[Info-vax] NFS server configuration

[Sum1]

So, once again I have dived into the world of NFS server performance on 
OpenVMS and find myself gasping for knowledge :)  Hence I again turn to 
the experts within COV...

$tcpip sho ver

  HP TCP/IP Services for OpenVMS Alpha Version V5.7
  on an AlphaServer DS10L 617 MHz running OpenVMS V8.4

Network File System:
  TCPIP$CFS_SHR;1            V5.7-13          21-APR-2010  SYS$COMMON:[SYSLIB]
  TCPIP$DNFSACP;1            V5.7-13          21-APR-2010  SYS$COMMON:[SYSEXE]
  TCPIP$DNFSDRIVER;1         V5.7-13          21-APR-2010  SYS$COMMON:[SYS$LDR]
  TCPIP$DNFS_MOUNT_SHR;1     V5.7-13          21-APR-2010  SYS$COMMON:[SYSLIB]
  TCPIP$LOCKD;1              V5.7-13          21-APR-2010  SYS$COMMON:[SYSEXE]
  TCPIP$MOUNTD;1             V5.7-13          21-APR-2010  SYS$COMMON:[SYSEXE]
  TCPIP$NFS_SERVER;1         V5.7-13          21-APR-2010  SYS$COMMON:[SYSEXE]
  TCPIP$NFS_SERVICES;1       V5.7-13          21-APR-2010  SYS$COMMON:[SYS$LDR]
  TCPIP$NFSSTAT;1            V5.7-13          21-APR-2010  SYS$COMMON:[SYSEXE]
  TCPIP$PCNFSD;1             V5.7-13          21-APR-2010  SYS$COMMON:[SYSEXE]
  TCPIP$STATD;1              V5.7-13          21-APR-2010  SYS$COMMON:[SYSEXE]

Having done a nicel clean install of VMS 8.4 with all patches and 
updates available as of 1-NOV-2013, I find that NFS cleint performance 
accessing an OSX server is still horrendous.  In particular, DNFS1ACP 
still hits and stays at 100% CPU for extended periods but, sincle 
updating from 8.3, the page faulting has stopped.  It just consumes 
100% of CPU and almost everyhting else stops.

Monitoring the OSX NFS server (10.9) statistics, I see that read calls 
increases slowly with access but lookup calls is four times as many and 
ticks over at a much faster rate.  When the VMS server is in 100% 
state, the OSX server lookup calls accelerates and the read calls stop.

Googking for information about how to change the VMS client to stop 
this behavious has produced nothing that seems relevant, however IBM 
and other sites make reference to similar behaviour (excessively high 
lookups instead of reads) as being client configuration related. Great.

What additional configuration information would you like to help 
determine possible improvements in client performance, specifically 
reducing the lookup rate and replacing it with reads?  Or some other 
configuration option....?