Time for the Infiniband group to move out of the dark ages.
The type of environments that could make the best use of Infiniband are exactly the environments most likely to use huge amounts of OSS.
Given that so many problems in that space today can be solved with standard Copper Gigabit, and that fiber 10 gigabit is continually getting cheaper I do not see how this group hopes to see IBT can see a future for their technology.
Sure, there are situations where infiniband is superior to Ethernet, but those spaces will continue to shrink, just like commodity clusters have shrunk the space of problems that absolutely need a traditional vector supercomputer.
Open, low per unit cost solutions dominate cluster computing. The IBT is unecessarily limiting themselves to niche where they could show leadership and grab hold of a market ready for higher interconnect speeds with reasonable terms.
It is a real shame when good technology has to be ignored because of archaic and narrow minded limitations of licenses prevent wide spread distribution.
The IBT would do well to change its policies to support OSS implementations of their technology, they have everything to gain and nothing to loose.
Copyright © 2017, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds