>At 06:07 PM 3/6/01, NetEng wrote:
>>Does HSRP work at the interface level or is the entire router on
>>acvtive/stand-by? In other words, if I have two routers working in HSRP and
>>a link goes down somewhere down the line, will the first router know to
>>fail-over to the second router (with a good link)?
>
>Interesting question. The first router would have to lose its connectivity
>to the second router. Routers that are running HSRP send and receive
>multicast UDP-based hello packets to detect router failure and to designate
>active and standby routers. HSRP detects when the designated active router
>fails because of the lack of hello packets, at which point a selected
>standby router assumes control of the Hot Standby group's MAC and IP
>addresses. A new standby router is also selected at that time.
>
>Remember HSRP stands for Hot Standby Router Protocol, not Hot Standby
>Routing protocol. It's the default router for LAN devices that's on
>standby. If you think of HSRP as a routing protocol, then you will tend to
>think it does more than it does. I think to solve your problem you need a
>"real" routing protocol, although without more info, it's hard to say for
>sure.
>
>Priscilla
>
>> I have one router
>>connected to one ISP and a second router connected to a second ISP. Can
>>these routers be run in HSRP or must they be running in parallel and let a
>>dynamic routing protocol (BGP on the outside and let's say EIGRP on the
> >inside) decide? TIA.
As Priscilla points out, HSRP is an interesting protocol that doesn't
easily fit into any architectural model. I've long advocated more
general use of the ATM architecture's split of protocols into:
-- user plane. Transfer user information (e.g., IP, FTP)
-- control plane: information from end host to first-hop router or switch
(e.g., Q.931, Q.2931, ICMP, IGMP). Can also operate between routers,
but there's a
distinct client/server aspect rather than the peer model of generally
accepted routing protocols
-- management plane: information between internal network elements.
originally SS7, PNNI, etc., but OSPF, ISIS, RIP, BGP, etc. fit very
nicely.
I started to call HSRP a control plane protocol, but that doesn't quite fit.
While it operates on the edge of the network, it is symmetric between edge
routers, and does not involve any protocol interaction with end hosts.
My thinking is that HSRP is a management protocol of very tightly defined
scope. "Scope" is a non-intuitive concept for most beginners, who intuitively
feel the more information available, the better. Scoping, however, is an
essential part of scalable networks.
In OSPF, for example, OSPF hellos have a scope of a single subnet: they don't
propagate beyond the local medium. Type 1 LSAs (router LSA) have a
scope of a single area. In the absence of summarization and
stubbiness, type 2 through 5
LSAs have a scope of the entire OSPF routing domain.
So I would call HSRP a management protocol of single medium scope,
with the additional constraint that this is intended as an edge
medium. The scope of its knowledge (as distinct from the scope of
its propagation) includes the interfaces of the routers in an HSRP
group, but not the routing tables of those routers. Interface
tracking fits into this model, but knowing whether a given ISP is
reachable over an interface is a function beyond the HSRP scope.
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
