[nsp] HSRP

Thu Jun 19 15:24:58 EDT 2003

one way to simplify load balancing is via GLBP the successor to HSRP.
if you are not aware of GLBP, and want a less configuration intensive
load balancing scheme, some investigation into GLBP may be worth your time.

http://www.cisco.com/en/US/products/sw/iosswrel/ps1839/products_feature_guide09186a00801541c8.html

GLBP makes use of a single virtual IP, and multiple virtual MAC addresses,
with an election scheme between the GLBP participant routers to split
the load of the end devices.  since each participant has a different VMAC,
it may go some way to avoiding the problem you saw.

cheers
.siva

On Thu, 19 Jun 2003, k. scott bethke wrote:

> Great thread guys..  Also note that multigroup HSRP is a way to load balance
> traffic, I always hated the fact that one router (or even just one router
> interface) was standby why not use it?  With mutigroup you can setup two
> VIP's essentually one homed on each interface and you can either point half
> your servers to one or half your servers to the other and balance traffic
> that way, or you can actually put two default routes on each server and it
> will do the right thing and balance traffic out to both routers.
>
> Has anyone noticed really strange issues with HSRP on dot1q based Vlan
> interfaces using extreme networks gear as the switches?  I used to run a
> network with really old 7500's and 7200's and we just had issues all the
> time with the vlans not communicating randomly.  It would work fine and just
> stop and then oneday just start working again.  Very strange.  A major issue
> with HSRP in this capacity is that both interfaces claim the other has died
> and BOTH take ownership of the VIP.  I can imagine that in this case a
> unicast solution MAY work better than multicast.
>
> -Scotty
> ----- Original Message -----
> From: "Thomas Renzy" <thomas.renzy at veritas.com>
> To: "'Peter B. Juul'" <peter.juul at uni-c.dk>; "mac" <mac at telvia.it>
> Cc: <cisco-nsp at puck.nether.net>
> Sent: Thursday, June 19, 2003 4:05 PM
> Subject: RE: [nsp] HSRP
>
>
> > Peter,
> >
> > HSRP uses Multicast and the address is 224.0.0.2. HSRP also uses UDP on
> port
> > 1985.
> >
> > You can also HSRP to track interfaces and to change the priority in case
> > there is a problem with it. Here is an example of this.
> >
> > interface Vlan10
> >  description Server-Vlan
> >  ip address 10.1.1.2 255.255.255.0
> >  standby ip 10.1.1.1
> >  standby priority 120 preempt
> >  standby track GigabitEthernet2/16 75
> >
> > Here this vlan is tracking interface G2/16. The number "75" tells how much
> > the decrement the priority when the tracked interface goes down.
> >
> > Thomas
> >
> > Thomas Renzy
> > IS&T Global Network Services
> > VERITAS Software
> > Office: +650-527-4734
> > Mobile: +650-248-1099
> > Fax: +650-527-2034
> >
> > "Some people drink from the fountain of knowledge, others just gargle." -
> > Author Robert Anthony
> >
> >
> >
> > -----Original Message-----
> > From: Peter B. Juul [mailto:peter.juul at uni-c.dk]
> > Sent: Thursday, June 19, 2003 12:17 PM
> > To: mac
> > Cc: cisco-nsp at puck.nether.net
> > Subject: Re: [nsp] HSRP
> >
> >
> > On Thu, Jun 19, 2003 at 08:29:37PM +0200, mac wrote:
> >
> > > can one give me a quick explanation of HSRP is and how works?
> >
> > Sure. HSRP, Hot Standby Router Protocol.
> >
> > The concept is that you have two routers connected to the same Layer 2
> > network, say 10.1.0.2/24 and 10.1.0.3/24.
> >
> > You configure the hosts on the net to use 10.1.0.1 as their default
> > gateway and you then configure HSRP with a virtual IP address of 10.1.0.1.
> >
> > The routers then decide amongst each other which of them is to be 10.1.0.1
> > now and the hosts use that router.
> >
> > Should that router die a horrible death (or should you decide to shutdown
> > the interface, yank the wire or somesuch) the other router notices and
> > quickly takes over the IP address and (as far as I remember) the MAC
> address
> > so the ARP tables doesn't make a fuss.
> >
> > You can set up priorities for the routers and you can ask them to
> pre-empt,
> > so that when the first router is back on line, it takes over again.
> >
> > It works a charm to my experience.
> >
> > (It is, afaik, the Cisco version of that which became VRRP. However, VRRP
> > uses unicast for checking the status whereas HSRP uses multicast. Please,
> > someone correct me if that's pure nonsense.)
> >
> > Peter B. Juul,
> > Uni·C (PBJ255-RIPE)
> > _______________________________________________
> > cisco-nsp mailing list  cisco-nsp at puck.nether.net
> > http://puck.nether.net/mailman/listinfo/cisco-nsp
> > archive at http://puck.nether.net/pipermail/cisco-nsp/
> >
> > _______________________________________________
> > cisco-nsp mailing list  cisco-nsp at puck.nether.net
> > http://puck.nether.net/mailman/listinfo/cisco-nsp
> > archive at http://puck.nether.net/pipermail/cisco-nsp/
> >
> >
>
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> http://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>