As far as removing the AV altogether?
One reason (in Microsoft at least), is file system filter drivers.
AV applications, and many other applications, are adding these "fsfd's" into
their products. Another reason, on any platform, is the intense and constant
struggle & constraint for server platform resources between the number of
applications Server paltofrms are running these days:
- File system fragmentation.
- "Un-Delete" applications.
- Anti-Virus Applications.
- Security applications.
- Various "snap shot" technologies.
- Backup products.
- Replication products.
Low overhead, detailed performance monitoring, 24x7x365, tells you all these
things and more, not only in U2 but across the entire application server
technology spectrum.
Remember when application servers had scheduled downtime, for periodic
platform maintenance? Disk integrity checks, defragmentation, etc? Patches
and updates applied? These days, it can almost take an act of congress at
some places to get a server reboot accomplished.
Remember when hardware vendors ran industry standard benchmarks and
characterization - simulation testing, to push Systems / Servers under "real
life application" load to see how to best, code, configure & tune them?
Seems like these functions have fairly much fallen by the wayside, left to
the individual company/organization, due to the uniqueness & peculiaralities
of "your actual mileage may vary", and explosion of technology that has
occurred over the years. This is not a bad thing, necessarily, but how does
a company / organization or indivdual go about such montioring, in any given
environment, be it production or pre-production, without the overhead of
such montioring interfereing in the performance of the platform?
How does one know, in their specific environment, with their given hardware
platforms, network configurations, user loads, and data set sizes, and
operational business dynamics, what is "normal" performance, what is
bottle-necked or throttling server platform performance, and why? It is
predictable? Is it a problem that can be fixed, re-coded, adjusted, tuned,
or configured better? More hardware required? What is the "cost" of running
multple applications at once and where/when do they "conflict for
resources"? What resources? Why? How does my Windows application server
platform performance compare with a similar Linux or other Unix platform
performance, "apples to apples"?? Can my current application server
platform support our anticipated growth ? How much will running "month-end"
impact our normal business processing cycles?
It's always a lot easier when debugging any problem, tuning any application
or server platform, or explaining / justifying capital expenses, (or even
justifying a reboot!), to have easy to see/understand detailed graphical,
factual and specific current real-time data to help point out the
operational dynamics of any problem in terms everyone can see/understand,
and then pro-actively monitor and verify as time moves forward, in real
time, and changes are made in hopes of alleviating or resovling the problem.
Have proof positive that your suggestion or changes did indeed identify
and/or resolve a problem.
Coming from an operational, uptime, and perfomance minded "server platforms
used in business" background, I can not understand why
companies/organizations/individuals would not want/need/require such data
pro-actively, to continually improve, justify, and ensure that your
technology infastructure is delivering all it can, 24x7x365.
/ad on
Performance Montoring technology exisits to allow very similar, very
detailed server wide and process level montioring of numerous diverse Server
application platforms; hardware, operating systems and
databases/applications - in one common interface, with extremely low
overhead & foot print.
- multiple processors / cores
- distributed applications
- network configuration and throughput
- storage sub-systems
- Windows
- Linux
- AIX
- Solaris
One such performance monitoing application can be found at deltek.us
/ad off
<(PS: I tried real hard to not jump into this one, but in the end, I could
not help myself - as I deal with this stuff all the time, successfully ;^)>.
Regards,
Scott Richardson
Senior Systems Engineer / Consultant
Product Support Engineer
Marlborough, MA 01752
****************************************
DPMonitor - http://www.deltek.us
----- Original Message -----
From: "Smith, Robert" <[EMAIL PROTECTED]>
To: <u2-users@listserver.u2ug.org>
Sent: Monday, March 31, 2008 9:45 AM
Subject: RE: [U2] RE: UNIDATA 7.1 performance measures
Good Morning John,
Our Network Manager asked why you recommended removing the anti-virus
software vs. disabling them. Could/would you please share your rationale
for making that recommendation?
The feedback I am getting as a result of this posting has been
incredibly helpful.
Rob Smith
Philadelphia University
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of John Jenkins
Sent: Saturday, March 29, 2008 8:29 PM
To: u2-users@listserver.u2ug.org
Subject: RE: [U2] RE: UNIDATA 7.1 performance measures
Ibd start by looking for specific bottlenecks. Itbs too easy to see
one of the system constraints and assume that's your bottleneck (been
there). Run some mete
As others have said, see a specific bottleneck first then take it from
there.
Some suggestions on things to look at in Windows and hardware - I'm
assuming all the usual has been done inside UniData itself.
__________________
Anti-Virus software killing you?
REMOVE it first for the avoidance of doubt - then decide. Donbt just
reconfigure or disable - REMOVE. Tweak afterwards - know whether itbs
worth tweaking first.
Paging (yes/no/how much)?
Disk performance
*read
*write
*wait I/O
Are you using RAID?
Should be 0+1, try and avoid RAID 5
Use hardware RAID not software
Turn off read-ahead, optimise for random, not sequential Steps vary for
different SANS (Hitachi / Veritas / Shark etc).
If using an external SAN:
* has the memory battery failed? (turns off write caching - a killer)
Put paging space on a separate disk, and UniData TMP also.
* Do you have remote mirroring on the SAM turned on (requiring remote
commit on writes can be a killer).
_________________________
If 3Gb is does turn out to be an issue consider Unix, much better memory
management. Linux on your x86 platform seems the logical move (same
hardware). You also get better CPU resource usage on Unix then Windows.
Memory could be a bottleneck for you, but Ibd be surprised. The usual
memory related issues I see on Windows (or Unix come to that) are due to
file import/export programs READING and WRITING truly enormous text
files from the O/S file structures. READSEQ and WRITESEQ are the cure
there/
Regards
JayJay
________________________________
From: Smith, Robert
Sent: Thursday, March 27, 2008 11:37 AM
To: u2-users@listserver.u2ug.org
Subject: UNIDATA 7.1 performance measures
Good Morning,
We are a UNIDATA shop and presently we are struggling with identifying
the best way to increase the performance level of our environment. We
presently are running UD on a Windows Enterprise server (SP2) with 16gig
of ram and 8 - 3gig processors. We have read through some MS Knowledge
Base articles that suggest a possible performance increase if we use the
/PAE and /3GB boot.ini switches. We were wondering whether anyone has
deployed these settings before and if there were any problems that
resulted...with either the OS or database.
Does anyone have any suggestions regarding ways to increase UD
performance levels in a Win2k3 environment?
Thanks in advance,
Rob Smith
Philadelphia University
-------
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
-------
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
-------
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
-------
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
-------
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
-------
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
