SCTP using questions (API etc.)

[Vadim Goncharov]

Hi!

I've seen that FreeBSD 7.0 is a reference implementation for SCTP, which is
a great protocol, good work!

But as I was looking through man pages for new API, I couldn't understand
some issues with high-load servers.

Currently, I am developing client-server application for my needs, and I want
to adjust protocol right now to ease possible transition to SCTP later when
it becomes wider available. Protocol uses TCP, it breaks stream to
variable-sized records (length in header up to 64K) and multiplexes several
"substreams". SCTP can do it for me, it's wonderful, but in practice there
are some questions.

How long can be one particular SCTP message? Can I rely on the fact that it
can be unbounded, e.g. I want to emulate a stream with transfer of
6 Gig-sized file?

Can I use 32-bit PPID field for my own headers/arbitrary values or I must
register that numbers somewhere before using?

Can a message be of zero-length data (only headers) ?

What is the relation between SCTP streams in both directions? Can streams
be opened and closed on-demand, like SSH port forwarding (yet again
multiplexing example) or they are preallocated at connection setup all
together? What is the minimum number of streams application can rely upon
(or it just one stream 0 in general case) ?

Another important questions are related with blocking and non-blocking I/O.

With TCP servers I have two models: in one a thread/process per client
(usually blocking), in other - a single-threaded FSM (Finite-State Machine)
serving all clients using select()/poll()/kqueue()/etc. Both rely on the
usual UNIX way of creading new file descriptor on accept().

My server currently uses FSM model, but question is interesting for both.
I didn't find any words about descriptor duplication or non-blocking I/O
in SCTP man pages.

How can I put request to kernel for a connect, for example, and then sleep
until connect will complete or event in some another descriptor will occur?

How can I put each client to it's fd and then do a kqueue()/kevent() on a
set of those fd's (and other items) ? It is very handy to have this
architecture as kevent() allows to store an arbitrary void* in it's
structure which I can later use to quickly dispatch events.

And, of course, all this usual C10K-problem-solving-TCP-server tricks I want
with basic SCTP SEQPACKET benefits: multiple streams and message record
separation (I don't need other SCTP features currently). Where can I find
answers to these questions, like it was with W.R.Stevens books for TCP ?..

-- 
WBR, Vadim Goncharov. ICQ#166852181       mailto:vadim_nuclight at mail.ru
[Moderator of RU.ANTI-ECOLOGY][FreeBSD][http://antigreen.org][LJ:/nuclight]