ftp speed from hpux to aix box

[zxcv]

Hi ,

i have two boxes..hp and aix

hp 11.23  rx2660 

root #/app_dumps >lanadmin -x 0
Speed = 1000 Full-Duplex.
Autonegotiation = On.

aix 5.3 p-780

drpbkpsrv1:/->lsattr -EH -l ent0 -a media_speed
attribute   value            description           user_settable

media_speed 1000_Full_Duplex Requested media speed True

the actual link is 45 Mbps for which we shuld be able to get 100MB in 20 sec but for this configuration we are getting

164984119 bytes sent in 466.54 seconds (345.35 Kbytes/s)

which is **bleep** slow then expectd..

Had a chat with n/w team they say autoport negn is not enabled on n/w switches.

Also i had made a change in /etc/inetd.conf on both servers ;

ftp stream tcp6 nowait root /usr/lbin/ftpd ftpd -B 16382 -l -i -o

[Stephan.]

If you use autoneg. on the systems it needs to be enabled as well on the switches.

All involved ports needs to be configured to the same setting, otherwise you will have this  **bleep** performance.

[Steven Schweda · Accepted solution]

> i have two boxes..hp and aix [...]

   Connected how?  Doing what, exactly?  (Who's the client?  Who's the
server?  Which way is the transfer going?  ASCII or binary?  As usual,
showing actual commands with their actual output can be more helpful
than vague descriptions or interpretations.)

> the actual link [...]

   So this is some complex scheme, not two boxes on the same LAN
segment?

> [...] we shuld be able to get [...]

   Have you tried any obvious tests, like say, a transfer from/to each
system to/from some local (to it) system, without involving this
(unspecified) "actual link"?  Where's the bottleneck?

> Also i had made a change in /etc/inetd.conf on both servers ;

   Why?  And did that help or hurt?

> Had a chat with n/w team they say autoport negn is not enabled on n/w
> switches.

   Really?  I know nothing, but I thought that gigabit interfaces
_required_ autonegotiation.

[zxcv]

Thanks Steven ,

gigabit interfaces
required autonegotiation solved my problem.