Hi
When we were previously using a Sun V880 (with 8Gb memory on Solaris 9), CA
Unicentre monitoring tool reported memory utilization at about 65% (I was told by
a colleague).
Last month we migrated the data & the very same version of oracle10g from this
V880 to a faster Sun T5120 (with 16Gb memory). "df -k" on this T5120 showed
the /swap partition has a total of 20Gb space.
When I reviewed CA Unicentre's report for this T5120 since it went live last month,
the memory utilization is at 98% while the swap space utilization is at 3.2%.
Questions :
a) can I safely say that the T5120 is not slowed by memory constraints as
the swap utilization is low?
b) a question was posed to me by the management : why is the old V880's
memory utilization only 65% while the T5120(running Solaris 10) which
has double the memory was reported to have 98% memory utilization?
c) supposed we plan to add 6Gb more data (there's enough disk space) to
the database, mainly images, will the current memory be sufficient (so
that slowness will not be experienced by users). I suppose if swap
utilization increases, the user is going to experience slowness, is this
right?
BTW, what's the command to check memory utilization in Solaris?
The /etc/system settings of the T5120 (don't have it anymore for V880)
is as follows & attached is the "prtdiag -v" from T5120 (before it went live) :
set noexec_user_stack=1
set noexec_user_stack_log=1
set semsys:seminfo_semmsl=256
# previous value in old system is shown below
# set shmsys:shminfo_shmmax=4294
967295
set shmsys:shminfo_shmmax=8589
934592
set shmsys:shminfo_shmseg=50
set shmsys:shminfo_shmmin=1
set shmsys:shminfo_shmmni=100
set semsys:seminfo_semmns=1024
set semsys:seminfo_semmni=4096
* reduce risk of buffer overflow (hardening spec)
set nfssrv:nfs_portmon=1
set c2audit:audit_load = 1
Start Free Trial
