 |
The Question is:
I am supporting a VAX 7640 running an "X" application to 5 servers. The
system ran like a dream when the servers were VXT's. Now we are sending
windows to vaxstations and PC's. The CPU went from 130% to 400% and
Buffered I/O went from around 500 to 650
0. I have tested with both DECnet and TCP/IP. DECNet reduces buffered I/O
back to normal. What makes the VXT so special? I have increased buffers
for TCP/IP and cannot find anything that looks out of order. No dynamic
buffer allocation and paging loo
ks ok. I can turn the VXT back on and I don't run out of CPU. Any ideas?
The Answer is :
Rather more information is required on the particular configuration
and environment.
You say you have lowered buffered I/O by switching back from (what?)
to DECnet... From IP?
This may be a result of differences in the capabilities of the basic
protocols -- various VXT terminals used LAT, which was specifically
optimized for on-LAN and non-routed communications. This may be a
difference in the server configuration. This may be a difference in
the X Windows server requirements on the VAXstation and/or on the PC
systems, or a lack of resources on these systems.
Software (EWS) to turn various older VAXstation systems into X Windows
terminals is available on the OpenVMS Freeware, if you wish to consider
using it on those VAXstation systems it is compatable with.
Without rather more details, it is difficult to say what is going on.
The OpenVMS Wizard would first recommend an upgrade to the current
version and ECO for the particular TCP/IP stack -- if this is TCP/IP
Services, that would typically be V4.2 with an ECO kit. If that does
not resolve this, then a call to the Compaq Customer Support Center
would likely be in order, as determining the particular cause will
likely involve rather more research.
 |
|
|
 |
|