Blue Lang, on January 21. 2001, wrote:
: On Sun, 21 Jan 2001 [EMAIL PROTECTED] wrote:
:
: > > and whether the server is running OpenSSH or SSH.com. Thus, the
: > > only thing that would be seen on the wire is the encrypted
: > > challange.
: >
: > may be it's not a concern in this case, but won't the session be
: > vulnerable to session hijacking (post-authentication) if encryption is
: > turned off?
:
: eh, it doesn't actually solve the performance problem anyway.. as the
: original poster pointed out, it's not CPU that's at issue, it's latency in
: the transmissions. i'm doing a bunch of tests and trying my hand at a
: little hacking on it - will post results later.
The latency is because of the protocol:
client server
READ request \ |
\ | t
\ | i
> DATA | m
/ | e
/ |
/ |
Writes the data |
to disk, sends \/
a new READ request
Because the requests are not sent in advance, the client is in I/O
wait, waiting for an event to occur in the socket.
I'm currently writing code to fix this problem, so that we would send
multiple requests ahead of time.
Cheers,
--
[[EMAIL PROTECTED] -- Sami J. Lehtinen -- [EMAIL PROTECTED]]
[work:+358 9 85657425][gsm:+358 50 5170 258][http://www.iki.fi/~sjl]
[SSH Communications Security Corp http://www.ssh.com/]
