hsrp group for traffic sharing

[suthomas1]

Hi Everyone,

Having 2 6506 switches, is it advisable to configure the load sharing using hsrp multiple groups. Is there any other way to achieve a condition where both switches are being active at the same time.

Is it efficient to have such a setup. please help me to see the merits and demerits of such setup.

Thanks in advance to all.

[paolo bevilacqua]

With L3 switches or powerful routers, sharing traffic is not necessary, because they can process traffic at linerate under all conditions.

[Joseph W. Doherty]

Disclaimer

The Author of this posting  offers the information contained within this posting without   consideration and with the reader's understanding that there's no      implied or expressed suitability or fitness for any purpose.    Information provided is for informational purposes only and should not   be construed as rendering professional advice of any kind.  Usage of   this posting's information is solely at reader's own risk.

Liability Disclaimer

In  no event shall Author be liable for any damages whatsoever (including,   without limitation, damages for loss of use, data or profit) arising  out  of the use or inability to use the posting's information even if  Author  has been advised of the possibility of such damage.

Posting

As Paolo notes, powerful L3 switches that can offer line rate don't really need to share.  The 6500 platform doesn't always offer line rate, but unless you're working with lots of busy gig with older "classic" or CEF256 cards, you rarely exceed a 6500's capacity.  Of course, it is possible, which is why there's now the Nexus series. 

If you describe what the 6506 are populated with, along with typical traffic, we could guesstimate how close you might be, or not, to your 6506's capacity.

If you do want to use both 6506's actively, you might find GLBP an easier way to load balance rather than HSRP or mHSRP.  If you do load balance, you'll need to be especially careful you don't set yourself up for unicast flooding.

Besides gateway balancing, you might also be able to balance using PVSPT or, if Cisco switches downstream, perhaps with Flexlink.

Personally, I prefer to have all active devices active to know they are working correctly.  I seem to bump into situations where hot/warm configurations don't work right when the hot fails and warm becomes active.  However, the other side of the coin is, with active/active when there is a problem it can be much harder to identify where it's at.

[Jon Marshall]

MHSRP is a pain because it requires manual config of the default-gateway on the client.

If you want to load balance then use GLBP as Joseph suggested. But it really depends on the topology ie.

you have 2 distro switches ds1 and ds2 and an access-layer switch as1.

In a typical setup the 2 disto switches are interconnected via a L2 trunk and the access switch is connected to both distro switches.

ds1 is STP root and HSRP active for odd vlans

ds2 is STP root and HSRP active for even vlans.

as1 link to ds1 forwards for odd vlans and blokcs for even vlans

as1 link to ds2 forwards for even vlans and blocks for odd vlans

the above is a common and well proven setup. Now if you then replaced HSRP with GLBP what happens is that the interconnect link between ds1 and ds2 now gets a lot more traffic. The reason for this is that as1 -> ds2 is blocking for odd vlans. But GLBP will hand out both ds1 and ds2 addresses when clients in an odd vlan arp out for their default-gateway. So if a client in an odd vlan gets ds2 mac-address back the only path it can take is as1 -> ds1 -> ds2 which is suboptimal when you compare it with the previous setup.

Where GLBP is useful is if you have implemented a L3 interconnect between ds1 and ds2 instead of a L2 trunk. Then both links from as1 will be forwarding for ALL vlans and now packets can go straight to either ds1 or ds2 from as1.

As Paolo noted it's not that much of an issue. If you truly are overloading one link then load balancing may be the answer but bear in mind that if one link fails you are simply back to the overloaded link which may or may not be acceptable to the business ie. you may need to upgrade links rather than implement load balancing.

Jon

[suthomas1]

Thanks to all for their valuable inputs.

Please allow me to pose another query related to same setup.where i have 2 coreswitches , few access switches. these access switches are trunked to each other and also have individual uplink to both core1 and core2.

in this case, if i use stp priority for vlans like:

core1 is primary root for vlan 100 & secondary for vlan 200

core2 is primary root for vlan200 & secondary for vlan300 , so forth.

and setup manual priorities. how efficient will this be and what will be the demerits for it.

Appreciate all inputs & suggestions.

Thanks.