Talk:Potato routing: Difference between revisions
Jump to navigation
Jump to search
imported>Howard C. Berkowitz (New page: {{subpages}}) |
imported>Howard C. Berkowitz (→Some of my published observations -- conflict of interest?: new section) |
||
| Line 1: | Line 1: | ||
{{subpages}} | {{subpages}} | ||
== Some of my published observations -- conflict of interest? == | |||
In ''Building Service Provider Networks'' (Wiley, 2002), I wrote (pp. 355-358 deals generally with potatoes and routing): | |||
:"Routing protocol design follows Darwinian principles. The first priority is survival — of the local router and of the routing system as a whole. Just as the first priority of hippopotamus reproduction is for the hippos to look good to other hippos, the first priority of routers is they work well with other routers. | |||
:Optimal routing is not the first priority in making a robust riytubg system. Indeed route optimality may mean different things at different time. For a given application, mimimizing latecy is optimal. For a different application, maximizing throughput is optimal. Survival often means maximizing closest-exit routing and minimizing routing table churn. The latter depends significantly on maximizing route aggregation, which causes a loss of detail. | |||
:In any hierarchical routing system, there is a very basic issue of routing policy, variously called closest exit versus optimal exit or hot potato versus cold potato..." | |||
and a lot more detail. On page 484, in the specific context of BGP routing where service level agreements are in place, I say: | |||
:"'''The Second Prime Directive''' Cold potato is the most rational strategy for premium services where quality of service is more important than minimizing cost. | |||
Anyone got someone else to cite? I certainly didn't invent the term, but I haven't seen many tutorials about it. [[User:Howard C. Berkowitz|Howard C. Berkowitz]] 23:23, 15 July 2008 (CDT) | |||
Latest revision as of 22:23, 15 July 2008
Some of my published observations -- conflict of interest?
In Building Service Provider Networks (Wiley, 2002), I wrote (pp. 355-358 deals generally with potatoes and routing):
- "Routing protocol design follows Darwinian principles. The first priority is survival — of the local router and of the routing system as a whole. Just as the first priority of hippopotamus reproduction is for the hippos to look good to other hippos, the first priority of routers is they work well with other routers.
- Optimal routing is not the first priority in making a robust riytubg system. Indeed route optimality may mean different things at different time. For a given application, mimimizing latecy is optimal. For a different application, maximizing throughput is optimal. Survival often means maximizing closest-exit routing and minimizing routing table churn. The latter depends significantly on maximizing route aggregation, which causes a loss of detail.
- In any hierarchical routing system, there is a very basic issue of routing policy, variously called closest exit versus optimal exit or hot potato versus cold potato..."
and a lot more detail. On page 484, in the specific context of BGP routing where service level agreements are in place, I say:
- "The Second Prime Directive Cold potato is the most rational strategy for premium services where quality of service is more important than minimizing cost.
Anyone got someone else to cite? I certainly didn't invent the term, but I haven't seen many tutorials about it. Howard C. Berkowitz 23:23, 15 July 2008 (CDT)