[j-nsp] Re: M5 - outbound load balancing

Korey Verlsteffen korey at webstream.net
Wed Jan 19 17:02:39 EST 2005


>Try configuring these two peers under the same peer group as in the following:

I change it to one group as you suggested, same results.

:Inactive reason: Router ID



----- Original Message ----- 
From: Fouant, Stefan A 
To: Korey Verlsteffen ; Harry Reynolds ; juniper-nsp at puck.nether.net 
Sent: Wednesday, January 19, 2005 4:48 PM
Subject: RE: [j-nsp] Re: M5 - outbound load balancing


Hi,

The "show route extensive" output shows that the second path is inactive
because of the Router-ID which indicates that multipath is not working
correctly - proper operation of multipath skips the Router-ID and Peer-ID
evaluation during the BGP Path Selection algorithm.  I don't think multipath
is working in your case because you've configured these two peers under
different groups.  

Try configuring these two peers under the same peer group as in the
following:

bgp {
    log-updown;
    damping;
    group to-ISPs {
        type external;
        multipath;
        neighbor 1.2.3.4 {
            import provider1-policy;
            export my-bgp-network;
            peer-as 1234;
        neighbor 5.6.7.8 {
            import provider2-policy;
            export my-bgp-network;
            peer-as 5678;
        }
    }
}

HTHs.
Stefan Fouant
Senior Network Engineer

-----Original Message-----
From: juniper-nsp-bounces at puck.nether.net
[mailto:juniper-nsp-bounces at puck.nether.net] On Behalf Of Korey Verlsteffen
Sent: Wednesday, January 19, 2005 4:16 PM
To: Harry Reynolds; juniper-nsp at puck.nether.net
Subject: [j-nsp] Re: M5 - outbound load balancing

65.0.0.0/19 (2 entries, 1 announced)
        *BGP    Preference: 170/-101
                Source: 64.31.144.93
                Next hop: 64.31.144.93 via t3-0/0/1.0,, selected
                Protocol next hop: 64.31.144.93 Indirect next hop: 845d088
4861
                State: <Active Ext>
                Local AS: 22275 Peer AS: 19094
                Age: 26:52      Metric2: 0 
                Task: BGP_19094.64.31.144.93+179
                Announcement bits (2): 0-KRT 2-Resolve inet.0 
                AS path: 19094 7911 6389 7893 I
                Localpref: 100
                Router ID: 24.56.99.16
         BGP    Preference: 170/-101
                Source: 205.198.5.1
                Next hop: 205.198.5.1 via t3-0/0/0.0,, selected
                Protocol next hop: 205.198.5.1 Indirect next hop: 845d000
4860
                State: <Ext>
                Inactive reason: Router ID
                Local AS: 22275 Peer AS:  4200
                Age: 26:57      Metric2: 0 
                Task: BGP_4200.205.198.5.1+179
                AS path: 4200 174 6389 7893 I
                Localpref: 100
                Router ID: 205.198.4.9


----- Original Message ----- 
From: Harry Reynolds 
To: Korey Verlsteffen ; juniper-nsp at puck.nether.net 
Sent: Wednesday, January 19, 2005 4:09 PM
Subject: RE: [j-nsp] Re: M5 - outbound load balancing


Can you do the show route again, but w/the detail or extensive switch?
The display should tell us something about why the other route is not
active. Note that multipath has no effect if, for example, MED is being
used to break a tie.

Cheers
 

> -----Original Message-----
> From: juniper-nsp-bounces at puck.nether.net 
> [mailto:juniper-nsp-bounces at puck.nether.net] On Behalf Of 
> Korey Verlsteffen
> Sent: Wednesday, January 19, 2005 1:06 PM
> To: juniper-nsp at puck.nether.net
> Subject: [j-nsp] Re: M5 - outbound load balancing
> 
> >Can we see your configuration under [protocols bgp]?
> 
> Here you go minus the real as and ip info.  Thank you for your help.
> 
> bgp {
>     log-updown;
>     damping;
>     group provider1 {
>         type external;
>         multipath;
>         neighbor 1.2.3.4 {
>             import provider1-policy;
>             export my-bgp-network;
>             peer-as 1234;
>         }
>     }
>     group provider2 {
>         type external;
>         multipath;
>         neighbor 5.6.7.8 {
>             import provider2-policy;
>             export my-bgp-network;
>             peer-as 5678;
>         }
>     }
> }
> 
> 
> 
> 
> ----- Original Message -----
> From: Fouant, Stefan A
> To: Korey Verlsteffen ; Erdem Sener ; juniper-nsp at puck.nether.net
> Sent: Wednesday, January 19, 2005 3:40 PM
> Subject: RE: [j-nsp] Re: M5 - outbound load balancing
> 
> 
> Hi,
> 
> Problem is here that BGP is making the decision as to which 
> route to make active (in the Local-RIB) and then sending that 
> single route to the routing table.  So from the routing table 
> perspective you don't' have multiple equal-cost paths to 
> choose from, hence the load-balance per-packet policy does not work.
> 
> You might want to try BGP multipath.  Can we see your 
> configuration under [protocols bgp]?
> 
> Stefan Fouant
> Senior Network Engineer
> UUNET / MCI
> 
> -----Original Message-----
> From: juniper-nsp-bounces at puck.nether.net
> [mailto:juniper-nsp-bounces at puck.nether.net] On Behalf Of 
> Korey Verlsteffen
> Sent: Wednesday, January 19, 2005 3:17 PM
> To: Erdem Sener; juniper-nsp at puck.nether.net
> Subject: [j-nsp] Re: M5 - outbound load balancing
> 
> I tried the per-packet policy but it don't think I have it 
> setup right.
> 
> A show route 65.0.0.1 shows me two equal paths
> 
> net.0: 151671 destinations, 302507 routes (151522 active, 0 
> holddown, 321
> hidden)
> + = Active Route, - = Last Active, * = Both
> 
> 65.0.0.0/19        *[BGP/170] 1d 00:43:22, localpref 100
>                       AS path: XXX XXX XXX XXX I
>                     > to 1.2.3.4 via t3-0/0/1.0
>                     [BGP/170] 1d 00:43:28, localpref 100
>                       AS path: XXX XXX XXX XXX I
>                     > to 5.6.7.8 via t3-0/0/0.0
> 
> But a show route forwarding-table destination 65.0.0.1 shows 
> me only one destination
> 
> Routing table: inet
> Internet:
> Destination        Type RtRef Next hop          Type Index NhRef Netif
> 65.0.0.0/19        user     0 1.2.3.4      ucst    25 101673 
> t3-0/0/1.0
> 
> What am I doing wrong?
> 
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net 
> http://puck.nether.net/mailman/listinfo/juniper-nsp
> 

_______________________________________________
juniper-nsp mailing list juniper-nsp at puck.nether.net
http://puck.nether.net/mailman/listinfo/juniper-nsp



More information about the juniper-nsp mailing list