Sybase NNTP forums - End Of Life (EOL)

The NNTP forums from Sybase - forums.sybase.com - are now closed.

All new questions should be directed to the appropriate forum at the SAP Community Network (SCN).

Individual products have links to the respective forums on SCN, or you can go to SCN and search for your product in the search box (upper right corner) to find your specific developer center.

Your chance to help influence the future ASE direction

5 posts in ,  UnixPerformance and TuningGeneral DiscussionWindows NT Last posting was on 2000-04-03 12:35:47.0Z
Lance Andersen Posted on 2000-03-30 16:57:30.0Z
Message-ID: <38E3877A.B92CC13@sybase.com>
Date: Thu, 30 Mar 2000 11:57:30 -0500
From: Lance Andersen <lancea@sybase.com>
X-Mailer: Mozilla 4.72 [en] (X11; U; SunOS 5.7 sun4u)
X-Accept-Language: en
MIME-Version: 1.0
CC: sethu
Subject: Your chance to help influence the future ASE direction
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Newsgroups: sybase.public.sqlserver.general,sybase.public.sqlserver.nt,sybase.public.sqlserver.unix,team_fibonacci.technical,sybase.public.sqlserver.performance+tuning
Lines: 51
NNTP-Posting-Host: alder.sybase.com 157.133.125.15
Path: forums-1-dub!forums-master.sybase.com!forums.sybase.com
Xref: forums-1-dub sybase.public.sqlserver.general:3728 sybase.public.sqlserver.nt:586 sybase.public.sqlserver.unix:258 sybase.public.sqlserver.performance+tuning:290
Article PK: 1074687

Folks, we are investigating a proposal to allow for dynamic ASE
reconfiguration without requiring a reboot of ASE.

With this
changing total memory, locks, open objects, user connections will
be done dynamically and doesn't need a forced reboot.

Engineering has several ideas, which have been listed below. If you
have a specific opinion, here is your 1 chance to provide direct
feedback to Sethu, our lead ASE architect. If you have a specific
opinon, please email sethu and I (sethu@sybase.com,
lancea@sybase.com). Please do not mail ANY other questions or
requests.

Thankyou.

Lance

Here are some possible options:


1) Make the procedure cache percentage from a percentage
to a whole round value (in terms of pages or bytes)

What we will do is as part of the upgrade, we will calculate the
total memory used by the procedure cache % and remove the old config
param and introduce a new config param that will give you the
option to set procedure cache as a percentage.

2) Make the default data cache value from DEFAULT (which is calculated
as (total memory) - (kernel+server datastructures) - proc. cache %)

Like proc. cache % param, this one also will be calculated and
updated during upgrade. Subsequently an upgraded server will
disallow a DEFAULT value for default data cache.

3) Remove the 'total memory' configuration.

With the removal and chance in functionality of #1 and #2 we can
remove the total memory and make it dynamic.

Pls. send us your comments as soon as possible
--
===============================================================================
Lance J. Andersen Email: lancea@sybase.com
Sybase Product Support Engineering Phone:(781) 564-6336
77 South Bedford Street Fax: (781) 564-7001
Burlington, MA 01803

The Dark Knight Returns!!! Let's Go Penguins!!!
===============================================================================


Anthony Mandic <amandic Posted on 2000-03-31 15:58:55.0Z
Message-ID: <38E4CB3F.EE2F61A8@_start.com.au>
Date: Sat, 01 Apr 2000 01:58:55 +1000
From: Anthony Mandic <amandic@_start.com.au>
Organization: Mandic Consulting Pty. Ltd.
X-Mailer: Mozilla 4.61 [en] (WinNT; I)
X-Accept-Language: en
MIME-Version: 1.0
To: lancea@sybase.com
Subject: Re: Your chance to help influence the future ASE direction
References: <38E3877A.B92CC13@sybase.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Newsgroups: sybase.public.sqlserver.general,sybase.public.sqlserver.nt,sybase.public.sqlserver.unix,sybase.public.sqlserver.performance+tuning
Lines: 68
NNTP-Posting-Host: pix208a.magna.com.au 203.111.111.208
Path: forums-1-dub!forums-master.sybase.com!forums.sybase.com
Xref: forums-1-dub sybase.public.sqlserver.general:3726 sybase.public.sqlserver.nt:584 sybase.public.sqlserver.unix:256 sybase.public.sqlserver.performance+tuning:288
Article PK: 1074683


Lance Andersen wrote:

> Folks, we are investigating a proposal to allow for dynamic ASE
> reconfiguration without requiring a reboot of ASE.

I've been hoping for this for a long time. However, shouldn't
this go thru the ISUG enhancements process?

> With this
> changing total memory, locks, open objects, user connections will
> be done dynamically and doesn't need a forced reboot.
...
> Here are some possible options:
>
> 1) Make the procedure cache percentage from a percentage
> to a whole round value (in terms of pages or bytes)

It would probably have to be in pages rather than bytes since
the minimum useful size is a page. However, I think it should
be in terms of megabytes. Ideally, though, I'd like to see the
single shared memory segment used split up. This way, proc cache,
data cache and user cache (yes, I'd like to see three separate
shared memory segments at least) can be dynamically resized
independantly of the others.

> What we will do is as part of the upgrade, we will calculate the
> total memory used by the procedure cache % and remove the old config
> param and introduce a new config param that will give you the
> option to set procedure cache as a percentage.

Or have three config options setting the proc, data and user
cache sizes in megabytes.

> 2) Make the default data cache value from DEFAULT (which is calculated
> as (total memory) - (kernel+server datastructures) - proc. cache %)
>
> Like proc. cache % param, this one also will be calculated and
> updated during upgrade. Subsequently an upgraded server will
> disallow a DEFAULT value for default data cache.

Or have a DEFAULT for all three caches. Although the user cache
size can probably be determined from just the number of users
and the stack size. Get rid of the silly stack guard word and
monitor stack overflow correctly instead of sweeping thru the
stacks regularly and checking for overflows into the guard word
area.

> 3) Remove the 'total memory' configuration.
>
> With the removal and chance in functionality of #1 and #2 we can
> remove the total memory and make it dynamic.

Yes. And the number of locks and open objects can be dynamically
resizing internal features no longer needing to be set by users.
Rather they act like stacks that grow as needed. If they then sit
at their max size, this will give usrs an indication of what they
should be. A tripwire feature could then be added to log when they
grow over a defined boundary - warning users of excessive locking
without actually running out of locks (until they run out of
memory perhaps?).

-am


Jim Egan Posted on 2000-03-30 22:19:57.0Z
From: Jim Egan <eganjp@compuserve.com>
Subject: Re: Your chance to help influence the future ASE direction
Date: Thu, 30 Mar 2000 16:19:57 -0600
Message-ID: <MPG.134d8f093b778dd298a77a@forums.sybase.com>
References: <38E3877A.B92CC13@sybase.com>
Reply-To: eganjp@compuserve.com
X-Newsreader: MicroPlanet Gravity v2.20
Newsgroups: sybase.public.sqlserver.general,sybase.public.sqlserver.nt,sybase.public.sqlserver.unix,team_fibonacci.technical,sybase.public.sqlserver.performance+tuning
Lines: 6
NNTP-Posting-Host: user-33qs14i.dialup.mindspring.com 199.174.4.146
Path: forums-1-dub!forums-master.sybase.com!forums.sybase.com
Xref: forums-1-dub sybase.public.sqlserver.general:3727 sybase.public.sqlserver.nt:585 sybase.public.sqlserver.unix:257 sybase.public.sqlserver.performance+tuning:289
Article PK: 1074684

I like the option of setting a specific amount of RAM instead of a
percentage.
--
Jim Egan [TeamSybase]
Houston, TX

Sybase Developers Network
http://sdn.sybase.com/sdn/mec/mec_home.stm


Alberto da Silva Posted on 2000-03-31 15:10:13.0Z
Reply-To: "Alberto da Silva" <Alberto.daSilva@rmb.co.za>
From: "Alberto da Silva" <Alberto.daSilva@rmb.co.za>
References: <38E3877A.B92CC13@sybase.com>
Subject: Re: Your chance to help influence the future ASE direction
Date: Fri, 31 Mar 2000 17:10:13 +0200
Lines: 104
Organization: Rand Merchant Bank
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2919.6600
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600
Message-ID: <H65pHQym$GA.52@forums.sybase.com>
Newsgroups: sybase.public.sqlserver.general
NNTP-Posting-Host: 196.4.165.116
Path: forums-1-dub!forums-master.sybase.com!forums.sybase.com
Xref: forums-1-dub sybase.public.sqlserver.general:8219
Article PK: 1079201

I've always felt that there are too many parameters to configure in ASE.

This remindes me of Unix System V Rel 3 - lots of interelated kenel
paramters.
With the introduction of Unix System V Rel 4 - 99% of these parameters were
removed and made dynamic.

The point is that less is often better.

There are really only a few parameters that should be required - and usually
to introduce limits.
eg. Total Memory should default to maximum shm available not 8Mb, if you
need less, then you lower it.
Why must a user decide how to "partition" memory:
1) % procedure cache,
2) 2kb buffers
3) 4kb buffers
4) 8kb buffers
5) 16kb buffers,
6) named buffers,
7) additional network memory
etc.
By segmenting memory like this, you will cause degarded performance.
The dataserver has all the statistics and with a bit of intellegence should
be able to decide how best
to manage its memory.

Isn't a 16kb buffer the same as 8 x 2kb - so why must I partition off a
block for 16kb buffers.
Surely the dataserver can join 8 x 2kb buffers together when it needs to do
a large i/o.

Make as much as possible dynamic - give dataserver some intellegence - let
it use the statistics to manage memory.

Regards,

Alberto da Silva
mailto:ads@rmb.co.za

"Lance Andersen" <lancea@sybase.com> wrote in message
news:38E3877A.B92CC13@sybase.com...
> Folks, we are investigating a proposal to allow for dynamic ASE
> reconfiguration without requiring a reboot of ASE.

"Lance Andersen" <lancea@sybase.com> wrote in message
news:38E3877A.B92CC13@sybase.com...
> Folks, we are investigating a proposal to allow for dynamic ASE
> reconfiguration without requiring a reboot of ASE.
>
> With this
> changing total memory, locks, open objects, user connections will
> be done dynamically and doesn't need a forced reboot.
>
> Engineering has several ideas, which have been listed below. If you
> have a specific opinion, here is your 1 chance to provide direct
> feedback to Sethu, our lead ASE architect. If you have a specific
> opinon, please email sethu and I (sethu@sybase.com,
> lancea@sybase.com). Please do not mail ANY other questions or
> requests.
>
> Thankyou.
>
> Lance
>
> Here are some possible options:
>
>
> 1) Make the procedure cache percentage from a percentage
> to a whole round value (in terms of pages or bytes)
>
> What we will do is as part of the upgrade, we will calculate the
> total memory used by the procedure cache % and remove the old config
> param and introduce a new config param that will give you the
> option to set procedure cache as a percentage.
>
> 2) Make the default data cache value from DEFAULT (which is calculated
> as (total memory) - (kernel+server datastructures) - proc. cache %)
>
> Like proc. cache % param, this one also will be calculated and
> updated during upgrade. Subsequently an upgraded server will
> disallow a DEFAULT value for default data cache.
>
> 3) Remove the 'total memory' configuration.
>
> With the removal and chance in functionality of #1 and #2 we can
> remove the total memory and make it dynamic.
>
> Pls. send us your comments as soon as possible
> --
>
============================================================================
===
> Lance J. Andersen Email: lancea@sybase.com
> Sybase Product Support Engineering Phone:(781) 564-6336
> 77 South Bedford Street Fax: (781) 564-7001
> Burlington, MA 01803
>
> The Dark Knight Returns!!! Let's Go Penguins!!!
>
============================================================================
===


Kenny MacLeod Posted on 2000-04-03 12:35:47.0Z
From: "Kenny MacLeod" <kennym@mutant-tech.com>
References: <38E3877A.B92CC13@sybase.com> <H65pHQym$GA.52@forums.sybase.com>
Subject: Re: Your chance to help influence the future ASE direction
Date: Mon, 3 Apr 2000 13:35:47 +0100
Lines: 129
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2919.6600
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600
Message-ID: <RzriEnWn$GA.266@forums.sybase.com>
Newsgroups: sybase.public.sqlserver.general
NNTP-Posting-Host: 212.82.88.2
Path: forums-1-dub!forums-master.sybase.com!forums.sybase.com
Xref: forums-1-dub sybase.public.sqlserver.general:8210
Article PK: 1079192

I agree absolutely with Alberto on this one. While lots of manual
configuration options can be a good thing, the defaults for those parameters
should be dynamic and chosen automatically by ASE. The server has run-time
statistics and usage patterns, and it should use those to get the best out
of the available resources.

Perhaps the DBA could choose a "performance" option over, say, "low
footprint" ?

cheers

kenny

"Alberto da Silva" <Alberto.daSilva@rmb.co.za> wrote in message
news:H65pHQym$GA.52@forums.sybase.com...
> I've always felt that there are too many parameters to configure in ASE.
>
> This remindes me of Unix System V Rel 3 - lots of interelated kenel
> paramters.
> With the introduction of Unix System V Rel 4 - 99% of these parameters
were
> removed and made dynamic.
>
> The point is that less is often better.
>
> There are really only a few parameters that should be required - and
usually
> to introduce limits.
> eg. Total Memory should default to maximum shm available not 8Mb, if you
> need less, then you lower it.
> Why must a user decide how to "partition" memory:
> 1) % procedure cache,
> 2) 2kb buffers
> 3) 4kb buffers
> 4) 8kb buffers
> 5) 16kb buffers,
> 6) named buffers,
> 7) additional network memory
> etc.
> By segmenting memory like this, you will cause degarded performance.
> The dataserver has all the statistics and with a bit of intellegence
should
> be able to decide how best
> to manage its memory.
>
> Isn't a 16kb buffer the same as 8 x 2kb - so why must I partition off a
> block for 16kb buffers.
> Surely the dataserver can join 8 x 2kb buffers together when it needs to
do
> a large i/o.
>
> Make as much as possible dynamic - give dataserver some intellegence - let
> it use the statistics to manage memory.
>
> Regards,
>
> Alberto da Silva
> mailto:ads@rmb.co.za
>
> "Lance Andersen" <lancea@sybase.com> wrote in message
> news:38E3877A.B92CC13@sybase.com...
> > Folks, we are investigating a proposal to allow for dynamic ASE
> > reconfiguration without requiring a reboot of ASE.
>
> "Lance Andersen" <lancea@sybase.com> wrote in message
> news:38E3877A.B92CC13@sybase.com...
> > Folks, we are investigating a proposal to allow for dynamic ASE
> > reconfiguration without requiring a reboot of ASE.
> >
> > With this
> > changing total memory, locks, open objects, user connections will
> > be done dynamically and doesn't need a forced reboot.
> >
> > Engineering has several ideas, which have been listed below. If you
> > have a specific opinion, here is your 1 chance to provide direct
> > feedback to Sethu, our lead ASE architect. If you have a specific
> > opinon, please email sethu and I (sethu@sybase.com,
> > lancea@sybase.com). Please do not mail ANY other questions or
> > requests.
> >
> > Thankyou.
> >
> > Lance
> >
> > Here are some possible options:
> >
> >
> > 1) Make the procedure cache percentage from a percentage
> > to a whole round value (in terms of pages or bytes)
> >
> > What we will do is as part of the upgrade, we will calculate the
> > total memory used by the procedure cache % and remove the old config
> > param and introduce a new config param that will give you the
> > option to set procedure cache as a percentage.
> >
> > 2) Make the default data cache value from DEFAULT (which is calculated
> > as (total memory) - (kernel+server datastructures) - proc. cache %)
> >
> > Like proc. cache % param, this one also will be calculated and
> > updated during upgrade. Subsequently an upgraded server will
> > disallow a DEFAULT value for default data cache.
> >
> > 3) Remove the 'total memory' configuration.
> >
> > With the removal and chance in functionality of #1 and #2 we can
> > remove the total memory and make it dynamic.
> >
> > Pls. send us your comments as soon as possible
> > --
> >
>
============================================================================
> ===
> > Lance J. Andersen Email: lancea@sybase.com
> > Sybase Product Support Engineering Phone:(781) 564-6336
> > 77 South Bedford Street Fax: (781) 564-7001
> > Burlington, MA 01803
> >
> > The Dark Knight Returns!!! Let's Go Penguins!!!
> >
>
============================================================================
> ===
>
>