Input : 1469518591 26072 515087563009 66790808
>>> Output: 1479039012 25262 764318666407 106322328
>>> Link:
>>> so-0/1/1.0
>>> Input : 1440546432 25294 498347148297 63414280
>>> Output: 560001305 8109 123610734671 13742872
>>> so-0/1/0.0
>>> Input : 28972159 778 16740414712 3376528
>>> Output: 919037707 17153 640707931736 92579456
>>>
>>> With this this bandwidth it seem 2 x stm-1 underutilized and the
>>> traffic over this interface experiencing heavy congestion.
>>> Do you have any suggestion to workaround this?
>>> If we remove the sonet links from aggregated sonet and created two
>>> independent sonet interface do we have 2 x stm-1 bandwidth?
>>>
>>> Thanks before
>>>
>>> Best Regads
>>>
>>>
>>> On Wed, Feb 18, 2009 at 2:34 PM, As ad Arafat <asad.arafat at gmail.com>
>>> wrote:
>>>> hehehe :D
>>>>
>>>> Thanks Nilesh
>>>>
>>>> On Wed, Feb 18, 2009 at 2:18 PM, Nilesh Khambal <nkhambal at juniper.net>
>>>> wrote:
>>>>> I see that you have a JTAC case opened. So I will shut up now :) and let
>>>>> JTAC drive it forward.
>>>>>
>>>>> Just as a note, load balancing in JUNOS, largely depends on what kind of
>>>>> traffic you are trying to load balace. I am sure JTAC will ask you about
>>>>> it
>>>>> during the progress of the case. Unfortunately, there is no one solution
>>>>> that fits all the traffic patterns. It needs to be fine tuned as per
>>>>> your
>>>>> your requirement.
>>>>>
>>>>> Thanks,
>>>>> Nilesh.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> On Feb 17, 2009, at 9:49 PM, "As ad Arafat" <asad.arafat at gmail.com>
>>>>> wrote:
>>>>>
>>>>>> Nilesh,
>>>>>>
>>>>>> I already add "family inet" as hash-key on both PE, but it still cant
>>>>>> achieve load sharing in as1
>>>>>>
>>>>>> Best Regards
>>>>>> As'ad
>>>>>>
>>>>>>
>>>>>> On Wed, Feb 18, 2009 at 12:26 PM, Nilesh Khambal <nkhambal at juniper.net>
>>>>>> wrote:
>>>>>>> Is it the traffic coming in from "pacnet" vrf that you are trying to
>>>>>>> load
>>>>>>> balance when going over as1 towards MPLS cloud? If so, I think you
>>>>>>> should
>>>>>>> configure "family inet" hashing as well on the PE.
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Nilesh
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> As ad Arafat wrote:
>>>>>>>> Hi Nilesh,
>>>>>>>>
>>>>>>>> I attached "show route a.b.c.d extensive" and "show route
>>>>>>>> forwarding-table destination a.b.c.d extensive"
>>>>>>>> i use hash like this.
>>>>>>>>
>>>>>>>> hash-key {
>>>>>>>> family mpls {
>>>>>>>> label-1;
>>>>>>>> payload {
>>>>>>>> ip;
>>>>>>>> }
>>>>>>>> }
>>>>>>>> }
>>>>>>>>
>>>>>>>> On Wed, Feb 18, 2009 at 12:03 PM, Nilesh Khambal
>>>>>>>> <nkhambal at juniper.net>
>>>>>>>> wrote:
>>>>>>>>> Can you please send the output of below 2 commands for the
>>>>>>>>> destination
>>>>>>>>> that
>>>>>>>>> you want to load balance?
>>>>>>>>>
>>>>>>>>> "show route a.b.c.d extensive"
>>>>>>>>> "show route forwarding-table destination a.b.c.d extensive"
>>>>>>>>>
>>>>>>>>> What kind of hashing have you configured under "edit
>>>>>>>>> forwarding-options"?
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Thanks,
>>>>>>>>> Nilesh.
>>>>>>>>>
>>>>>>>>> As ad Arafat wrote:
>>>>>>>>>> Hi list,
>>>>>>>>>>
>>>>>>>>>> I got m7i and m10i linked aggregated sonet interface with 2 sonet
>>>>>>>>>> interface inside it.
>>>>>>>>>> I already add load-balancing per-packet policy and
>>>>>>>>>> forwarding-option
>>>>>>>>>> cinfiguration
>>>>>>>>>> But one of sonet interface seem underutilized.
>>>>>>>>>>
>>>>>>>>>> Any workaround for this issue?
>>>>>>>>>>
>>>>>>>>>> Best Regads
>>>>>>>>>>
>>>>>>>>>> Asad
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> ------------------------------------------------------------------------
>>>>>>>>>>
>>>>>>>>>> _______________________________________________
>>>>>>>>>> juniper-nsp mailing list juniper-nsp at puck.nether.net
>>>>>>>>>> https://puck.nether.net/mailman/listinfo/juniper-nsp
>>>>>> <PE1.Equinix6>
>>>>>> <PE1.Batam6>
More information about the juniper-nsp
mailing list