本文发表在 rolia.net 枫下论坛Similar thinking has been there from the beginning of Internet - where the intelligence should reside - on the network side, or on the server side... now they replaced "server" with "application"..
This is similar to the idea of enterprise message bus - to intelligently directing packets (rather than messages) to the right place dynamically based on time of the day, real time traffic situation, priority of the packets, etc etc...from a control brain outside of the network - meaning from different applications...comparing to today's easy solution - the network does not care, and does not have an end to end view, either geographically, or from the seven layers' perspective. Today's router and switches just forward the packets based on pre-defined forwarding table as fast as they can, with pre-defined QoS policies...
In my view, overly complicated technologies are hard to survive - as nobody can understand it and thus no one will use it. This programmable network is an interesting enough idea...just i can't see how it can be implemented in a large scale... maybe within a corporate network where security and end to end information are easier to obtain - otherwise, who will win the war to forward the packets faster than others based on what criteria ?
This is a good market strategy for Juniper to differentiate from Cisco's SONA... but at the end of the day, it may attract eye balls, but by no means it will become a reality any time soon..更多精彩文章及讨论,请光临枫下论坛 rolia.net
This is similar to the idea of enterprise message bus - to intelligently directing packets (rather than messages) to the right place dynamically based on time of the day, real time traffic situation, priority of the packets, etc etc...from a control brain outside of the network - meaning from different applications...comparing to today's easy solution - the network does not care, and does not have an end to end view, either geographically, or from the seven layers' perspective. Today's router and switches just forward the packets based on pre-defined forwarding table as fast as they can, with pre-defined QoS policies...
In my view, overly complicated technologies are hard to survive - as nobody can understand it and thus no one will use it. This programmable network is an interesting enough idea...just i can't see how it can be implemented in a large scale... maybe within a corporate network where security and end to end information are easier to obtain - otherwise, who will win the war to forward the packets faster than others based on what criteria ?
This is a good market strategy for Juniper to differentiate from Cisco's SONA... but at the end of the day, it may attract eye balls, but by no means it will become a reality any time soon..更多精彩文章及讨论,请光临枫下论坛 rolia.net