Prev: modern firefox on Solaris 10
Next: Solaris licensing after Oracle takeover
From: underh20 on 14 Apr 2010 14:14
Dear Sir/Madam,

Our Solaris 10 server has eight 2.16Gb single-threaded cpus and 40Gb
memory.
It's running Symantec/Veritas Storage Foundation for Oracle 4.1,
Websphere 6.0.2,
and Oracle 11G. The root disk is encapsulated and mirrored.

We are getting user complaints of slowness at the server when running
program utilizing the Oracle 11G database. Looking at the "top"
output below, we noticed that there are 2 Oracle jobs (pid 26404 and
9111) which are consistently taking over 90% cpus.

How does one go about resolving this performance issue ? What commands
do we use for
trouble-shooting?

Your expert advice is appreciated. Thanks Bill


# top

last pid: 10152; load avg: 1.86, 1.85, 1.71; up
26+12:43:28 13:42:47
450 processes: 446 sleeping, 4 on cpu
CPU states: 63.6% idle, 30.3% user, 6.0% kernel, 0.0% iowait, 0.0%
swap
Memory: 40G phys mem, 11G free mem, 20G swap, 20G free swap

PID USERNAME LWP PRI NICE SIZE RES STATE TIME CPU COMMAND
26404 ocwts 11 0 0 1239M 783M cpu 33:13 99.10% oracle
9111 ocwts 14 0 0 1260M 806M cpu 0:22 90.24% oracle
9730 wasadm 113 31 2 1512M 1081M cpu 100:07 20.71% java
5073 cwats 11 60 4 1233M 772M sleep 1:12 10.39% oracle
9833 wasadm 114 26 11 3456M 2745M sleep 523:37 8.00% java
9832 wasadm 108 26 11 2106M 1446M sleep 110:44 7.76% java
9600 wasadm 113 26 11 2501M 1812M sleep 206:07 4.90% java
9835 wasadm 109 26 11 2045M 1236M sleep 99:16 3.10% java
9834 wasadm 110 26 11 2102M 1416M sleep 113:03 2.30% java
From: Richard B. Gilbert on 14 Apr 2010 17:15
underh20 wrote:
> Dear Sir/Madam,
>
> Our Solaris 10 server has eight 2.16Gb single-threaded cpus and 40Gb
> memory.
> It's running Symantec/Veritas Storage Foundation for Oracle 4.1,
> Websphere 6.0.2,
> and Oracle 11G. The root disk is encapsulated and mirrored.
>
> We are getting user complaints of slowness at the server when running
> program utilizing the Oracle 11G database. Looking at the "top"
> output below, we noticed that there are 2 Oracle jobs (pid 26404 and
> 9111) which are consistently taking over 90% cpus.
>
> How does one go about resolving this performance issue ? What commands
> do we use for
> trouble-shooting?
>
> Your expert advice is appreciated. Thanks Bill
>
>
> # top
>
> last pid: 10152; load avg: 1.86, 1.85, 1.71; up
> 26+12:43:28 13:42:47
> 450 processes: 446 sleeping, 4 on cpu
> CPU states: 63.6% idle, 30.3% user, 6.0% kernel, 0.0% iowait, 0.0%
> swap
> Memory: 40G phys mem, 11G free mem, 20G swap, 20G free swap
>
> PID USERNAME LWP PRI NICE SIZE RES STATE TIME CPU COMMAND
> 26404 ocwts 11 0 0 1239M 783M cpu 33:13 99.10% oracle
> 9111 ocwts 14 0 0 1260M 806M cpu 0:22 90.24% oracle
> 9730 wasadm 113 31 2 1512M 1081M cpu 100:07 20.71% java
> 5073 cwats 11 60 4 1233M 772M sleep 1:12 10.39% oracle
> 9833 wasadm 114 26 11 3456M 2745M sleep 523:37 8.00% java
> 9832 wasadm 108 26 11 2106M 1446M sleep 110:44 7.76% java
> 9600 wasadm 113 26 11 2501M 1812M sleep 206:07 4.90% java
> 9835 wasadm 109 26 11 2045M 1236M sleep 99:16 3.10% java
> 9834 wasadm 110 26 11 2102M 1416M sleep 113:03 2.30% java

It's hard to say what's going on except that Oracle is eating your system.

There are good and not so good ways to design your database. There are
smart ways and not so smart ways to query and/or update your database.
A good DBA should be able to help you optimize your database and your
queries and updates. One rule of thumb that I recall is "Don't put both
a table and an index to that table on the same disk." This is far from
being the only thing to look at.

The last time I installed Oracle, the directions said that a *minimum*
of five disk drives were required in order to get good performance.

The Oracle documentation goes into the subject of optimizing your
database layout and your queries in some detail. If you haven't studied
it, please do so.

I think you need Oracle database help rather than Solaris help.


From: Stefaan A Eeckels on 15 Apr 2010 16:57
On Wed, 14 Apr 2010 11:14:26 -0700 (PDT)
underh20 <underh20.scubadiving(a)gmail.com> wrote:

> We are getting user complaints of slowness at the server when running
> program utilizing the Oracle 11G database. Looking at the "top"
> output below, we noticed that there are 2 Oracle jobs (pid 26404 and
> 9111) which are consistently taking over 90% cpus.
>
> How does one go about resolving this performance issue ? What commands
> do we use for trouble-shooting?

Enterprise Manager will tell you which queries cause the problem. In
all likelihood, it's a bunch of queries that do full table scans.

If you don't know how to start EM (hint: "emctl start dbconsole") or
use it (hint "http://oracle-host:5500/em") then you need the assistance
of a DBA (someone who doesn't need EM :).

Take care,

--
Stefaan A Eeckels
--
Life itself is a misery and nobody can tell what can be of it.
Those that can tell what can be of it are those who cannot tell
us because they are far from us (dead). -- Very profound scam
 | 
Pages: 1
Prev: modern firefox on Solaris 10
Next: Solaris licensing after Oracle takeover