GroupStudy.com GroupStudy.com - A virtual community of network engineers
 Home  BookStore  StudyNotes  Links  Archives  StudyRooms  HelpWanted  Discounts  Login
RE: Dlsw load balancing posted 09/30/2004
[Chronological Index] [Thread Index] [Top] [Date Prev][Date Next] [Thread Prev][Thread Next]


A quandry, it is....

Once you enable load-balancing, then circuit weight is the measurement that
is looked at first.  If you are not load-balancing, then peer cost is used
to determine the best path to a destination.

HTH,

 
Scott Morris, CCIE4 (R&S/ISP-Dial/Security/Service Provider) #4713, CISSP,
JNCIP, et al.
IPExpert CCIE Program Manager
IPExpert Sr. Technical Instructor
swm@xxxxxxxxxx/smorris@xxxxxxxxxxxx
http://www.ipexpert.net
 
 

-----Original Message-----
From: ccie2be [mailto:ccie2be@xxxxxxxxxx] 
Sent: Thursday, September 30, 2004 1:58 PM
To: swm@xxxxxxxxxx; 'Group Study'
Subject: Re: Dlsw load balancing

Thanks Scott, u da mon !!!

And, while we're on the topic of dlsw...

I was just reviewing some of my notes and cam across an interesting load
balancing issue.

Suppose there's a "conflict" between circuit weight and cost.  For example,

                     /  RTR -B
                /
RTR-A
                \
                    \   RTR -C


rtr-A is configured with:

dlsw remote-peer 0 tcp  rtr-B circuit weight 10 dlsw remote-peer 0 tcp
rtr-C circuit weight 10

So, rtr-A should distribute new circuits equally between rtr-B and rtr-C
based on above.

But, rtr-B is configured with:

dlsw local-peer peer-id rtr-B cost 2

And, rtr-C is configured with:

dlsw local-peer peer id rtr-C cost 4

What happens now?

Based on rtr-B's and rtr-C's config, rtr-A should only connect to rtr-B
because it has the lowest cost.
But, based on rtr-A config, connections should be balanced between the 2
remote peers.

And, if the config's were changed so that rtr-A's remote-peer commands were
config'd with cost, but rtr-B and rtr-C local peer commands were config's
with circuit weight in a "conflicting" way, what would happen?

In other words, does cost take precedence over circuit weight or does what's
configured on remote peer still take precedence over what's configured on
local peer?

And, 1 last scenario.  Is there a "default" cost?  For example, suppose
using the same rtr's as above, rtr-A should prefer rtr-B, but no special
config is allowed on rtr-A.  If rtr-B has NO cost configured and rtr-C has
cost X,  will rtr-A prefer rtr-B or rtr-C?

I don't know if these type of issues are going way beyond what might be
found on the lab, but it seems like Cisco keeps upping the ante in terms of
difficulty, so I figure I better just make sure I know this stuff.

Thanks, Tim

----- Original Message -----
From: "Scott Morris" <swm@xxxxxxxxxx>
To: "'ccie2be'" <ccie2be@xxxxxxxxxx>; "'Group Study'"
<ccielab@xxxxxxxxxxxxxx>
Sent: Thursday, September 30, 2004 1:09 PM
Subject: RE: Dlsw


> Correct.  While advertised during the peer's capabilities exchange, I may
> tell you one thing, but in your remote-peer statement to me, you "know
> better" and whatever value you have locally for our peering relationship
> overrides what I may try to tell you.
>
> HTH,
>
>
> Scott Morris, CCIE4 (R&S/ISP-Dial/Security/Service Provider) #4713, CISSP,
> JNCIP, et al.
> IPExpert CCIE Program Manager
> IPExpert Sr. Technical Instructor
> swm@xxxxxxxxxx/smorris@xxxxxxxxxxxx
> http://www.ipexpert.net
>
>
>
> -----Original Message-----
> From: nobody@xxxxxxxxxxxxxx [mailto:nobody@xxxxxxxxxxxxxx] On Behalf Of
> ccie2be
> Sent: Thursday, September 30, 2004 12:29 PM
> To: Group Study
> Subject: Dlsw
>
> Hi guys,
>
> I've noticed that some parameters e.g. cost, circuit weight, etc can be
used
> on both the dlsw local peer and dlsw remote peer commands.
>
> Is it always true that if the same parameter is configured on both dlsw
peer
> (local & remote), the parameter configured on the remote command takes
> precedence?
>
>
> TIA, Tim
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html