Re: [Skunkworks] Redundancy across Data Centres networks.

Hi Skunks,
Thinking of a way of giving a redundancies across NOC. I.e If I terminates several links to one location ( ISP Style) with different providers on p2p. Like maybe using the appropriate connection, safaricom Wimax, KDN Fibre, Jamii Fibre, AK Fibre/ Canopy, Kenstream serial/E1 connection name them depending where the customer is. Am looking to a way, where I can give a backup, such that NOC1 goes down, NOC2 takes up. Note one requirement is the NOC should be 8 or more Kilometers apart. NOC meaning, where all terminates before going out.
Not sure if your questions relate only to the traffic between the data centre alone, all all traffic including the incoming traffic from customers. For the former, I know only of BGP, but if your data centre can operate independently of each other, there is two ways you can go about this: You can run a BGP session between the two data centre. Its more involving in that you need to have a big IP block so that they can allocate you a AS number, a reliable link between the two data centre and good network know how. You need three ISP, one into data centre 1 , another into data centre 2 and one between them. Its petty resilient though when well implemented in that you likely to survive most network outage, but boy, its expensive and in my opinion hard to maintain if you are a small team The method I would recommend is using some DNS companies that has fail over capabilities. I have used Neustar and they are amazing. No change to your existing infrastructure except you will have to outsource DNS services :( In normal operation, they load balance your traffic between data centres. If you have an outage, they fail over the traffic to the working data centre immediately. You select when the fail over kicks in. By that I mean, they have probes all across the globes. You can for example set it so that fail over take place only when 2 or 3 probes can not reach your systems. This ovoid fail over when upfront ISP have transient issues, like convergence after route change. Unlike BGP, current sessions do die, but you have a far simpler solution. I am assuming you can live with that http://www.neustar.biz/enterprise/dns-services/external-dns-advanced-service...
Regards,
Edwin Ngige
William
254 722 841853
participants (1)
-
William Muriithi