I was participating in a storage design discussion and a Storage person threw up their most common complaint about networking – “we don’t know how to build reliable networks”. Let me take a shot at that.
Between networking and storage, there is a large difference in scale and investment. The Storage markets have been able to make substantial investments ( by networking standards, excessive amounts of money) in high scale equipment and redundancy, possibly because the business case was easy to make. AKA if the disk stop working, everything stops working ergo give me lots of money to stop that from happening.
Networking has a thirty year pedigreee where near enough was usually good enough and this attitude is what many of us have inherited today. It is a historical and environmental reality rather than a shortcoming of our own making. In my experience, Management still feels that a short outage on the data side is acceptable. Until that attitude changes and the investment matches the reality this will remain true.
For example, when I tell management that building a five sigma network solution will cost 500K, their eyes bug out and they have a fit, but when the Storage folks pops up with a 1.2 million for equivalent benefits, they don’t even flinch.
Meet HA Networking experts
If you want to see really top notch networking, meet Data Centre Networking people fully focussed on delivering very high quality solutions. Get a high level Data Centre expert in your storage team and you should notice a real difference in attitude and delivery.
Going forward, Data Centre networking is going to become an expert field in its own right (if it isn’t already)
FC networks are often quite small
Also, remember that FC networks are quite small, compared to Ethernet networks. A couple of hundred FC ports is not a big network compared to a DCE that has thousands. Complexity, in my experience, increases exponentially with size.
I feel that part of FC’s eventual demise will be its inability to scale beyond several hundred ports.