Bundling and redundency

I'm not sure what your question is? Sounds like you need to go back to your provider and ask them if you can get another link for redundancy, or go to another provider if you really want redundancy in case one provider cuts their own fiber. You won't be able to avoid all hops, but two providers should be able to get you most of the way there.

After re-reading your submission, do you mean 'bundling' as in etherchannel for doubling your service? If that is the case, and presuming these are layer 3 routed links and not switched (else it would be 100 meg or gig), you would accomplish this through routing and ensuring you have equal cost load balancing. This means terminating to two different routers, and each of those routers participates in the same routing protocol back to your core, which then has equal cost paths to the destination. Let me know if I have misunderstood.

Have you asked your service provider? Some service providers [eg BT in the UK] can provide a diversely routed circuit for an additional fee. You may, however, find that the extra fee is greater than getting another circuit from a different provider, which in theory would provide greater redundancy.

You will have to implement BGP routing. I have a similar environment where I use a 6 MEG pipe as my primary link via a ethernet connection. My alt side is just a 1.5 meg connection as backup only and it is serial port connection to on my router. There are many different flavors of this that you could consider.

Here is one way of providing redundancy and load balancing:

|-----------Router--------| Service Provider 1 |---------Router------| Corporate Remote Site |-----------Router--------| Service Provider 2 |---------Router------|

Routing protocol (e.g. EIGRP) provides load balaning and quick failover. The routing have to be properly designed- for instance, the resources have to be equal cost from both routers so that one route is not preferred over other. Similarly it is ideal to have same bandwidth on both links.

If budget is prohibitive, you can opt for one router two links, use variance to control different bandwidths etc.

Good luck

I agree the above posts. Your best bet is to go back to your provider and either verify they are providing the resilient infrastructure or ask them to provide you a solution that will. Keep in mind to ask them at what layers is this resilient/redundant. Just because you have layer 3 in the cloud doesn't mean you'll have layer 1 between you and the CO.

There are many factors that come into play considering what kind of a budget you have and how redundant you want to go, since there are potentially many points of failure. Considering the connection, your best bet would be to get service from a completely different provider. If your single provider gives you two separate fiber services, but their fiber gets cut a 1/2 mile up the road - both of your lines will be down - all that extra money for nothing. Have a separate provider bring in service and make sure they are coming from a different path as well as providing them a separate entrance into your building. Might be overkill, you just need to consider what down time will cost your business. Then, as has been mentioned in this thread, you can handle the redundancy yourself through routing.

Hope that helps,

Jim

You don't have redundancy unless your second circuit is from another vendor (and both vendors aren't selling the same service). If the carrier has a problem on their network and both links go down, well you get the point... Remember EVERYTHING must be redundant and on different carriers, including local loops. Of course sometimes you can't eliminate all points of failure especially if you only have a single carrier entrance into your buildings.

the assumption you make with this is that carriers do not provide "real" resilience - some do.

however you are right in that lots of carriers use others for some plumbing (and in turn another subcontractor might be involved) and even the carrier who has access to the info on the service build can have problems tracing how and where things get mapped to a common link, fibre etc.

If the carrier has a

which is when you know the resilience isnt there :) - which is why i suggest that someone test resilience if possible before they need it.

agreed.

The unambiguous way is often to use 2 different access loops that have to be different - point to point wireless and fibre is 1 that comes up sometimes.

FWIW 100 Mbps on microwave is now fairly cheap equipment (at least for a carrier service) and good for 10's of Km links, given a clear line of sight.

Just to try to put your options into perspective, redundancy is merely one approach to better availability. As has been pointed out, what you need to evaluate is just how important availability is to your business. Until you know what downtime costs, it is merely guesswork trying to determine when the cost of additional availability improvements is no longer cost effective. Keep in mind that 100% availability, like 100% security, is an impossible objective. You can have infinite redundancy between you and your service provider, but if a fire wipes out your data center, you are off the air. Ditto if you are unable to cope with a distributed denial of service attack. Etc. etc.

You will typically find that a second link to a second POP of your ISP will provide the biggest (and cheapest) payoff. Adding multiple ISPs and BGP gives another increment. Attention to network monitoring and configuration management also provides a major boost. Adding redundancy only works if it is implemented correctly. I've dealt with numerous "redundant" networks where the added redundancy only contributed to cost and in some cases wound of reducing the overall availability. Remember that for redundancy to help, you need three things, the ability to detect a problem, the ability to use the redundant equipment to restore service, and diversity so that whatever killed the one path does not also kill the backup path.

Good luck and have fun!