From: Volkov, Dmitry (Toronto - BCE) (dmitry_volkov@xxxxxxxxx)
Date: Wed Aug 28 2002 - 23:22:52 GMT-3
Yes.
This is kind a trick. Order (age of route) - how codes compares BGP routes
is very important,
despite that age of route is only factor Nr10:
http://www.cisco.com/warp/public/459/25.shtml:
10. "When both paths are external, prefer the path that was received first
(the oldest one). "
This effect may not be visible in simple lab setups. But i "real life" :)
setups:
"This can occur when the same route is received from multiple ASs or
confederation sub-ASs, with exactly the same path length, but different
MEDs. "
It's why: "Cisco recommends enabling bgp deterministic-med in all new
network deployments"
Dmitry
> -----Original Message-----
> From: Volkov, Dmitry (Toronto - BCE)
> Sent: Wednesday, August 28, 2002 10:11 PM
> To: 'Nick Shah'; Volkov, Dmitry (Toronto - BCE); 'Mingzhou Nie'; Jon
> O'Nan
> Cc: ccielab@groupstudy.com
> Subject: RE: deterministic med
>
>
> I run 12.1(15) on R1 & R10 and 12.0(23) on R8
> The guy who compares meds is 12.1(15) - R1
>
> I've just read it http://www.cisco.com/warp/public/459/37.html again.
> And I guess I understand. BTW I don't remember any bgp determ
> med in the examples in Doyle and halabi when they ecplain MEDs.
>
> Key phrase here: If you disable bgp deterministic-med, the
> ORDER in which routes are received may impact MED-based best
> path decisions.
> ---skiped---
> The order in which the BGP routes were received is entry3,
> entry2, and entry1 (entry3 is the oldest entry in the BGP
> table and entry1 is the newest one).
> ---skiped---
> A BGP router with bgp deterministic-med disabled chooses
> entry2 over entry1
> So router compared between entry 1) and entry 2) - if it's
> disabled (1st example)
> ---skiped---
> With bgp deterministic-med enabled, paths are grouped by
> peer-AS and the best path for each group is selected. So it
> compares between entry 1) and entry 3)
> AFTER THAT winner is compared with entry 2
>
> Dmitry
>
> > -----Original Message-----
> > From: Nick Shah [mailto:nshah@connect.com.au]
> > Sent: Wednesday, August 28, 2002 10:02 PM
> > To: Volkov, Dmitry (Toronto - BCE); 'Mingzhou Nie'; Jon O'Nan
> > Cc: ccielab@groupstudy.com
> > Subject: Re: deterministic med
> >
> >
> > hmm.. That's weird.
> > I am just testing it as we speak, and I am seeing different
> > behaviour to
> > what you are seeing.
> >
> > Running 12.1(5)T9
> > What are you runnign ?
> >
> > Nick
> > ----- Original Message -----
> > From: "Volkov, Dmitry (Toronto - BCE)" <dmitry_volkov@ca.ml.com>
> > To: "'Mingzhou Nie'" <mnie@yahoo.com>; "Nick Shah"
> > <nshah@connect.com.au>;
> > "Jon O'Nan" <jononan@columbus.rr.com>
> > Cc: <ccielab@groupstudy.com>
> > Sent: Thursday, August 29, 2002 11:53 AM
> > Subject: RE: deterministic med
> >
> >
> > > Guys,
> > >
> > > I just tested it. It's NOT TRUE. MEDs are compared BY DEFAULT !
> > >
> > > I have 3 routes on common ethernet segment
> > > r1 160.100.2.1; r10 160.100.2.250 and R8 160.100.2.254
> > > r10 R8
> > > | |
> > > r1-----------------
> > >
> > > R10 & R8 in AS 2001, R1 in AS 2010
> > >
> > > R8 advert networks via network statements
> > >
> > > R1 sees routes:
> > > Network Next Hop Metric LocPrf Weight Path
> > > *> 10.10.10.0/28 133.7.4.4 0 100
> > 0 (65302) 2020
> > i
> > > * 160.100.100.0/24 160.100.2.254 0
> 0 2001 i
> > > *> 160.100.2.254
> 0 2001 i
> > > * 160.100.128.0/24 160.100.2.254 0
> 0 2001 i
> > > *> 160.100.2.254
> 0 2001 i
> > > * 160.100.129.0/24 160.100.2.254 0
> 0 2001 i
> > > *> 160.100.2.254
> 0 2001 i
> > > * 160.100.130.0/24 160.100.2.254 0
> 0 2001 i
> > > *> 160.100.2.254
> 0 2001 i
> > >
> > > No any metrics in place no any bgp determenistic meds in place.
> > > Tiebreiker in BGP selection process is Oldest route.
> > > Whenever I did on R1: neighbor R8 shut no shut or neigbor
> > R10 shut no shu
> > > best route on R1 was changing.
> > >
> > > After I put route-map on R10 advertising network
> > 160.100.100.0 with metric
> > > 121 towards to R1,
> > > this network also has best route via R8 !!! - So MED plays
> > his role in BGP
> > > best path desision process ! without any bgp determ med on
> > any routers.
> > >
> > > r1#sh ip bgp 160.100.100.0
> > > BGP routing table entry for 160.100.100.0/24, version 40
> > > Paths: (2 available, best #1, table Default-IP-Routing-Table)
> > > Advertised to non peer-group peers:
> > > 133.7.2.2 160.100.2.250
> > > 2001
> > > 160.100.2.254 from 160.100.2.254 (192.190.102.1)
> > > Origin IGP, metric 0, localpref 100, valid, external, best
> > > 2001
> > > 160.100.2.254 from 160.100.2.250 (101.101.101.101)
> > > Origin IGP, metric 121, localpref 100, valid, external
> > >
> > >
> > > Dmitry
> > >
> > >
> > >
> > > > -----Original Message-----
> > > > From: Mingzhou Nie [mailto:mnie@yahoo.com]
> > > > Sent: Wednesday, August 28, 2002 9:40 PM
> > > > To: Nick Shah; Volkov, Dmitry (Toronto - BCE); Jon O'Nan
> > > > Cc: ccielab@groupstudy.com
> > > > Subject: Re: deterministic med
> > > >
> > > >
> > > > Nick,
> > > >
> > > > Don't believe that's true. MED is always compared whether "bgp
> > > > deterministic med " enabled to not. the command simply
> changes the
> > > > comparison behavior a little bit by grouping routes from
> > same neiborgh
> > > > and compare.
> > > >
> > > > In the given example, if you change the order as below.
> > > >
> > > > entry1: AS(PATH) 500, med 150, external, rid 172.16.13.1
> > > > entry2: AS(PATH) 500, med 100, internal, rid 172.16.8.4
> > > > entry3: AS(PATH) 100, med 200, external, rid 1.1.1.1
> > > >
> > > > even w/o the command, entry2 will be chosen over entry1.
> > > >
> > > > ming
> > > >
> > > > --- Nick Shah <nshah@connect.com.au> wrote:
> > > > > > Can we say that by default (bgp deterministic-med is
> > not enabled)
> > > > > > Metric(MED) is not compared between ANY (doesn't
> > matter from the
> > > > > same AS
> > > > > or
> > > > > > not) routes with the same path lenghts ?
> > > > >
> > > > > Think it this way, whats the BGP path decision
> process, Weight,
> > > > > localpref,
> > > > > origin code, ASPATH, MED (stop here). Now unless bgp
> > > > > deterministic-med is
> > > > > enabled, MED comparison is skipped.
> > > > >
> > > > > So if all else be the same till here (till MED) and you
> > really want
> > > > > MED to
> > > > > influence path, then turn on 'BGP DETERMINISTIC MED'
> > > > >
> > > > > > I.e. MED - is not taken into consideration in the
> > > > > > BGP best path selection alghorithm ? (by default) So
> > forget about
> > > > > MED
> > > > > > completely until enable "bgp-deterministic-med" or "bgp
> > > > > > always-compare-med" ?
> > > > >
> > > > > Correct
> > > > >
> > > > > > Also this sentence:
> > > > > > (MED is used in this comparison by default because
> > the winners of
> > > > > each
> > > > > group
> > > > > > are now from different ASs - enabling bgp always-compare-med
> > > > > changes this
> > > > > > default behavior).
> > > > > > from http://www.cisco.com/warp/public/459/37.html
> > > > > >
> > > > > > Is it mistake ?? Should I read (MED is NOT used in this
> > > > > comparision....)
> > > > > > because it doesn't compare MEDs from different ASs
> until "bgp
> > > > > > always-compare-med" is enabled ?
> > > > >
> > > > > That looks like a'convoluted statement' and a typo.
> > > > >
> > > > > The correct document is : (meaning, one which confirms
> > the actual
> > > > > behaviour
> > > > > is )
> > > > > http://www.cisco.com/warp/public/459/bgp-med.html
> > > > >
> > > > > //snip//Enabling the bgp deterministic-med command ensures the
> > > > > comparison of
> > > > > the MED variable when choosing routes advertised by
> > different peers
> > > > > in the
> > > > > same autonomous system (AS). Enabling the bgp
> always-compare-med
> > > > > command
> > > > > ensures the comparison of the MED for paths from neighbors in
> > > > > different
> > > > > autonomous systems. The bgp always-compare-med command is
> > > > useful when
> > > > > multiple service providers or enterprises agree on a
> > uniform policy
> > > > > for
> > > > > setting MED. Thus, for network X, if Internet Service
> > Provider (ISP)
> > > > > A sets
> > > > > the MED to 10, and ISP B sets the MED to 20, then both
> > ISPs agree
> > > > > that ISP A
> > > > > has the better performing path to X. //snip//
> > > > >
> > > > > rgds
> > > > >
> > > > > Nick
> > > > >
> > > > > >
> > > > > > > -----Original Message-----
> > > > > > > From: Nick Shah [mailto:nshah@connect.com.au]
> > > > > > > Sent: Wednesday, August 28, 2002 7:36 PM
> > > > > > > To: Jon O'Nan; ccielab@groupstudy.com
> > > > > > > Subject: Re: deterministic med
> > > > > > >
> > > > > > >
> > > > > > > Jon
> > > > > > >
> > > > > > > I hope you have already read this paper..
> > > > > > > http://www.cisco.com/warp/customer/459/bgp-med.html
> > > > > > >
> > > > > > > regardless of that. I will list down the effect of these 2
> > > > > commands.
> > > > > > >
> > > > > > > * By default (with none of these commands entered)
> > MED is NOT
> > > > > > > compared when
> > > > > > > selecting the best path.
> > > > > > > * Turning on bgp deterministic-med, turns on the
> feature to
> > > > > > > compare MED,
> > > > > > > however it only compares MED, if prefix is received
> > from same
> > > > > > > neighbor and
> > > > > > > the best man wins
> > > > > > > * Turning on bgp always-compare-med, starts
> comparing 'same
> > > > > > > prefix' recd.
> > > > > > > from 'different neighbors' and the best man wins
> > > > > > > * Turning on both of them may really not be
> required, but it
> > > > > basically
> > > > > > > compares if prefixes are recd. in groups of AS's (say same
> > > > > > > prefix from AS
> > > > > > > 200, via 2 neighbors, and same prefix from AS 100 via 1
> > > > > neighbor).
> > > > > > >
> > > > > > > rgds
> > > > > > > Nick
> > > > > > > ----- Original Message -----
> > > > > > > From: "Jon O'Nan" <jononan@columbus.rr.com>
> > > > > > > To: <ccielab@groupstudy.com>
> > > > > > > Sent: Thursday, August 29, 2002 8:58 AM
> > > > > > > Subject: deterministic med
> > > > > > >
> > > > > > >
> > > > > > > > Am I correct that:
> > > > > > > >
> > > > > > > > bgp deterministic med - compares med within AS
> > > > > > > >
> > > > > > > > VS.
> > > > > > > >
> > > > > > > > bgp always-compare-med - compares med from another AS
> > > > > > > >
> > > > > > > >
> > > > > > > > Thanks
> > > > > > > > Jon
> > > > > > > >
> > > > >
This archive was generated by hypermail 2.1.4 : Sat Sep 07 2002 - 19:48:41 GMT-3