Re: [wsrm] What about this tradeoff solution for the persistence issue?

[Paolo Romano <Paolo.Romano@dis.uniroma1.it>]

Tom Rutt <tom@coastin.com> ha detto:

> Paolo Romano wrote:
>
> >Persistent storage usage is the default mode.
> >An apposite <mustpersist="false"/> tag is used when applications do not require
> >a crash tolerance reliability level, but only tolerance to transport layer
> >failures.
> >
> Is not "guaranteed delivery" alone, without duplicate elim or ordered
> delivery the only thing
> that can be done without persistence?
>
> We should not put in new parameters which are not needed.
>
> Tom Rutt
> Fujitsu
>

I think there is a misunderstanding on the persistence terme. I can claim that
without persisting you can still do duplicate elim and ordered delivery because
I'm assuming that you MUST store the messages (or its metadata) in a non
persistent storage (i.e. RAM, main memory). In fact you can still do reliable
messaging, but the point is that, in this case, you are reliable as long as
there ane no crashes (e.g. power off), but only communication channel failures.

The <mustpersist ...> tag in my mind would be used to indicate that the rm
processor will still store the protocol information state somewhere, in a non
persistent way. The specification should prescribe the same behavior for state
information updates in the case that either a persistent storage or
non-persistent is used. This is why I think that introducing this new feature
would not make our protocol much more complex, as long as we clearly define the
reliability level/reliability quality of service.







--
Paolo Romano