[Dyncast] Information on use of compute information in L7 vs L3/L3.5 solutions

[Dirk Trossen <[e-mail]>]

Dear all,

As additional information to the CAN gap analysis and requirements draft (https://www.ietf.org/archive/id/draft-liu-can-gap-reqs-00.html), our paper at the recent ACM SIGCOMM workshop on Future of Internet Routing & Addressing (FIRA) (available at https://dl.acm.org/doi/abs/10.1145/3527974.3545721) may provide some insights into the use of a compute aware traffic steering solution in architectures relying on an indirection approach (e.g., DNS, GSLB, ALTO) vs those provide in-band discovery capabilities (such as those advocated by CAN).

The paper illustrates the two systems-at-test and compare performance for crucial aspects in service delivery, such as initial request latency, as well as performs a use case driven comparison to shed light on latency as well as latency variance reduction but also resilience. The finding of this paper may be useful for this community.

The paper will also be presented in a planned sidemeeting at the IETF on Wednesday the 9th of November, starting at around 6.15pm; see https://wiki.ietf.org/meeting/115/sidemeetings for more info and the announcements on a few IETF lists.

Best,

Dirk