2010+ networking software: how to open up, how to speed new ideas to market
The onePK announcement Ric describes in the previous blog entry is game changing. It also intersects a trend which has gone fairly unnoticed in the networking standards areas. The importance of new standards is declining relative to advances in software and hardware.
Don’t get me wrong, standards are still very important. But you should not look for a plethora of new foundational IP networking standards anytime soon. The decade of 2010 will not be groundbreaking from a standards perspective relative to the 1980s and 1990s when concepts like TCP, Ethernet, BGP, ICMP, DHCP, and IPv6 were solidifying.
But networking continues to evolve quickly. Upcoming fundamental changes will be driven from computer industry innovations such as:
- Multicore CPU and process multithreading
- Virtual Machines and Linux Containers
- Open Source, 3rd party application, and autonomic integration
Intersecting these drivers gives some guidance on what to expect from the 2010s:
- Being able to leverage Multicore means that routers must continually support more threads/processes.
- Linux Containers will enable application processes (e.g., Open Source) to be safely isolated from critical infrastructure (e.g., forwarding or route topology calculation).
- Moore’s law means off-the-shelf CPU intensive applications can be brought into the network
- If these applications are run within a Linux Container, they can be spun up without requiring a full suite of network element regression testing and recertification.
The deadliest barrier to networking innovation is the long lead time from concept to embodiment. So when people say “speed kills”, think “slow speed kills good ideas”. All of the drivers described above remove real barriers slowing an idea’s time to market. Once an emerging idea has been vetted, morphed, and validated by the market, then it will be time to consider standardization. (Food for thought: don’t be surprised if the forums for standardization shift into code dominated areas involved with Linux and Open Source.)
The onePK announcement is a huge step to increasing the speed and flexibility for new applications entering the network. Complimentary advances in software and hardware will remove the majority of boundaries slowing the realization of the next wave of networking advances. Without tools like onePK, emerging applications will continue to congregate on servers attached to routers and switches. And operational costs and complexities will keep many worthy ideas from being realized.