[j-nsp] Load Balance in VRF by Junos

Addy Mathur addy.mathur at gmail.com
Tue Apr 6 09:14:15 EDT 2010


Gabriel:

What kind of flows and variations exist in your traffic?  Regardless
of the loadbalance configuration, without enough variance, traffic
distribution will be polarized.

Regards,
Addy.


On 4/5/10, Gabriel Farias <gabrielfarias03 at gmail.com> wrote:
> Hi,
>
> I configured the router the suggested parameters in URL:
>
> {master}[edit]
>
> RT110# show policy-options policy-statement load-balance
>
> term 1 {
>
>     then {
>
>         load-balance per-packet;
>
>     }
>
> }
>
>
>
> RT110# show routing-options
>
> forwarding-table {
>
>     export load-balance;
>
> }
>
> RT110# show forwarding-options
>
> load-balance {
>
>     indexed-next-hop;
>
> }
>
> hash-key {
>
>     family inet {
>
>         layer-3;
>
>         layer-4;
>
>     }
>
> }
>
> The continuing imbalance in the physical interfaces:
>
> {master}[edit]
>
> RT110# run show interfaces ge-3/0/3 | match rate
>
>   Input rate: 520010728 bps (98183 pps)
>
>   Output rate: 585910320 bps (176371 pps)
>
>
>
> {master}[edit]
>
> RT110# run show interfaces ge-3/0/0 | match rate
>
>   Input rate: 621317752 bps (107767 pps)
>
>   Output rate: 107517264 bps (17319 pps)
>
>
> The configuration is correct? Or we would have to configure within the
> routing-instances dados.
>
>
> Thanks
>
> Gabriel Farias
>
>
>
> 2010/4/5 Serge Vautour <sergevautour at yahoo.ca>
>
>> Hello,
>>
>> eBGP shouldn't come into play here. Your physical interfaces are bundled
>> as
>> 1 logical interface (ae0). Load balancing over L3 LAG interfaces works the
>> same as ECMP links. You have to configure this:
>>
>>
>> http://www.juniper.net/techpubs/en_US/junos10.0/information-products/topic-collections/config-guide-policy/policy-configuring-per-packet-load-balancing.html
>>
>> Stefan has suggested this below. Have you tried it?. Can you post your
>> relevant config?
>>
>> What I find weird is you do have some egress traffic on both physical
>> links. It's just not even close to being balanced. Without any load
>> balancing configs, I would expect to see 0 bps on one link. Do you have
>> enough flows? If most of your traffic is generated from just a few flows,
>> you may not be able to get much better.
>>
>> Serge
>>
>>
>> ----- Original Message ----
>> From: Gabriel Farias <gabrielfarias03 at gmail.com>
>> To: mail-list <mohan.nanduri at gmail.com>
>> Cc: juniper-nsp at puck.nether.net; uniper-nsp-bounces at puck.nether.net
>> Sent: Mon, April 5, 2010 1:20:33 PM
>> Subject: Re: [j-nsp] Load Balance in VRF by Junos
>>
>> Stefan,
>>
>> This configuration has no effect, because the connection (PE x CE) uses
>> EBGP
>> routing within the VRF dados.
>>
>> What I need is to balance the outbound traffic of physical interfaces of
>> RT110, you have any other suggestions?
>>
>> Thanks
>> Gabriel Farias
>>
>>
>>
>> 2010/4/5 mail-list <mohan.nanduri at gmail.com>
>>
>> > try this and see if it makes any difference...
>> >
>> > set forwarding-options hash-key family mpls label-1
>> >  set forwarding-options hash-key family mpls payload ip
>> > or
>> > set forwarding-options hash-key family mpls label-1
>> >  set forwarding-options hash-key family mpls label-2
>> > set forwarding-options hash-key family mpls payload ip
>> >
>> >
>> >
>> >
>> >
>> > On Mon, Apr 5, 2010 at 8:35 AM, Gabriel Farias <
>> gabrielfarias03 at gmail.com>wrote:
>> >
>> >> Stefan,
>> >>
>> >> Any suggestions to correct this imbalance?
>> >>
>> >> Thanks,
>> >> Gabriel Farias
>> >>
>> >> 2010/4/1 Gabriel Farias <gabrielfarias03 at gmail.com>
>> >>
>> >> > Sorry for my bad translation, follows the configuration:
>> >> >
>> >> > *Interfaces*:
>> >> > show configuration interfaces ge-3/0/0
>> >> > description "*BA* G2/47 - SW001";
>> >> > gigether-options {
>> >> >     802.3ad ae0;
>> >> > }
>> >> > show configuration interfaces ge-3/0/3
>> >> > description "*BA* G1/47 - SW001";
>> >> > gigether-options {
>> >> >     802.3ad ae0;
>> >> > }
>> >> > show configuration interfaces ae0
>> >> > description "*BA* Po150 - SW001";
>> >> > mtu 9192;
>> >> > unit 0 {
>> >> >     family inet {
>> >> >         address 10.251.40.53/30;
>> >> >     }
>> >> > }
>> >> >
>> >> > *VRF instances*:
>> >> > show configuration routing-instances dados
>> >> > instance-type vrf;
>> >> > interface ae0.0;
>> >> > ...
>> >> > protocols {
>> >> >     bgp {
>> >> >         group PEER-SW001 {
>> >> >             type external;
>> >> >             hold-time 30;
>> >> >             peer-as 65100;
>> >> >             as-override;
>> >> >             neighbor 10.251.40.54;
>> >> >         }
>> >> >     }
>> >> >
>> >> > What I need is to balance the outbound traffic of physical
>> >> > interfaces,
>> >> see
>> >> > that is constant unbalance
>> >> >
>> >> > {master}
>> >> > RT110> show interfaces ge-3/0/0 | match rate
>> >> >   Input rate     : 271677184 bps (69055 pps)
>> >> >   Output rate    : 37802544 bps (10429 pps)
>> >> >
>> >> > {master}
>> >> > RT110> show interfaces ge-3/0/3 | match rate
>> >> >   Input rate     : 293302512 bps (61095 pps)
>> >> >   Output rate    : 628471504 bps (134276 pps)
>> >> >
>> >> > {master}
>> >> > RT110> show interfaces ge-3/0/0 | match rate
>> >> >   Input rate     : 298597864 bps (70456 pps)
>> >> >   Output rate    : 34856992 bps (8756 pps)
>> >> >
>> >> > {master}
>> >> > RT110> show interfaces ge-3/0/3 | match rate
>> >> >   Input rate     : 288075200 bps (57232 pps)
>> >> >   Output rate    : 657782056 bps (129224 pps)
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > 2010/3/31 Stefan Fouant <sfouant at shortestpathfirst.net>
>> >> >
>> >> >> I'm sorry, your request is getting lost in translation. Why don't
>> >> >> you
>> >> just
>> >> >> show us your configs and do a 'show route x.x.x.x' and 'show route
>> >> >> forwarding-table destination x.x.x.x' for the route you are trying
>> >> >> to
>> >> >> load-balance.
>> >> >>
>> >> >> Thanks,
>> >> >>
>> >> >>
>> >> >> Stefan Fouant
>> >> >>
>> >> >> Sent from my Verizon Wireless BlackBerry
>> >> >> ------------------------------
>> >> >> *From: * Gabriel Farias <gabrielfarias03 at gmail.com>
>> >> >> *Date: *Wed, 31 Mar 2010 18:51:04 -0300
>> >> >> *To: *<sfouant at shortestpathfirst.net>
>> >> >> *Cc: *<juniper-nsp-bounces at puck.nether.net>; <
>> >> juniper-nsp at puck.nether.net
>> >> >> >
>> >> >> *Subject: *Re: [j-nsp] Load Balance in VRF by Junos
>> >> >>
>> >> >> I'm not using EBGP external type in PE (M120)
>> >> >>
>> >> >> Reagards,
>> >> >> Gabriel Farias
>> >> >>
>> >> >> 2010/3/31 Stefan Fouant <sfouant at shortestpathfirst.net>
>> >> >>
>> >> >>> Do you have multi-path enabled on your EBGP sessions?
>> >> >>>
>> >> >>> Stefan Fouant
>> >> >>>
>> >> >>> Sent from my Verizon Wireless BlackBerry
>> >> >>> ------------------------------
>> >> >>> *From: * Gabriel Farias <gabrielfarias03 at gmail.com>
>> >> >>> *Date: *Wed, 31 Mar 2010 18:32:18 -0300
>> >> >>> *To: *<sfouant at shortestpathfirst.net>
>> >> >>> *Cc: *<juniper-nsp-bounces at puck.nether.net>; <
>> >> >>> juniper-nsp at puck.nether.net>
>> >> >>> *Subject: *Re: [j-nsp] Load Balance in VRF by Junos
>> >> >>>
>> >> >>> I have a section between EBGP VRF PE x CE via two physical
>> connections
>> >> >>> (aggregate ethernet) and see the traffic unbalanced because the
>> Junos
>> >> uses
>> >> >>> the default flow.
>> >> >>>
>> >> >>> I need the traffic between the PE x CE is symmetrically balanced
>> way,
>> >> the
>> >> >>> previous setting and did not work, and within the configuration
>> >> >>> routing-instance <name> not have all the options.
>> >> >>>
>> >> >>> Thanks
>> >> >>> Gabriel Farias
>> >> >>>
>> >> >>> 2010/3/31 Stefan Fouant <sfouant at shortestpathfirst.net>
>> >> >>>
>> >> >>>> You need the following:
>> >> >>>>
>> >> >>>> 'set routing-option forwarding-table export load-balance'
>> >> >>>>
>> >> >>>> and
>> >> >>>>
>> >> >>>> 'set policy-options policy-statement load-balance then
>> >> >>>> load-balance
>> >> >>>> per-packet'
>> >> >>>>
>> >> >>>> Sorry for the top-post, I'm on my Blackberry.
>> >> >>>>
>> >> >>>> HTHs.
>> >> >>>>
>> >> >>>> Stefan Fouant
>> >> >>>> ------Original Message------
>> >> >>>> From: Gabriel Farias
>> >> >>>> Sender: juniper-nsp-bounces at puck.nether.net
>> >> >>>> To: juniper-nsp at puck.nether.net
>> >> >>>> Subject: [j-nsp] Load Balance in VRF by Junos
>> >> >>>> Sent: Mar 31, 2010 4:23 PM
>> >> >>>>
>> >> >>>> Hi members,
>> >> >>>>
>> >> >>>>
>> >> >>>> How to balance traffic between EBGP equipment Juniper M120 (PE) x
>> >> Cisco
>> >> >>>> 6509
>> >> >>>> (CE) Switch, which are connected via ethernet aggregated?
>> >> >>>>
>> >> >>>>
>> >> >>>>
>> >> >>>> I used the setup below and did not work the Juniper continues with
>> >> the
>> >> >>>> output unbalanced traffic.
>> >> >>>>
>> >> >>>>
>> >> >>>>
>> >> >>>> configuration forwarding-options
>> >> >>>>
>> >> >>>> load-balance {
>> >> >>>>    indexed-next-hop;
>> >> >>>> }
>> >> >>>> hash-key {
>> >> >>>>    family inet {
>> >> >>>>        layer-3;
>> >> >>>>        layer-4;
>> >> >>>>    }
>> >> >>>> }
>> >> >>>>
>> >> >>>>
>> >> >>>> Thanks,
>> >> >>>>
>> >> >>>> Gabriel Farias
>> >> >>>> _______________________________________________
>> >> >>>> juniper-nsp mailing list juniper-nsp at puck.nether.net
>> >> >>>> https://puck.nether.net/mailman/listinfo/juniper-nsp
>> >> >>>>
>> >> >>>>
>> >> >>>> Sent from my Verizon Wireless BlackBerry
>> >> >>>
>> >> >>>
>> >> >>>
>> >> >>
>> >> >
>> >> _______________________________________________
>> >> juniper-nsp mailing list juniper-nsp at puck.nether.net
>> >> https://puck.nether.net/mailman/listinfo/juniper-nsp
>> >>
>> >
>> >
>> _______________________________________________
>> juniper-nsp mailing list juniper-nsp at puck.nether.net
>> https://puck.nether.net/mailman/listinfo/juniper-nsp
>>
>>
>>
>>       __________________________________________________________________
>> Be smarter than spam. See how smart SpamGuard is at giving junk email the
>> boot with the All-new Yahoo! Mail.  Click on Options in Mail and switch to
>> New Mail today or register for free at http://mail.yahoo.ca
>>
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp
>

-- 
Sent from my mobile device


More information about the juniper-nsp mailing list