[c-nsp] Load Balancing of Unequal Ethernet Bandwidth
Arie Vayner (avayner)
avayner at cisco.com
Mon Feb 16 07:58:15 EST 2009
One trick you could do is to use VLANs to virtualized the links...
On the 100M link run a single VLAN while on the 200M link run 2 sub
interfaces, each on a different VLAN.
This would create 3 equal cost paths (well, you need to set the cost
manually to be the same) and you would get automatic load sharing.
Arie
-----Original Message-----
From: cisco-nsp-bounces at puck.nether.net
[mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Andy Saykao
Sent: Monday, February 16, 2009 09:47
To: Ben Steele; td_miles at yahoo.com
Cc: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] Load Balancing of Unequal Ethernet Bandwidth
Tony - we run OSPF as our IGP.
Ben - on our WAN link we run BGP. We also have ethernet circuits between
our POPS in each capital city which we run OSPF over.
I'll give that link a read. Thanks
________________________________
From: Ben Steele [mailto:illcritikz at gmail.com]
Sent: Monday, 16 February 2009 6:06 PM
To: td_miles at yahoo.com
Cc: Andy Saykao; cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] Load Balancing of Unequal Ethernet Bandwidth
Alternatively if you are using BGP, have a look at BGP Link Bandwidth
http://www.cisco.com/en/US/docs/ios/12_2t/12_2t2/feature/guide/ftbgplb.h
tml
On Mon, Feb 16, 2009 at 5:32 PM, Tony <td_miles at yahoo.com> wrote:
Hi Andy,
What do you run as IGP then so that we can help you out ?
If static routes, then you can do it using by having multiple
routes that are to the same destination.
eg. on 2x serial links you might have:
serial1 = 200Mbps (10.1.1.1/30)
serial2 = 100Mbps (10.1.1.5/3)
You would then add static routes like this:
ip route x y serial1
ip route x y 10.1.1.2
ip route x y serial2
This way when you do "show ip route x" you would see something
like:
* directly connected via serial1
Route metric is 0, traffic share count is 1
* directly connected via serial2
Route metric is 0, traffic share count is 1
* 10.1.1.2
Route metric is 0, traffic share count is 1
Your router would then divide the traffic into three with one
third going to each of the destinations configured. The fact that two of
those destinations are the same link means that two thirds will go down
your 200Mbps link and one third down your 100Mbps link.
This is fairly basic and doesn't scale very well, but will work.
regards,
Tony.
--- On Mon, 16/2/09, Andy Saykao
<andy.saykao at staff.netspace.net.au> wrote:
> From: Andy Saykao <andy.saykao at staff.netspace.net.au>
> Subject: Re: [c-nsp] Load Balancing of Unequal Ethernet
Bandwidth
> To: "Ben Steele" <illcritikz at gmail.com>
> Cc: cisco-nsp at puck.nether.net
> Date: Monday, 16 February, 2009, 5:39 PM
> Hi Ben,
>
> When I googled around, there were many discussions abvout
> using the
> variance command with eigrp but we don't run eigrp
> internally as our
> IGP.
>
> This is a typical setup where we need to upgrade some of
> our links, so
> we might upgrade 50M on the second leg and end up with a
> situation where
> the first leg is100M and the second leg is 150M. As you may
> know, some
> providers aren't so flexible so you can't just
> upgrade 25M on each leg
> because they increment by 50M per leg only. Hence my
> question if it was
> possible to load balance across unequal ethernet circuits
> without buying
> additional bandwidth for both circuits.
>
> Thanks.
>
> Andy
>
>
> ________________________________
>
> From: Ben Steele [mailto:illcritikz at gmail.com]
> Sent: Monday, 16 February 2009 5:29 PM
> To: Andy Saykao
> Subject: Re: [c-nsp] Load Balancing of Unequal Ethernet
> Bandwidth
>
>
> You could do this with variance in eigrp, just add variance
> 2 into the
> eigrp config and it will load balance on a 2:1 ratio, if
> your links are
> equally matched in terms of latency you can look at
> enabling per-packet
> load sharing on the 2 egress interfaces to get an even more
> granular
> distribution, this can wreck some havoc with unequal paths
> and out of
> sequence packets though, however if equally similar in
> characteristics
> then performance is usually very good.
>
> Ben
>
>
> On Mon, Feb 16, 2009 at 4:01 PM, Andy Saykao
> <andy.saykao at staff.netspace.net.au> wrote:
>
>
> Is it possible to aggregate and then load balance
unequal
> ethernet
> circuits like so:
>
> I have two ethenet circuits on my Cisco router. Both
have
> equal
> costs to
> the next hop.
>
> Ethernet Circuit #1- 200M
> Ethernet Circuit #2 - 100M
>
> Can I aggregate both ethernet circuits so that the total
> amount
> of
> bandwidth available to the next hop is is 300M?
> Can I then load balance it so both circuits are equally
> utilized?
>
> For example...
>
> * If I have 150M of traffic flowing to the next hop then
> the
> router
> would spread the load across both links like so:
>
> 100M through Ethernet Circuit #1.
> 50M through Ethernet Circuit #2.
>
> * The formula to use for this would be something like:
>
> Utilization / Total Bandwidth = percentage of
utilization
> required per
> link
> 150/300 = 0.5
>
> 0.5 x bandwidth of Ethernet #1 = 0.5 x 200 = 100M
> 0.5 x bandwidth of Ethernet #1 = 0.5 x 100 = 50M
>
> * If there was a total of 250M of traffic flowing to the
> next
> hop, and
> applying the formula above, the router would work out
that
> the
> load
> distributed across both ethernet links would be:
>
> 166M through Ethernet Circuit #1.
> 84M through Ethernet Circuit #2.
>
> Any ideas???
>
> Thanks.
>
> Andy
_______________________________________________
cisco-nsp mailing list cisco-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/
______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email
______________________________________________________________________
This email and any files transmitted with it are confidential and
intended
solely for the use of the individual or entity to whom they are
addressed.
Please notify the sender immediately by email if you have received this
email by mistake and delete this email from your system. Please note
that
any views or opinions presented in this email are solely those of the
author and do not necessarily represent those of the organisation.
Finally, the recipient should check this email and any attachments for
the presence of viruses. The organisation accepts no liability for any
damage caused by any virus transmitted by this email.
_______________________________________________
cisco-nsp mailing list cisco-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/
More information about the cisco-nsp
mailing list