On Mon, Nov 30, 2020 at 06:36:16PM -0500, Michael Richardson wrote: > > Brian E Carpenter <[email protected]> wrote: > > Perhaps there is something we should specify in ANIMA to prevent the > > ANIMA infrastructure falling into this sort of trap: when there is a > > system-wide issue (such as hitting an O/S resource limit everywhere at > > the same time) it also prevents the autonomic mechanisms from working. > > I think you mean, that it should not also prevent the autonomic system? > I think that key is: > 1) allocate resources (including threads) up-front > 2) do not dynamically allocate threads per interface, but rather use async > routines.
Reminds me of university times in the 80th. In my university, they had built a network layer packet accounting system (midle box like the firewall, i let you guess the network layer protocol). Using: https://en.wikipedia.org/wiki/PEARL_(programming_language) I took a semester programming PEARL. No dynamic resource management. Pain in undisclosed body parts. But gets you into the mood to think about how to handle fixed constraints. Didn't help to avoid buggy software though. One month it claimed i had used 60,000 German Marks in network resources. Turned out it was "only" 6,000 though. Programming bug related to insufficient testing with corner situations (high numbers). I thought Facebook praised itself by creating regularily "A Series of Unfortunate Events" into its software to ensure the system is continuously hardened. > But, I don't know what we could write into the specification to make this > happen. It seems that we really just need smart implementers. Unless there is documentation for what constitutes "smart", you can only count on "experienced". Oh wait: Networking is a commodity. "Experienced" is too expensive for our industry. Cheers Toerless > -- > Michael Richardson <[email protected]> . o O ( IPv6 IøT consulting ) > Sandelman Software Works Inc, Ottawa and Worldwide > > > > > _______________________________________________ > Anima mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/anima -- --- [email protected] _______________________________________________ Anima mailing list [email protected] https://www.ietf.org/mailman/listinfo/anima
