View Single Post
  #11  
Old 01-06-2006, 10:29 AM
tnygren's Avatar
tnygren tnygren is offline
Registered User
 
Join Date: May 2005
Posts: 1,408
Hi Jimserio,

Getting consistent sftp performance has been a problem since
the beginning of sftp. Things have improved, but there are a
number of variables:

- SSH2 Window size
- SFTP packet size
- number of parallel sftp requests

The best performance is achieved when the client and the
server have made similar assumptions. Thus far, we've been
unable to tweak our clients to perform consistently across a
range of various vendors servers. This same obstacle holds
true for our efforts to tweak our VShell server to perform
consistently across a range of various vendor clients.

Our client and server products are using the same basic
assumptions for SSH2 Window size, SFTP packet size, and the
number of parallel SFTP requests.

Some client will allow for changes to these variables to
help improve the performance.

For example, using OpenSSH, it is possible to adjust the
buffer size (-B) and the the outstanding requests (-R) with
the SFTP command to increase file transfer performance.

However, if the network does not have a high latecy, then
increasing the outstanding requests will not have a
significant impact on performance.

Does this help explain why SecureFX has a better transfer
rate then other clients?

As for the line drawing, do any of the suggestions in my
previous post help?

If not, could you also send me a raw log of the connection?
__________________
Thanks,

Teresa

Teresa Nygren
Reply With Quote