To remove any confusion caused by the apparent contradiction between
the answers that Randy and I gave:

- It's ok to have connections open to more than one cache server at a
  time, but if you're going to do that, you have to buffer the results
  for each cache separately.

- It's -not- ok to -use- the data from more than one cache server at a
  time, attempt to combine the data from multiple cache servers in any
  way, or otherwise get clever about comparing data from multiple
  cache servers.

- The only point of allowing multiple connections is to allow fast
  transfer to a new cache if the cache one has been using dies.  That
  is: we're trying to avoid the situation where you realize that the
  cache you're using has died, and it takes several minutes to get a
  full load from an alternate cache.

- The main way of detecting that the cache you're using has died is
  the (required) periodic poll; if other mechanisms happen to give you
  a hint sooner, fine, but that's the real check.

- This is all covered in the draft, albeit somewhat tersely.  If you
  think it's unclear, please send text.
_______________________________________________
sidr mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/sidr

Reply via email to