[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: making psamp export congestion-aware



The use of UCP/TCP/SCTP doent make a big difference when a service provider
exports to a separate network for network management.But,sometimes the 
non-local network used for network management will become a hindrance as
the SP network grows leading to surging export cost .Also,sometimes a 
combination of TCP and increasing gap between exporter and collector(because 
of non-local management takes extra buffer and unusual fail over schemes.
So,I think use of separate network for management is itself makes things
complex leave alone use of TCP/SCTP.

-Senthil



>* In general I think the concept is great of allowing a protocol like UDP to
>be used with a higher level congestion aware protocol to handle the backoff.
>However, from past conversations (with Randy) on this topic my understanding
>is that the goal of IETF's mandate in this area is to push protocols like
>TCP/SCTP/etc. since it protects the internet from service providers that
>might otherwise blast out UDP.  Proposals within IPFIX context to use rate
>limiting were dismissed since it was too easy for service providers to
>mis-configure that or to never enable it altogether.  The final comment was
>that there might cases where a service provider might desire a stateless
>protocol like UDP within their network (maybe it is an internal network for
>mgmt data) but that IETF does not standardize protocols for that
>environment.
>	-- Doesn't this proposal suffer in that it is to easy for service providers
>to not enable your mechanism and therefore end up getting to happily use UDP
>without any controls forced on them ?
>
>* In the past I have heard many service providers indicate they really don't
>want boxes like a collector changing configuration of the router through
>MIB/CLI especially if they are designed by 3rd party vendors.  Do you see
>this as a concern with this proposal ?
>
>Will


--
to unsubscribe send a message to psamp-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/psamp/>