Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. And see here for current known issues.

Bronze

network resiliency or high availability

Hi fellows,

I am trying to implement a solution by which my company can achieve 100% network availability, the scenario is as following.

we have two internal routers 3361 running IOS 12.2 (26) as backup and 12.3(11)T as active with EIGRP + many static routes to our partner networks and remote branches connecting through either TDM or MPLS connections.

what I need is that if for instance one the router fails due to any possible reason HW or SW, the other router should immediately be able to route all the traffic with any downtime.

In my head I have two options to work on

1. IP SLA

2. HSRP

if there is any other way of achieving the same or if i use any one of the above mentioned method which should be the most preferred one and likely to achieve the desired results.

7 REPLIES

Re: network resiliency or high availability

upon ur case

i can see u have routing between routers and it is a routed network

so u dont need HSRP becuase if u count on routing protocols redundancy better that HSRP in regard to convergence time

just u need to make sure that each router in ur network has multiple path to be reach from others through routing protocols or static route

and through rouign u can achive which one will be the primary and which will be the fall back

also with eigrp u can do load balncing and if one gos down the other one will handel the whole network load

by the way,, with routing u cant make neighboring with VIP i mean virtual ip address such as the one used by HSRP u need to making the peering with peer itself

HSRP will be useful if uhave clients or for example servers and they use those routers as a exit gateway then u can use HSRP or GLBP to achive high availablity...

good luck

please, if helpful rate

Bronze

Re: network resiliency or high availability

many thanks for the valuable input,

I have gone through what you have said and what I had in my mind and if I am not wrong HSRP and GLBP are very similar to each other except that GLBP has load balancing capibilities. Addtional to that, while using HSRP I would have to configure my entire LAN clients with the new default gateway IP (which will be virtual); whereas while using GLBP I didn't recall reading such configuration. Please correct me if I am not wrong.

Because changing the default gateway on all clients (static or DHCP) is a big administrative task espacially if you are taking large number.

hoping for a reply

Hall of Fame Super Silver

Re: network resiliency or high availability

Hello Zulqurnain,

you can change the ip addresses used by the two routers and the current default-gateway will be defined as the HSRP VIP or GLBP VIP (GLBP also uses a Virtual IP) they need to be all on the same subnet so this is an easy step.

Then you need to use the tracking options of GLBP or HSRP so that the correct reaction is performed in case of failure or restore.

see:

http://www.cisco.com/en/US/docs/ios/ipapp/configuration/guide/ipapp_glbp.html

Hope to help

Giuseppe

Bronze

Re: network resiliency or high availability

hello anybody ?

Re: network resiliency or high availability

Routing protocol on the wan side will give you what you are looking for - load balanced links that upon possible failure of linkA all traffic will go over to linkB .... Assuming that you have n+1 routers at remote sites :)

While GLBP is a great feature it does not provide load balancing as a 1,2,1,2,1,2 ....

Rather server #1 picks GW#1, server #2 picks GW#2, server #3 picks GW#1 It is possible that if server #2 generates 4 times as much traffic as #1 you will have uneven loads on the outbound links.

If your internal servers are NIC-teamed to n+1 L3 switches (or 3750s /65xx_w_VSS) you can reduce this issue by routing between servers switches and the wan routers.

Super Bronze

Re: network resiliency or high availability

Avoiding "any downtime" can be difficult and expensive to achieve, although minimal downtime, as in the subsecond range, is often good enough. Almost zero downtime might require routers that support NSF and have hardware redundancy.

IP SLA would be best for use if you intend to retain you static routes. As the other posters have mentioned, if you can move from static routing to dynamic routing, dynamic routing should handle loss of a path.

HSRP is one of the methods to provide a virtual gateway address for hosts. As is GLBP, also mentioned by another poster. There's also VRRP. (If you move to any of these, you can often re-use your existing gateway IP, but the MAC usually changes and these can confuse some hosts until they ARP for the new gateway MAC. [I've also seen some devices get somewhat confused by GLBP.])

Virtual gateways address network resiliency for hosts. Dynamic routing or IP SLAs, address routing resiliency. Different aspects of total network resiliency, but often used together.

Another method to handle host gateway resiliency, would be to have the router provide proxy gateways (although I wouldn't recommend this).

Another method to handle router path resiliency, would be usage of OER/PfR. These are not supported with EIGRP, but static routing is. However, feature only available with later IOSs and not available on all platforms.

New Member

Re: network resiliency or high availability

You cannot achieve 100% uptime. You can try it.

You must consider all aspects and not only the network aspects.

What about power outage?

What about some catastrophic event?

What about personnel fault?

What about other?

For network first hop redundancy you have: HSRP, VRRP, GLBP

You can use Routing protocol in the Access layer.

Or you can use VSS but you need other equipment for that.

Static routers are ok until you don't have more paths of how to reach the branches and you have only one leaving point for your downstream.

347
Views
0
Helpful
7
Replies
CreatePlease to create content