Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

We currently host all of our clients in one datacenter. We would like to expand to two new datacenters by the end of the year. My job is to figure out how to extend our current network to those two locations.

I have some basic designs drawn up (site to site VPNs between the datacenters, OSPF for internal routing, round-robin DNS to distribute the load), but I could really use some advice from someone who has done this before. My major concern is making a design choice that will constrain us or require a redesign in the future. We already have some design choices that will hamper us in the future (e.g. every client gets their own 10.x.x.x/24 subnet with their own VLAN, which will work great right up until our 4093rd client).

Does anyone know some good resources on how to create a scalable network design?

EDIT: Our business resembles VPS hosting, so as you can imagine, there's all types of traffic. Most of it is web (80,443) and mail though, so throughput is the usual priority followed by latency.

The business reason we want to expand is so our clients can host their servers either on the East/West Coast US or South Pacific Asia. The IT reason is to give us a disaster recovery site in case of natural disaster.

EDIT 2: Just to be clear I'm looking for resources on how to design the network. The details on how to do it (VPNs, routing, DNS, etc.) I can do.

share|improve this question
    
I hate to state the obvious... but "This is your Job. Not mine." As for scalable network designs... 80% depends on how you define "scaleable" and 19.9999% depends on what services you're offering. The remaining fraction of a fraction of a percent is what you can fit into a diagram. I can't even begin to guess at all the variables you have brought to the table. –  TheCompWiz Jun 6 '11 at 21:01
    
Well, I'm not asking anyone to design the network for me (at least not for free). Also, I figured this was an old problem with a lot of different solutions to pick from. –  zippy Jun 6 '11 at 21:27
    
Without more specific details about what kinds of services you're trying to scale, and how you are defining scaleable... you might have well asked "how can I build the entire universe?" –  TheCompWiz Jun 6 '11 at 21:30
    
802.1ad (aka QinQ) is an extension to 802.1q tagging to allow up to ~16M vLANs. That should keep you busy for a while. –  Chris S Jun 7 '11 at 2:50
    
Updated my question w/ some details about the traffic type and reasoning for expanding. As for defining scalable, I'm not sure what you're looking for. –  zippy Jun 7 '11 at 17:16

1 Answer 1

Sun used to push a model that they called the Service Delivery Network Architecture.

This won't answer all of your questions (it's an architecture, not a how-to guide) but you may find it useful.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.