Drawing a bead on network complexity - today's scale barrier
According to Stu Bailey we're now passing through a unique time in the networking market. Complexity, he says, is the scale barrier to network development and growth, just as bandwidth (or the lack of it) was the scale barrier 10 to 20 years ago.
Once this is recognised, he says, the industry will be able to get its arms around the problem and start quantifying complexity and factoring in the cost of reducing it when investment, equipment or architectural decisions must be made.
Clearly 'complexity' and its measurement is significant in the development of SDN/NFV and the resulting network automation, since complexity is the thing that these developments are designed to mitigate.
Bailey has been chewing on this issue for some time and with his long-time research partner, Prof. Robert Grossman of the University of Chicago has developed a formula for complexity "that accounts for the number of endpoints on a network and how they interact, rather than the traditional approach of simply counting network infrastructure and the wires that connect them."
The resulting tool, Tapestry will be a free download from the FlowForwarding.Org project and is expected to be available next month.
The Tapestry software will run on any computer and generates a Network Complexity Index (NCI) number based on endpoint interaction data from network-wide control systems such as the Domain Name System (DNS).
Tapestry is an SDN application and will run on a free, open-source SDN control plane from FlowForwarding.org called Loom. Loom in turn controls SDN white boxes, built on inexpensive programmable Ethernet processors and available for as little as $300, to collect the NCI data. Bailey says the white boxes can be deployed in front of a network’s DNS servers without disrupting existing infrastructure or operations.
So does Bailey want to develop an industry-standard for complexity measurement so positioning Infoblox front and centre in the SDN field?
"Actually, I don't really care whether it's our software or something else. If this kicks off a process, and people to come up with an even better method and algorithm than the one we've come up with, then that would be great. The important thing is that we start the conversation," said Bailey.