RE: Have we hit the wall with Sharepoint?

Tech-Archive recommends: Repair Windows Errors & Optimize Windows Performance



I agree with Dave's idea to check the cpu usage. Whice process occupies the
cpu and memory. Just shared some experience.

We met a slow performance when a sub site owner to upload an invalid ppt to
the portal and then sent the link to users. After users tried to open the ppt
on portal, the w3svc process heavvily consumed the cpu. So we just stopped
the process and restarted it. Then the sps portal came back to a reasonbale
performance.

And then we decided a new strategy for the portal. We monthly maintained the
porta via shut down the portal and restart it. This could help us clear some
unepcted high-resource-cunsumer processes.

br
ted

"dave nixon" wrote:

> Dave
> I suspect this has nothing to do with network loads etc...this is an
> internal sps issue...
> I run a medium farm with 200+ gb data (why did I enable mysite!) and for the
> whole it works fine.
> I suspect your cpu usage is down to sps processes...can u identify in task
> magaer which is taking all the cpu/mem resource?
> I would look at the index/search/scopes/prifle import scheduling...make sure
> all are set incrementally and maybe do your full schedule once a week at 2am.
> Also look at your search scopes and see if you are indexing any huge content
> that can be slimlined.
>
> also it may be worth doing an iis reset on your web servers to see if it
> calms down.
> If you want to add more boxes to the farm then set them as index/search
> servers to try and alleviate the load.
>
> hope this helps
> Dave
>
> hope this helps
>
> "Dave" wrote:
>
> > Hi,
> > we have been running a large farm for about 2 years now. We use the SPS
> > portal as our homepage for all users (about 2000) and also use Team Sites
> > (WSS). Our total system size is now 225 GB which I have been told is
> > enormous for a Sharepoint installation. Recently our system has slowed
> > considerably, front end web servers run at 100% CPU for most of the day and
> > adding additonal load balanced front end servers has done little to help.
> > Response time is poor. We still have one database collection since we are no
> > where the limit to split off # of sites wise. In fact we only have about 30
> > WSS sites. Our SQL is a cluster and its performance has been great
> > throughout.
> >
> > Has anyone had experience with large installations and noticed a severe
> > downgrade in performance once they hit a certain size? Any ideas on tweaking
> > performance? We have read Microsoft's whitepaper outlining their settings
> > but there is nothing in there that has caused any improvement for us -
> > http://www.microsoft.com/technet/itsolutions/msit/infowork/spsperfnote.mspx#EEAA.
> > We are also up to date on Service Packs. Sharepoint was great up until we
> > approached 200G in size.
.



Relevant Pages

  • Kernel Panics from FC4 Server
    ... We have some web servers in production running FC4, ... Thanks to a serial console setup and conserver, I was able to get the ... NMI Watchdog detected LOCKUP on CPU 1 ... RAX: 0000000076c5d58e RBX: ffff8100be64b820 RCX: 0000000012eed80a ...
    (Linux-Kernel)
  • Re: HZ=100: not necessarily better?
    ... eating 10% more cpu to get a few ... Kernel compiled without any SMP support. ... some sort of test run at lower contention levels. ... running fully loaded web servers. ...
    (freebsd-performance)
  • 2.6.15.4: NFS-related BUG in radix_tree_tag_set()
    ... after a kernel upgrade from 2.6.11 to 2.6.15.4, ... These web servers obtain their data ... After one of the crashes, ... which is currently in the middle of being handled by the first CPU? ...
    (Linux-Kernel)
  • Re: 500AU Cache
    ... You can probably just change the clock setting for your CPU. ... As far as the cache module goes, the best one is the 4 meg unit. ... The systems are being used as web servers db ...
    (comp.unix.tru64)
  • Re: Yet another SVCHOST.EXE posting (my workaround)
    ... SVCHOST.EXE is an important part of Windows. ... Process Explorer you can see which and that is why I asked about the Command ... about the Windows Automatic Update service -a common cause of excessive CPU ... First off I killed PID 1302 and noticed that CPU usage fell down ...
    (microsoft.public.windowsxp.general)