Using the current splitfile spec, given that GJ eventually implements
"streaming within a segment"...
How many blocks would be needed for the first output chunk?
=> what would the startup time be
Would that be any particular blocks?
What would the granularity be? Can we stream within a block as well as
within a segment?

Separately:
we need the mainport options to be configurable, and documented, in the
config file. One way to do this is just to make them regular
Params/Config options. Any other suggestions?

Also:
When/how can we make splitfiles more transparent? Without sub-segment
streaming, this seems unreasonable except maybe for really small files,
and files that we have already downloaded... Keeping a list of recent
successful splitfile downloads and doing those transparently might be
helpful. Once we have sub-segment streaming, we could have a list of
MIME types that we always stream, or at least a per-type size limit.

Finally:
Splitfiles are the basis for many killer apps on freenet. Unfortunately
they are slow. The main reason for this is the low number of threads
used to request and insert by default. We cannot simply increase the
defaults, because then multiple simultaneous downloads will overload the
node. So we need to determine a reasonable maximum number of requests
for fproxy to load the node with (my preferred total guesstimate is 20%
of maxThreads), and dynamically allocate them according to usage to
"normal" downloads (and uploads), and FEC downloads (and uploads).

Any comment on any of this is welcome and solicited.
-- 
Matthew Toseland
toad at amphibian.dyndns.org
amphibian at users.sourceforge.net
Freenet/Coldstore open source hacker.
Employed full time by Freenet Project Inc. from 11/9/02 to 11/1/03
http://freenetproject.org/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: 
<https://emu.freenetproject.org/pipermail/devl/attachments/20021122/45725729/attachment.pgp>

Reply via email to