Packet watermarks were introduced to address these issues in the early 2000s; but, Cisco and Juniper generally promoted the notion of "free" bandwidth and bandwidth allocation schemes limited to coarse estimates (MPLS, etc.). Packet watermarks also enable differential QoS at far higher efficiency in computational overhead and better management of "flow". The other push comes from search and poor notions of where the network (*your* network / infrastructure) starts. Innovation suffers when legacy models are not subject to "restart"!
The problem with the current network address approach is that it takes each packet at face value (This one says it came from google.com? Good enough for me!). The idea of integrating authentication is a real step forward, whether it is done in a content-centric approach or an overlay onto the current structure. This seems to be a clean-sheet-of-paper approach, which has technical advantages and huge logistical challenges. Step one: throw away all of your infrastructure. Step two: doesn't matter until you get people to do step one.
Content-centric networks should be a natural for any company to use for all their product design activities. The PARC announcement also strongly suggests the re-discovery of the use of hashed binary trees, which are the strongest data structures (read also of course files, networks, clouds, etc.), capable of organizing and providing immense amounts information, especially for hierarchical product and library design information. I believe you will find that NSA is also using this basic data and network paradigm for their huge huge collections of information. My own background here is that this is the approach we took at Honeywell Large Information Systems EDA hierarchical design data using what Honeywell called indexed-sequential files. We implemented a EDA database we called MUSER which was actively used by Honeywell / Bull from 1969 through 1997, and was only shut down due to the french Bull folks having to use some many off the shelf design tools. --Steve Grout
Blog Doing Math in FPGAs Tom Burke 2 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...