Growing Pains–Network Scaling And Maturation
When I signed up with my current employer around 8 years ago, the world was a different place. We did not have much of a budget, so we relied on some creative thinking and duct tape to get the job done. The network was a much simpler place too. Every office had their own local servers and WAN traffic definitely fit a simpler profile. Planning for the future was more targeted at which products provided the most functionality at the lowest (or no) cost.
However, that was then and this is now. Suddenly, we have budgets and business sees value in investing in a cost-center, like IT. The business is growing. More and more traffic is passing over the WAN and the profile has changed drastically. Voice traffic, ICA/RDP, backup needs, and just plain more data is increasing the load on the network. Plus, we have new offices opening in places where getting a static IP address is extremely expensive and not realistic.
The network was designed a decade ago and it is becoming time to remodel so we can continue to grow. The issue becomes how. Which change is going to provide the biggest bang for our buck?
We have the backing and we believe we have made some core infrastructure purchases that allows us to operate right now and grow into the future.
So… where do we go from here?
- Dynamic Routing Protocol: Static routes have been great as we use a hub-and-spoke model for the time being. However, as we expand into China and Europe, the networking design will need to change and routing will become more and more important. Something like EIGRP will probably be the best fit for us.
– Inter-office communications: Some offices have MPLS connections. Others do not… so, their communications travel all over the world to reach other offices. Being able to bring up VPN tunnels on demand to facilitate more point-to-point connectivity is going to be important. Hello DMVPN. Sure, I can create VPN configurations on each office router for other office routers, but that becomes increasingly difficult and monotonous as a new office comes onboard and the branch offices can only handle so many connections.
– Geographic routing blocks: Conceptually, just having geographic regions in the same IP addressing area makes for simpler routing. That means re-IPing a handful of offices to meet the needs.
- Regional hubs: Does it make sense to aggregate core services in a single location or create regional hubs and focus on higher performing hub-to-hub communications. I believe the answer is ‘Yes’!
All of this is really showing me a brief glimpse into the pains and considerations that really need to take place to create a more mature networking environment. Concepts like:
– Router hardware limitation (CPU availability, encryption offloading, etc…)
– Router limitations impact on network throughput
– Resiliency design
– Limitations and Advantages of various routing protocols
– Packets Per Second (versus they typical Mbps style measurement)
– plus much more!
Based on my research, this is the next logical step for us to move. I am sure there are 1 million other ways to go (always open to comments). But, as long as I try to be forward looking and pay more attention to the little details, we should be good to go!
Implementing the options from above sounds like a lot of fun. My inner geek is squee’ing with excitement. Hopefully, these projects will turn into reality and we can begin using a smarter, larger, and more robust network in the near future.