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.

Snapshot Isolation/MultiVersioning

17 posts in Product Futures Discussion Last posting was on 2004-10-31 10:19:43.0Z
Eugene Korolkov Posted on 2004-08-16 17:45:16.0Z
Message-ID: <4120EEA5.3F916EF7@davidsohn.com>
From: Eugene Korolkov <ekorolkov@davidsohn.com>
X-Mailer: Mozilla 4.77 [en] (Windows NT 5.0; U)
X-Accept-Language: en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
Subject: Snapshot Isolation/MultiVersioning
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Original-NNTP-Posting-Host: 12.3.91.165
X-Original-Trace: 16 Aug 2004 10:50:47 -0700, 12.3.91.165
Lines: 8
X-Original-NNTP-Posting-Host: forums-2-dub.sybase.com
X-Original-Trace: 16 Aug 2004 10:36:03 -0700, forums-2-dub.sybase.com
NNTP-Posting-Host: forums-master.sybase.com
X-Original-NNTP-Posting-Host: forums-master.sybase.com
Date: 16 Aug 2004 10:45:16 -0700
X-Trace: forums-1-dub 1092678316 10.22.108.75 (16 Aug 2004 10:45:16 -0700)
X-Original-Trace: 16 Aug 2004 10:45:16 -0700, forums-master.sybase.com
X-Authenticated-User: ngsysop
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1579
Article PK: 96574

Hi,

Just read that it is implemented in SQL Server 2005.
When Sybase is going to implement that absolutely necessary feature eventually ?

Regards,
Eugene


Eugene Korolkov Posted on 2004-08-16 17:55:53.0Z
Message-ID: <4120F126.AA6D6F29@davidsohn.com>
From: Eugene Korolkov <ekorolkov@davidsohn.com>
X-Mailer: Mozilla 4.77 [en] (Windows NT 5.0; U)
X-Accept-Language: en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
To: Eugene Korolkov <ekorolkov@davidsohn.com>
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Original-NNTP-Posting-Host: 12.3.91.165
X-Original-Trace: 16 Aug 2004 11:01:28 -0700, 12.3.91.165
Lines: 12
X-Original-NNTP-Posting-Host: forums-2-dub.sybase.com
X-Original-Trace: 16 Aug 2004 10:46:41 -0700, forums-2-dub.sybase.com
NNTP-Posting-Host: forums-master.sybase.com
X-Original-NNTP-Posting-Host: forums-master.sybase.com
Date: 16 Aug 2004 10:55:53 -0700
X-Trace: forums-1-dub 1092678953 10.22.108.75 (16 Aug 2004 10:55:53 -0700)
X-Original-Trace: 16 Aug 2004 10:55:53 -0700, forums-master.sybase.com
X-Authenticated-User: ngsysop
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1580
Article PK: 96578

BTW Oracle has it for 20 years.

Eugene Korolkov wrote:

> Hi,
>
> Just read that it is implemented in SQL Server 2005.
> When Sybase is going to implement that absolutely necessary feature eventually ?
>
> Regards,
> Eugene


Robert Greig Posted on 2004-08-16 22:18:22.0Z
From: Robert Greig <robertAT@nospam.greig.plus.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.7) Gecko/20040616
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com> <4120F126.AA6D6F29@davidsohn.com>
In-Reply-To: <4120F126.AA6D6F29@davidsohn.com>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
X-Original-NNTP-Posting-Host: greig.plus.com
Message-ID: <412133fd@forums-2-dub>
X-Original-Trace: 16 Aug 2004 15:23:57 -0700, greig.plus.com
Lines: 23
X-Original-NNTP-Posting-Host: forums-2-dub.sybase.com
X-Original-Trace: 16 Aug 2004 15:09:08 -0700, forums-2-dub.sybase.com
NNTP-Posting-Host: forums-master.sybase.com
X-Original-NNTP-Posting-Host: forums-master.sybase.com
Date: 16 Aug 2004 15:18:22 -0700
X-Trace: forums-1-dub 1092694702 10.22.108.75 (16 Aug 2004 15:18:22 -0700)
X-Original-Trace: 16 Aug 2004 15:18:22 -0700, forums-master.sybase.com
X-Authenticated-User: ngsysop
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1581
Article PK: 96579


Eugene Korolkov wrote:

> BTW Oracle has it for 20 years.
>
>>Just read that it is implemented in SQL Server 2005.
>>When Sybase is going to implement that absolutely necessary feature eventually ?

Even Postgres has it!

Prompted by your posting, I've just been looking at the list of features
in SQL Server 2005 and it does look quite impressive. (I'm not a SQL
Server user).

I'd also like to see Common Table Expressions (basically a powerful way
of avoiding temp tables while getting a recursive feature that makes
things like Bill of Materials relatively straightforward).

Multiple Active Resultsets - allowing you to submit new commands before
you have finished processing previous ones - might be handy too.

Even the feature to add try/catch to T-SQL looks good.

Robert


Eugene Korolkov Posted on 2004-08-17 14:05:14.0Z
Message-ID: <41220C92.BED6B356@davidsohn.com>
From: Eugene Korolkov <ekorolkov@davidsohn.com>
X-Mailer: Mozilla 4.77 [en] (Windows NT 5.0; U)
X-Accept-Language: en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
To: Robert Greig <robertAT@nospam.greig.plus.com>
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com> <4120F126.AA6D6F29@davidsohn.com> <412133fd@forums-2-dub>
Content-Type: multipart/alternative; boundary="------------7AC2F3E221662A63B7C29B03"
X-Original-NNTP-Posting-Host: 12.3.91.165
X-Original-Trace: 17 Aug 2004 07:10:52 -0700, 12.3.91.165
Lines: 72
X-Original-NNTP-Posting-Host: forums-2-dub.sybase.com
X-Original-Trace: 17 Aug 2004 06:55:57 -0700, forums-2-dub.sybase.com
NNTP-Posting-Host: forums-master.sybase.com
X-Original-NNTP-Posting-Host: forums-master.sybase.com
Date: 17 Aug 2004 07:05:14 -0700
X-Trace: forums-1-dub 1092751514 10.22.108.75 (17 Aug 2004 07:05:14 -0700)
X-Original-Trace: 17 Aug 2004 07:05:14 -0700, forums-master.sybase.com
X-Authenticated-User: ngsysop
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1582
Article PK: 96575

Even the feature to add try/catch to T-SQL looks good


Though I think they should go with with VB-style  error handling syntax, not Java :-)

Robert Greig wrote:

Eugene Korolkov wrote:

> BTW Oracle has it for 20 years.
>
>>Just read that it is implemented in SQL Server 2005.
>>When Sybase is going to implement that absolutely necessary feature eventually ?

Even Postgres has it!

Prompted by your posting, I've just been looking at the list of features
in SQL Server 2005 and it does look quite impressive. (I'm not a SQL
Server user).

I'd also like to see Common Table Expressions (basically a powerful way
of avoiding temp tables while getting a recursive feature that makes
things like Bill of Materials relatively straightforward).

Multiple Active Resultsets - allowing you to submit new commands before
you have finished processing previous ones - might be handy too.

Even the feature to add try/catch to T-SQL looks good.

Robert


Jason L. Froebe [TeamSybase] Posted on 2004-08-18 15:43:53.0Z
From: "Jason L. Froebe [TeamSybase]" <jason@NOSPAMMY.froebe.net>
Reply-To: jason@NOSPAMMY.froebe.net
Organization: TeamSybase
User-Agent: Mozilla Thunderbird 0.7.3 (Windows/20040803)
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com> <4120F126.AA6D6F29@davidsohn.com>
In-Reply-To: <4120F126.AA6D6F29@davidsohn.com>
Content-Type: multipart/mixed; boundary="------------080906040603090306030407"
NNTP-Posting-Host: 65.82.134.145
X-Original-NNTP-Posting-Host: 65.82.134.145
Message-ID: <41237939$1@forums-1-dub>
Date: 18 Aug 2004 08:43:53 -0700
X-Trace: forums-1-dub 1092843833 65.82.134.145 (18 Aug 2004 08:43:53 -0700)
X-Original-Trace: 18 Aug 2004 08:43:53 -0700, 65.82.134.145
Lines: 46
X-Authenticated-User: TeamSybase
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1585
Article PK: 96582


Eugene Korolkov wrote:

> BTW Oracle has it for 20 years.
>

your math is a bit off but point taken

--
Jason L. Froebe

"There is usually a balance between the left and the right... checks &
balances... the bane of the government but the boon of the people" -
Jason L. Froebe

Bookcrossing (http://www.bookcrossing.com)
WebBlog http://www.livejournal.com/users/jfroebe

TeamSybase (http://www.teamsybase.com)
ISUG member (http://www.isug.com)
Chicago Sybase Tools User Group (http://www.cpbug.com)


Download VCard jason.vcf


Jason L. Froebe [TeamSybase] Posted on 2004-08-18 15:43:30.0Z
From: "Jason L. Froebe [TeamSybase]" <jason@NOSPAMMY.froebe.net>
Reply-To: jason@NOSPAMMY.froebe.net
Organization: TeamSybase
User-Agent: Mozilla Thunderbird 0.7.3 (Windows/20040803)
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com>
In-Reply-To: <4120EEA5.3F916EF7@davidsohn.com>
Content-Type: multipart/mixed; boundary="------------010506000407080702000809"
NNTP-Posting-Host: 65.82.134.145
X-Original-NNTP-Posting-Host: 65.82.134.145
Message-ID: <41237922$1@forums-1-dub>
Date: 18 Aug 2004 08:43:30 -0700
X-Trace: forums-1-dub 1092843810 65.82.134.145 (18 Aug 2004 08:43:30 -0700)
X-Original-Trace: 18 Aug 2004 08:43:30 -0700, 65.82.134.145
Lines: 57
X-Authenticated-User: TeamSybase
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1584
Article PK: 96581


Eugene Korolkov wrote:

> Hi,
>
> Just read that it is implemented in SQL Server 2005.
> When Sybase is going to implement that absolutely necessary feature eventually ?
>
> Regards,
> Eugene
>

Hi Eugene,

Please submit your request for the versioning logging system to
http://www.isug.com/cgi-bin/ISUG2/submit_enhancement

jason

--
Jason L. Froebe

"There is usually a balance between the left and the right... checks &
balances... the bane of the government but the boon of the people" -
Jason L. Froebe

Bookcrossing (http://www.bookcrossing.com)
WebBlog http://www.livejournal.com/users/jfroebe

TeamSybase (http://www.teamsybase.com)
ISUG member (http://www.isug.com)
Chicago Sybase Tools User Group (http://www.cpbug.com)


Download VCard jason.vcf


Eugene Korolkov Posted on 2004-08-18 17:00:52.0Z
Message-ID: <41238737.12A73AFE@davidsohn.com>
From: Eugene Korolkov <ekorolkov@davidsohn.com>
X-Mailer: Mozilla 4.77 [en] (Windows NT 5.0; U)
X-Accept-Language: en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
To: jason@NOSPAMMY.froebe.net
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com> <41237922$1@forums-1-dub>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
NNTP-Posting-Host: 12.3.91.165
X-Original-NNTP-Posting-Host: 12.3.91.165
Date: 18 Aug 2004 10:00:52 -0700
X-Trace: forums-1-dub 1092848452 12.3.91.165 (18 Aug 2004 10:00:52 -0700)
X-Original-Trace: 18 Aug 2004 10:00:52 -0700, 12.3.91.165
Lines: 54
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1586
Article PK: 96583

Hi Jason,

Already did (couple years ago, I guess, do not know the story cause
I am not the member of ISUG since then) and constantly asking for years about that
in this newsgroup (see for example MVC thread on 06/02/02 and
particularly Sybase's representative responds and Sybase's gurus trying to defend obsolete
moreover wrong IMHO ANSI isolation levels model lock implementation
by Sybase, MSQL inherited from DB2 I guess, but in that article (1995!) which has
been mentioned in that thread one of the the creator of this mess
lock implementation looks like confessed that it was wrong )) and Peter Twaley personally on
the seminar at 1998 or 1999 (when RLL was introduced) .

BTW right now I am fighting with deadlock situations for the major client
which also inspired as a result of one unimplemented feature
which Sybase promised to implement already couple years ago:
Lifting transaction above the connection.

Regards,
Eugene

"Jason L. Froebe [TeamSybase]" wrote:

> Eugene Korolkov wrote:
>
> > Hi,
> >
> > Just read that it is implemented in SQL Server 2005.
> > When Sybase is going to implement that absolutely necessary feature eventually ?
> >
> > Regards,
> > Eugene
> >
>
> Hi Eugene,
>
> Please submit your request for the versioning logging system to
> http://www.isug.com/cgi-bin/ISUG2/submit_enhancement
>
> jason
>
> --
> Jason L. Froebe
>
> "There is usually a balance between the left and the right... checks &
> balances... the bane of the government but the boon of the people" -
> Jason L. Froebe
>
> Bookcrossing (http://www.bookcrossing.com)
> WebBlog http://www.livejournal.com/users/jfroebe
>
> TeamSybase (http://www.teamsybase.com)
> ISUG member (http://www.isug.com)
> Chicago Sybase Tools User Group (http://www.cpbug.com)


Jason L. Froebe [TeamSybase] Posted on 2004-08-19 04:28:00.0Z
From: "Jason L. Froebe [TeamSybase]" <jason@NOSPAMMY.froebe.net>
Reply-To: jason@NOSPAMMY.froebe.net
Organization: TeamSybase
User-Agent: Mozilla Thunderbird 0.7.3 (Windows/20040803)
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com> <41237922$1@forums-1-dub> <41238737.12A73AFE@davidsohn.com>
In-Reply-To: <41238737.12A73AFE@davidsohn.com>
Content-Type: multipart/mixed; boundary="------------000905020900080707070905"
NNTP-Posting-Host: 65.82.140.138
X-Original-NNTP-Posting-Host: 65.82.140.138
Message-ID: <41242c50@forums-1-dub>
Date: 18 Aug 2004 21:28:00 -0700
X-Trace: forums-1-dub 1092889680 65.82.140.138 (18 Aug 2004 21:28:00 -0700)
X-Original-Trace: 18 Aug 2004 21:28:00 -0700, 65.82.140.138
Lines: 83
X-Authenticated-User: TeamSybase
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1588
Article PK: 96585


Eugene Korolkov wrote:

> Hi Jason,
>
> Already did (couple years ago, I guess, do not know the story cause
> I am not the member of ISUG since then) and constantly asking for years about that
> in this newsgroup (see for example MVC thread on 06/02/02 and
> particularly Sybase's representative responds and Sybase's gurus trying to defend obsolete
> moreover wrong IMHO ANSI isolation levels model lock implementation
> by Sybase, MSQL inherited from DB2 I guess, but in that article (1995!) which has
> been mentioned in that thread one of the the creator of this mess
> lock implementation looks like confessed that it was wrong )) and Peter Twaley personally on
> the seminar at 1998 or 1999 (when RLL was introduced) .
>
> BTW right now I am fighting with deadlock situations for the major client
> which also inspired as a result of one unimplemented feature
> which Sybase promised to implement already couple years ago:
> Lifting transaction above the connection.
>
> Regards,
> Eugene

Hi Eugene,

Well, I would say that the transaction mechanism the Sybase uses is
different than say Oracle but certainly not obsolete. That's just my
opinion which doesn't count for much ;-)

I'm at TechWave and will bring up your request for the versioning
transcation system to the Sybase engineers. Keep in mind that I will be
relaying the message... I'm not an employee of Sybase so I won't be
privy to any internal meetings they may or may not have. I will pass on
any response that I receive.

I'm travelling back to Chicago from Orlando tomorrow so I may not reply
until Friday depending how late I get in.

The "Lifting transaction above the connection" ... I don't know what
you mean. Can you elaborate more on this?

Deadlocking on index pages, self deadlocks, or other? Maybe we can help
you on this?

jason
--
Jason L. Froebe

"There is usually a balance between the left and the right... checks &
balances... the bane of the government but the boon of the people" -
Jason L. Froebe

Bookcrossing (http://www.bookcrossing.com)
WebBlog http://www.livejournal.com/users/jfroebe

TeamSybase (http://www.teamsybase.com)
ISUG member (http://www.isug.com)
Chicago Sybase Tools User Group (http://www.cpbug.com)


Download VCard jason.vcf


Jason L. Froebe [TeamSybase] Posted on 2004-08-19 04:33:09.0Z
From: "Jason L. Froebe [TeamSybase]" <jason@NOSPAMMY.froebe.net>
Reply-To: jason@NOSPAMMY.froebe.net
Organization: TeamSybase
User-Agent: Mozilla Thunderbird 0.7.3 (Windows/20040803)
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com> <41237922$1@forums-1-dub> <41238737.12A73AFE@davidsohn.com> <41242c50@forums-1-dub>
In-Reply-To: <41242c50@forums-1-dub>
Content-Type: multipart/mixed; boundary="------------090608030804050705000401"
NNTP-Posting-Host: 65.82.140.138
X-Original-NNTP-Posting-Host: 65.82.140.138
Message-ID: <41242d85$1@forums-1-dub>
Date: 18 Aug 2004 21:33:09 -0700
X-Trace: forums-1-dub 1092889989 65.82.140.138 (18 Aug 2004 21:33:09 -0700)
X-Original-Trace: 18 Aug 2004 21:33:09 -0700, 65.82.140.138
Lines: 95
X-Authenticated-User: TeamSybase
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1589
Article PK: 96586


Jason L. Froebe [TeamSybase] wrote:

> Eugene Korolkov wrote:
>
>> Hi Jason,
>>
>> Already did (couple years ago, I guess, do not know the story cause
>> I am not the member of ISUG since then) and constantly asking for
>> years about that
>> in this newsgroup (see for example MVC thread on 06/02/02 and
>> particularly Sybase's representative responds and Sybase's gurus
>> trying to defend obsolete
>> moreover wrong IMHO ANSI isolation levels model lock implementation
>> by Sybase, MSQL inherited from DB2 I guess, but in that article
>> (1995!) which has
>> been mentioned in that thread one of the the creator of this mess
>> lock implementation looks like confessed that it was wrong )) and
>> Peter Twaley personally on
>> the seminar at 1998 or 1999 (when RLL was introduced) .
>>
>> BTW right now I am fighting with deadlock situations for the major client
>> which also inspired as a result of one unimplemented feature
>> which Sybase promised to implement already couple years ago:
>> Lifting transaction above the connection.
>>
>> Regards,
>> Eugene
>
>
> Hi Eugene,
>
> Well, I would say that the transaction mechanism the Sybase uses is
> different than say Oracle but certainly not obsolete. That's just my
> opinion which doesn't count for much ;-)
>
> I'm at TechWave and will bring up your request for the versioning
> transcation system to the Sybase engineers. Keep in mind that I will be
> relaying the message... I'm not an employee of Sybase so I won't be
> privy to any internal meetings they may or may not have. I will pass on
> any response that I receive.
>
> I'm travelling back to Chicago from Orlando tomorrow so I may not reply
> until Friday depending how late I get in.
>
> The "Lifting transaction above the connection" ... I don't know what
> you mean. Can you elaborate more on this?
>
> Deadlocking on index pages, self deadlocks, or other? Maybe we can help
> you on this?
>
> jason

Info on Snapshot Isolation on MS SQL Server 2005 (beta)
http://www.devx.com/codemag/Article/21570

--
Jason L. Froebe

"There is usually a balance between the left and the right... checks &
balances... the bane of the government but the boon of the people" -
Jason L. Froebe

Bookcrossing (http://www.bookcrossing.com)
WebBlog http://www.livejournal.com/users/jfroebe

TeamSybase (http://www.teamsybase.com)
ISUG member (http://www.isug.com)
Chicago Sybase Tools User Group (http://www.cpbug.com)


Download VCard jason.vcf


Eugene Korolkov Posted on 2004-08-20 21:09:24.0Z
Message-ID: <4126646A.8C7CD141@davidsohn.com>
From: Eugene Korolkov <ekorolkov@davidsohn.com>
X-Mailer: Mozilla 4.77 [en] (Windows NT 5.0; U)
X-Accept-Language: en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
To: jason@NOSPAMMY.froebe.net
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com> <41237922$1@forums-1-dub> <41238737.12A73AFE@davidsohn.com> <41242c50@forums-1-dub>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
NNTP-Posting-Host: 12.3.91.165
X-Original-NNTP-Posting-Host: 12.3.91.165
Date: 20 Aug 2004 14:09:24 -0700
X-Trace: forums-1-dub 1093036164 12.3.91.165 (20 Aug 2004 14:09:24 -0700)
X-Original-Trace: 20 Aug 2004 14:09:24 -0700, 12.3.91.165
Lines: 111
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1590
Article PK: 96587

I think that transaction mechanism
(more precisely ANSI isolation model locking implementation)
is more than obsolete , it is just wrong .

Without snapshot/isolation/multiversioning all your results
from simple select stmt in multiuser env. might be unreliable
and be dependable from access paths, indexes, CPU speed,
other users activity (locks), query optimizer decisions and so on ,
shortly "situation on the road" or artificially correct and consistent by
blocking other users.

With the absence of the snapshot concept (point of the time) you even
could easy end-up with result rows never being presented in the table at the
same moment. (BTW that is hiding on Walll Street by constant price fluctuation ,
so nobody really knows what is correct result IMHO :-))

Looks like talking to Sybase's TS engineers is useless,
they (TS) responded that they have (eventually) that request as CR
with very low priority and asking how our business will benefit on that feature
(it is not a feature, it is core ! IMHO) instead of thinking how their business
will benefit on that, though it might kill all data warehouse industry at all
(in that part of separated OLTP and decision support/rpts system cause they interfere
with each other otherwise )

About "Lifting transaction above the connection" (maybe mine definition is not very good :-))
we have something like below in VB/CRW code :
BEGIN TRAN
EXEC SP1
...
CRW.CTL
...
COMMIT TRAN

Code above and below ... belongs to one ASE connection, but
Crystal control generate another ASE connection.
(which BTW linked to just one sql stmt as opposite to ORACLE where in one connection you
can have many sql stmts from many tables and do ONLY select and
implement master-detail relationship very easy which means
that you do not need to create one big result table (DELETE/INSERT => locks/deadlocks),
which was the case with Crystal before subreports arrived, but roots of that
I think are in that stmt/connection 1:1 relationship )

Because in ASE transaction could not be propagated via different connections your
transaction mechanism could not handle 2 connections (imagine Internet which loves connect/disconnect)
in the same transaction. Long time ago Sybase promised to put transaction above the connection
as it opposite to what we have right now and did not make it.

Regards,
Eugene

"Jason L. Froebe [TeamSybase]" wrote:

> Eugene Korolkov wrote:
>
> > Hi Jason,
> >
> > Already did (couple years ago, I guess, do not know the story cause
> > I am not the member of ISUG since then) and constantly asking for years about that
> > in this newsgroup (see for example MVC thread on 06/02/02 and
> > particularly Sybase's representative responds and Sybase's gurus trying to defend obsolete
> > moreover wrong IMHO ANSI isolation levels model lock implementation
> > by Sybase, MSQL inherited from DB2 I guess, but in that article (1995!) which has
> > been mentioned in that thread one of the the creator of this mess
> > lock implementation looks like confessed that it was wrong )) and Peter Twaley personally on
> > the seminar at 1998 or 1999 (when RLL was introduced) .
> >
> > BTW right now I am fighting with deadlock situations for the major client
> > which also inspired as a result of one unimplemented feature
> > which Sybase promised to implement already couple years ago:
> > Lifting transaction above the connection.
> >
> > Regards,
> > Eugene
>
> Hi Eugene,
>
> Well, I would say that the transaction mechanism the Sybase uses is
> different than say Oracle but certainly not obsolete. That's just my
> opinion which doesn't count for much ;-)
>
> I'm at TechWave and will bring up your request for the versioning
> transcation system to the Sybase engineers. Keep in mind that I will be
> relaying the message... I'm not an employee of Sybase so I won't be
> privy to any internal meetings they may or may not have. I will pass on
> any response that I receive.
>
> I'm travelling back to Chicago from Orlando tomorrow so I may not reply
> until Friday depending how late I get in.
>
> The "Lifting transaction above the connection" ... I don't know what
> you mean. Can you elaborate more on this?
>
> Deadlocking on index pages, self deadlocks, or other? Maybe we can help
> you on this?
>
> jason
> --
> Jason L. Froebe
>
> "There is usually a balance between the left and the right... checks &
> balances... the bane of the government but the boon of the people" -
> Jason L. Froebe
>
> Bookcrossing (http://www.bookcrossing.com)
> WebBlog http://www.livejournal.com/users/jfroebe
>
> TeamSybase (http://www.teamsybase.com)
> ISUG member (http://www.isug.com)
> Chicago Sybase Tools User Group (http://www.cpbug.com)


Jason L. Froebe [TeamSybase] Posted on 2004-08-21 22:49:34.0Z
From: "Jason L. Froebe [TeamSybase]" <jason@NOSPAMMY.froebe.net>
Reply-To: jason@NOSPAMMY.froebe.net
Organization: TeamSybase
User-Agent: Mozilla Thunderbird 0.7.3 (Windows/20040803)
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com> <41237922$1@forums-1-dub> <41238737.12A73AFE@davidsohn.com> <41242c50@forums-1-dub> <4126646A.8C7CD141@davidsohn.com>
In-Reply-To: <4126646A.8C7CD141@davidsohn.com>
Content-Type: multipart/mixed; boundary="------------040809090401070603050607"
X-Original-NNTP-Posting-Host: d53-64-196-184.nap.wideopenwest.com
Message-ID: <4127d2df$1@forums-2-dub>
X-Original-Trace: 21 Aug 2004 15:55:27 -0700, d53-64-196-184.nap.wideopenwest.com
Lines: 104
X-Original-NNTP-Posting-Host: forums-2-dub.sybase.com
X-Original-Trace: 21 Aug 2004 15:39:50 -0700, forums-2-dub.sybase.com
NNTP-Posting-Host: forums-master.sybase.com
X-Original-NNTP-Posting-Host: forums-master.sybase.com
Date: 21 Aug 2004 15:49:34 -0700
X-Trace: forums-1-dub 1093128574 10.22.108.75 (21 Aug 2004 15:49:34 -0700)
X-Original-Trace: 21 Aug 2004 15:49:34 -0700, forums-master.sybase.com
X-Authenticated-User: ngsysop
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1591
Article PK: 96588


Eugene Korolkov wrote:

> I think that transaction mechanism
> (more precisely ANSI isolation model locking implementation)
> is more than obsolete , it is just wrong .
>
> Without snapshot/isolation/multiversioning all your results
> from simple select stmt in multiuser env. might be unreliable
> and be dependable from access paths, indexes, CPU speed,
> other users activity (locks), query optimizer decisions and so on ,
> shortly "situation on the road" or artificially correct and consistent by
> blocking other users.
>
> With the absence of the snapshot concept (point of the time) you even
> could easy end-up with result rows never being presented in the table at the
> same moment. (BTW that is hiding on Walll Street by constant price fluctuation ,
> so nobody really knows what is correct result IMHO :-))
>
> Looks like talking to Sybase's TS engineers is useless,
> they (TS) responded that they have (eventually) that request as CR
> with very low priority and asking how our business will benefit on that feature
> (it is not a feature, it is core ! IMHO) instead of thinking how their business
> will benefit on that, though it might kill all data warehouse industry at all
> (in that part of separated OLTP and decision support/rpts system cause they interfere
> with each other otherwise )
>
> About "Lifting transaction above the connection" (maybe mine definition is not very good :-))
> we have something like below in VB/CRW code :
> BEGIN TRAN
> EXEC SP1
> ....
> CRW.CTL
> ....
> COMMIT TRAN
>
> Code above and below ... belongs to one ASE connection, but
> Crystal control generate another ASE connection.
> (which BTW linked to just one sql stmt as opposite to ORACLE where in one connection you
> can have many sql stmts from many tables and do ONLY select and
> implement master-detail relationship very easy which means
> that you do not need to create one big result table (DELETE/INSERT => locks/deadlocks),
> which was the case with Crystal before subreports arrived, but roots of that
> I think are in that stmt/connection 1:1 relationship )
>
> Because in ASE transaction could not be propagated via different connections your
> transaction mechanism could not handle 2 connections (imagine Internet which loves connect/disconnect)
> in the same transaction. Long time ago Sybase promised to put transaction above the connection
> as it opposite to what we have right now and did not make it.
>
> Regards,
> Eugene

Hi Eugene,

I passed on your request to Sybase ASE engineering while at TechWave.

There is a white paper from IBM that I'm tracking down that points out
the versioning log system used by Postgres, Oracle and others may not be
all that it is cracked up to be... meaning there are pros and cons to
both types of transaction log systems. Neither one is obsolete or
better than the other.

Again, I'm trying to track down the paper.

--
Jason L. Froebe

"There is usually a balance between the left and the right... checks &
balances... the bane of the government but the boon of the people" -
Jason L. Froebe

http://www.froebe.net
Bookcrossing (http://www.bookcrossing.com)
WebBlog http://www.livejournal.com/users/jfroebe

TeamSybase (http://www.teamsybase.com)
ISUG member (http://www.isug.com)
Chicago Sybase Tools User Group (http://www.cpbug.com)


Download VCard jason.vcf


Eugene Korolkov Posted on 2004-08-23 15:05:29.0Z
Message-ID: <412A0389.522A92C2@davidsohn.com>
From: Eugene Korolkov <ekorolkov@davidsohn.com>
X-Mailer: Mozilla 4.77 [en] (Windows NT 5.0; U)
X-Accept-Language: en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
To: jason@NOSPAMMY.froebe.net
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com> <41237922$1@forums-1-dub> <41238737.12A73AFE@davidsohn.com> <41242c50@forums-1-dub> <4126646A.8C7CD141@davidsohn.com> <4127d2df$1@forums-2-dub>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
NNTP-Posting-Host: 12.3.91.165
X-Original-NNTP-Posting-Host: 12.3.91.165
Date: 23 Aug 2004 08:05:29 -0700
X-Trace: forums-1-dub 1093273529 12.3.91.165 (23 Aug 2004 08:05:29 -0700)
X-Original-Trace: 23 Aug 2004 08:05:29 -0700, 12.3.91.165
Lines: 99
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1595
Article PK: 96592

Jason,

I have read that paper when it was published 2 years ago. It calls
"A Technical Discussion of Multi Version Read Consistency".
Nothing interesting IMHO. The general idea that "...each transaction should
see current committed data ..." is wrong I think instead of "...old and possibly
out of date information..." which actually is correct.

"Current" depends from everything (see prev. e-mail) and not related to the point
when query started (or finished), result is the mixture of different timestamps,
because unfortunately it cannot be done immediately and absolutely inconsistent.

"Old" will give correct and consistent result and feeling that query executed immediately
(and all this without artificial locking other users !)

Regards,
Eugene

"Jason L. Froebe [TeamSybase]" wrote:

> Eugene Korolkov wrote:
>
> > I think that transaction mechanism
> > (more precisely ANSI isolation model locking implementation)
> > is more than obsolete , it is just wrong .
> >
> > Without snapshot/isolation/multiversioning all your results
> > from simple select stmt in multiuser env. might be unreliable
> > and be dependable from access paths, indexes, CPU speed,
> > other users activity (locks), query optimizer decisions and so on ,
> > shortly "situation on the road" or artificially correct and consistent by
> > blocking other users.
> >
> > With the absence of the snapshot concept (point of the time) you even
> > could easy end-up with result rows never being presented in the table at the
> > same moment. (BTW that is hiding on Walll Street by constant price fluctuation ,
> > so nobody really knows what is correct result IMHO :-))
> >
> > Looks like talking to Sybase's TS engineers is useless,
> > they (TS) responded that they have (eventually) that request as CR
> > with very low priority and asking how our business will benefit on that feature
> > (it is not a feature, it is core ! IMHO) instead of thinking how their business
> > will benefit on that, though it might kill all data warehouse industry at all
> > (in that part of separated OLTP and decision support/rpts system cause they interfere
> > with each other otherwise )
> >
> > About "Lifting transaction above the connection" (maybe mine definition is not very good :-))
> > we have something like below in VB/CRW code :
> > BEGIN TRAN
> > EXEC SP1
> > ....
> > CRW.CTL
> > ....
> > COMMIT TRAN
> >
> > Code above and below ... belongs to one ASE connection, but
> > Crystal control generate another ASE connection.
> > (which BTW linked to just one sql stmt as opposite to ORACLE where in one connection you
> > can have many sql stmts from many tables and do ONLY select and
> > implement master-detail relationship very easy which means
> > that you do not need to create one big result table (DELETE/INSERT => locks/deadlocks),
> > which was the case with Crystal before subreports arrived, but roots of that
> > I think are in that stmt/connection 1:1 relationship )
> >
> > Because in ASE transaction could not be propagated via different connections your
> > transaction mechanism could not handle 2 connections (imagine Internet which loves connect/disconnect)
> > in the same transaction. Long time ago Sybase promised to put transaction above the connection
> > as it opposite to what we have right now and did not make it.
> >
> > Regards,
> > Eugene
>
> Hi Eugene,
>
> I passed on your request to Sybase ASE engineering while at TechWave.
>
> There is a white paper from IBM that I'm tracking down that points out
> the versioning log system used by Postgres, Oracle and others may not be
> all that it is cracked up to be... meaning there are pros and cons to
> both types of transaction log systems. Neither one is obsolete or
> better than the other.
>
> Again, I'm trying to track down the paper.
>
> --
> Jason L. Froebe
>
> "There is usually a balance between the left and the right... checks &
> balances... the bane of the government but the boon of the people" -
> Jason L. Froebe
>
> http://www.froebe.net
> Bookcrossing (http://www.bookcrossing.com)
> WebBlog http://www.livejournal.com/users/jfroebe
>
> TeamSybase (http://www.teamsybase.com)
> ISUG member (http://www.isug.com)
> Chicago Sybase Tools User Group (http://www.cpbug.com)


Sybase User Posted on 2004-08-23 02:46:10.0Z
From: "Sybase User" <someone@somewhere.com>
Newsgroups: sybase.public.ase.product_futures_discussion
References: <4120EEA5.3F916EF7@davidsohn.com> <41237922$1@forums-1-dub> <41238737.12A73AFE@davidsohn.com> <41242c50@forums-1-dub> <4126646A.8C7CD141@davidsohn.com>
Subject: Re: Snapshot Isolation/MultiVersioning
Lines: 140
MIME-Version: 1.0
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1409
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1409
NNTP-Posting-Host: cupid.sybase.com
X-Original-NNTP-Posting-Host: cupid.sybase.com
Message-ID: <41295a72@forums-1-dub>
Date: 22 Aug 2004 19:46:10 -0700
X-Trace: forums-1-dub 1093229170 158.77.15.36 (22 Aug 2004 19:46:10 -0700)
X-Original-Trace: 22 Aug 2004 19:46:10 -0700, cupid.sybase.com
X-Authenticated-User: techsupp
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1592
Article PK: 96589

Mention of VB makes me to think that you'd be using Sybase ODBC Driver to
connect to ASE ?
If so, the limit of 1 active stmt/connection comes from using SelectMethod=1
(Direct) attribute
in your ODBC DSN pointing to ASE.
If you use SelectMethod= (Cursor), that limitation doesnt exist.
You may want to consider this...

"Eugene Korolkov" <ekorolkov@davidsohn.com> wrote in message
news:4126646A.8C7CD141@davidsohn.com...
> I think that transaction mechanism
> (more precisely ANSI isolation model locking implementation)
> is more than obsolete , it is just wrong .
>
> Without snapshot/isolation/multiversioning all your results
> from simple select stmt in multiuser env. might be unreliable
> and be dependable from access paths, indexes, CPU speed,
> other users activity (locks), query optimizer decisions and so on ,
> shortly "situation on the road" or artificially correct and consistent by
> blocking other users.
>
> With the absence of the snapshot concept (point of the time) you even
> could easy end-up with result rows never being presented in the table at
the
> same moment. (BTW that is hiding on Walll Street by constant price
fluctuation ,
> so nobody really knows what is correct result IMHO :-))
>
> Looks like talking to Sybase's TS engineers is useless,
> they (TS) responded that they have (eventually) that request as CR
> with very low priority and asking how our business will benefit on that
feature
> (it is not a feature, it is core ! IMHO) instead of thinking how their
business
> will benefit on that, though it might kill all data warehouse industry at
all
> (in that part of separated OLTP and decision support/rpts system cause
they interfere
> with each other otherwise )
>
> About "Lifting transaction above the connection" (maybe mine definition is
not very good :-))
> we have something like below in VB/CRW code :
> BEGIN TRAN
> EXEC SP1
> ...
> CRW.CTL
> ...
> COMMIT TRAN
>
> Code above and below ... belongs to one ASE connection, but
> Crystal control generate another ASE connection.
> (which BTW linked to just one sql stmt as opposite to ORACLE where in one
connection you
> can have many sql stmts from many tables and do ONLY select and
> implement master-detail relationship very easy which means
> that you do not need to create one big result table (DELETE/INSERT =>
locks/deadlocks),
> which was the case with Crystal before subreports arrived, but roots of
that
> I think are in that stmt/connection 1:1 relationship )
>
> Because in ASE transaction could not be propagated via different
connections your
> transaction mechanism could not handle 2 connections (imagine Internet
which loves connect/disconnect)
> in the same transaction. Long time ago Sybase promised to put transaction
above the connection
> as it opposite to what we have right now and did not make it.
>
> Regards,
> Eugene
>
>
> "Jason L. Froebe [TeamSybase]" wrote:
>
> > Eugene Korolkov wrote:
> >
> > > Hi Jason,
> > >
> > > Already did (couple years ago, I guess, do not know the story cause
> > > I am not the member of ISUG since then) and constantly asking for
years about that
> > > in this newsgroup (see for example MVC thread on 06/02/02 and
> > > particularly Sybase's representative responds and Sybase's gurus
trying to defend obsolete
> > > moreover wrong IMHO ANSI isolation levels model lock implementation
> > > by Sybase, MSQL inherited from DB2 I guess, but in that article
(1995!) which has
> > > been mentioned in that thread one of the the creator of this mess
> > > lock implementation looks like confessed that it was wrong )) and
Peter Twaley personally on
> > > the seminar at 1998 or 1999 (when RLL was introduced) .
> > >
> > > BTW right now I am fighting with deadlock situations for the major
client
> > > which also inspired as a result of one unimplemented feature
> > > which Sybase promised to implement already couple years ago:
> > > Lifting transaction above the connection.
> > >
> > > Regards,
> > > Eugene
> >
> > Hi Eugene,
> >
> > Well, I would say that the transaction mechanism the Sybase uses is
> > different than say Oracle but certainly not obsolete. That's just my
> > opinion which doesn't count for much ;-)
> >
> > I'm at TechWave and will bring up your request for the versioning
> > transcation system to the Sybase engineers. Keep in mind that I will be
> > relaying the message... I'm not an employee of Sybase so I won't be
> > privy to any internal meetings they may or may not have. I will pass on
> > any response that I receive.
> >
> > I'm travelling back to Chicago from Orlando tomorrow so I may not reply
> > until Friday depending how late I get in.
> >
> > The "Lifting transaction above the connection" ... I don't know what
> > you mean. Can you elaborate more on this?
> >
> > Deadlocking on index pages, self deadlocks, or other? Maybe we can help
> > you on this?
> >
> > jason
> > --
> > Jason L. Froebe
> >
> > "There is usually a balance between the left and the right... checks &
> > balances... the bane of the government but the boon of the people" -
> > Jason L. Froebe
> >
> > Bookcrossing (http://www.bookcrossing.com)
> > WebBlog http://www.livejournal.com/users/jfroebe
> >
> > TeamSybase (http://www.teamsybase.com)
> > ISUG member (http://www.isug.com)
> > Chicago Sybase Tools User Group (http://www.cpbug.com)
>


Jeff Tallman Posted on 2004-08-23 12:57:23.0Z
From: Jeff Tallman <tallman@sybase.com>
User-Agent: Mozilla Thunderbird 0.6 (Windows/20040502)
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com> <41237922$1@forums-1-dub> <41238737.12A73AFE@davidsohn.com> <41242c50@forums-1-dub> <4126646A.8C7CD141@davidsohn.com>
In-Reply-To: <4126646A.8C7CD141@davidsohn.com>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
NNTP-Posting-Host: pcp03383846pcs.potshe01.pa.comcast.net
X-Original-NNTP-Posting-Host: pcp03383846pcs.potshe01.pa.comcast.net
Message-ID: <4129e9b3$1@forums-1-dub>
Date: 23 Aug 2004 05:57:23 -0700
X-Trace: forums-1-dub 1093265843 68.80.126.219 (23 Aug 2004 05:57:23 -0700)
X-Original-Trace: 23 Aug 2004 05:57:23 -0700, pcp03383846pcs.potshe01.pa.comcast.net
Lines: 179
X-Authenticated-User: techsupp
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1593
Article PK: 96590

Actually, the transaction separation you are asking for was implemented
a ~5 years ago (12.0 - DTM option) - the correct terminology is "tightly
coupled transactions" and is part of the ANSI/XA spec as well as ODBC's
similar implementation (one of the check boxes in the driver dialog).
Prior to that it was available via the XA Server.

Note that this is not an argument/defense of MV vs. locking, but am
looking specifically at the situation you are raising as a rationale....

Not trying to start a war here, but I have noticed for years that VB and
in particular COM/COM+ components tend to bleed connections. I had an
experience at one customer site where a single (yes single) query to get
all the available time slots on a weekly schedule used in excess of 100
connections to the same server - all of which were open simultaneously.
You can immediately see the impact on user concurrency, memory usage,
etc. of this sort of behaviour. I asked a VB programmer (outside of
customer) as to whether this was normal and his reply was that VB is
'taught' that way due to massive attempts to decompose objects into and
have each object interact with the DB separately. His comment was that
it was a simple task to pass the connection handle from one VB object to
the other and consequently avoid the whole 2PC problem...

...in your case, is it possible to pass the VB connection handle to the
crystal control??? If not, I would also ask that you file a request to
add that feature to crystal.... Secondly, you may want to see if the
tightly coupled transaction aspect of 2PC is sufficient for what you
want vs. waiting years for MVC in ASE. I also am wondering why the
transactional consistency of the report is needed to span both the proc
(and whatever else is there) as well as the report - did the original
programmer not understand the difference between physical and logical
transactions (i.e. a logical unit of work can use exception
handling/recovery actions to avoid creating dependencies where they
really don't exist)???? Again, not trying to start an argument here,
but at first blush, the example you gave was actually a bad example as
it points more to a design/API problem that could have detrimental
impact on the server in terms of resource usage by implementing MV as a
work-around. I would love to see a better example (seriously)...

IRT the multi-versioning, it is most useful in DSS situations as larger
load activity is effectively blocked by long running DSS queries.
Hence, that is why Sybase IQ does use MV - but also solves the long
running query problem through it's speed. As far as OLTP is concerned,
with a simple query such as "What is the current balance for my
accounts?" in an MV situation will *NOT* (guaranteed unless system is
static) ever retrieve the correct answer. It will return an answer that
was correct at some point in time, but not the most current information
at the point in which the query commits/completes (iso 3) - which is
usually fairly critical in financial systems.

Regardless, MV vs. locking - both have advantages as well as
disadvantages - as to whether it is *CORE* or not sometimes can be
looked at from many different perspectives with even more opinions.
Sybase did attempt to resolve some of the locking issues in DOL
situations by using index latching, adding uncommitted insert bypass and
update return capabilities in 11.9.2.....I guess the question that
begs to be ask, if the customer app is dying from deadlocks, is the issue:

1) The excessive use of multiple connections w/o tightly coupled
transactions?

2) Does the system make use of DOL locking??

Eugene Korolkov wrote:

> I think that transaction mechanism
> (more precisely ANSI isolation model locking implementation)
> is more than obsolete , it is just wrong .
>
> Without snapshot/isolation/multiversioning all your results
> from simple select stmt in multiuser env. might be unreliable
> and be dependable from access paths, indexes, CPU speed,
> other users activity (locks), query optimizer decisions and so on ,
> shortly "situation on the road" or artificially correct and consistent by
> blocking other users.
>
> With the absence of the snapshot concept (point of the time) you even
> could easy end-up with result rows never being presented in the table at the
> same moment. (BTW that is hiding on Walll Street by constant price fluctuation ,
> so nobody really knows what is correct result IMHO :-))
>
> Looks like talking to Sybase's TS engineers is useless,
> they (TS) responded that they have (eventually) that request as CR
> with very low priority and asking how our business will benefit on that feature
> (it is not a feature, it is core ! IMHO) instead of thinking how their business
> will benefit on that, though it might kill all data warehouse industry at all
> (in that part of separated OLTP and decision support/rpts system cause they interfere
> with each other otherwise )
>
> About "Lifting transaction above the connection" (maybe mine definition is not very good :-))
> we have something like below in VB/CRW code :
> BEGIN TRAN
> EXEC SP1
> ...
> CRW.CTL
> ...
> COMMIT TRAN
>
> Code above and below ... belongs to one ASE connection, but
> Crystal control generate another ASE connection.
> (which BTW linked to just one sql stmt as opposite to ORACLE where in one connection you
> can have many sql stmts from many tables and do ONLY select and
> implement master-detail relationship very easy which means
> that you do not need to create one big result table (DELETE/INSERT => locks/deadlocks),
> which was the case with Crystal before subreports arrived, but roots of that
> I think are in that stmt/connection 1:1 relationship )
>
> Because in ASE transaction could not be propagated via different connections your
> transaction mechanism could not handle 2 connections (imagine Internet which loves connect/disconnect)
> in the same transaction. Long time ago Sybase promised to put transaction above the connection
> as it opposite to what we have right now and did not make it.
>
> Regards,
> Eugene
>
>
> "Jason L. Froebe [TeamSybase]" wrote:
>
>
>>Eugene Korolkov wrote:
>>
>>
>>>Hi Jason,
>>>
>>>Already did (couple years ago, I guess, do not know the story cause
>>>I am not the member of ISUG since then) and constantly asking for years about that
>>>in this newsgroup (see for example MVC thread on 06/02/02 and
>>>particularly Sybase's representative responds and Sybase's gurus trying to defend obsolete
>>>moreover wrong IMHO ANSI isolation levels model lock implementation
>>>by Sybase, MSQL inherited from DB2 I guess, but in that article (1995!) which has
>>>been mentioned in that thread one of the the creator of this mess
>>>lock implementation looks like confessed that it was wrong )) and Peter Twaley personally on
>>>the seminar at 1998 or 1999 (when RLL was introduced) .
>>>
>>>BTW right now I am fighting with deadlock situations for the major client
>>>which also inspired as a result of one unimplemented feature
>>>which Sybase promised to implement already couple years ago:
>>>Lifting transaction above the connection.
>>>
>>>Regards,
>>>Eugene
>>
>>Hi Eugene,
>>
>>Well, I would say that the transaction mechanism the Sybase uses is
>>different than say Oracle but certainly not obsolete. That's just my
>>opinion which doesn't count for much ;-)
>>
>>I'm at TechWave and will bring up your request for the versioning
>>transcation system to the Sybase engineers. Keep in mind that I will be
>>relaying the message... I'm not an employee of Sybase so I won't be
>>privy to any internal meetings they may or may not have. I will pass on
>>any response that I receive.
>>
>>I'm travelling back to Chicago from Orlando tomorrow so I may not reply
>>until Friday depending how late I get in.
>>
>>The "Lifting transaction above the connection" ... I don't know what
>>you mean. Can you elaborate more on this?
>>
>>Deadlocking on index pages, self deadlocks, or other? Maybe we can help
>>you on this?
>>
>>jason
>>--
>>Jason L. Froebe
>>
>>"There is usually a balance between the left and the right... checks &
>>balances... the bane of the government but the boon of the people" -
>>Jason L. Froebe
>>
>>Bookcrossing (http://www.bookcrossing.com)
>>WebBlog http://www.livejournal.com/users/jfroebe
>>
>>TeamSybase (http://www.teamsybase.com)
>>ISUG member (http://www.isug.com)
>>Chicago Sybase Tools User Group (http://www.cpbug.com)
>
>


Jason L. Froebe [TeamSybase] Posted on 2004-08-21 22:43:24.0Z
From: "Jason L. Froebe [TeamSybase]" <jason@NOSPAMMY.froebe.net>
Reply-To: jason@NOSPAMMY.froebe.net
Organization: TeamSybase
User-Agent: Mozilla Thunderbird 0.7.3 (Windows/20040803)
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: team_fibonacci.technical
Subject: Need the IBM white paper on ASE's type of Transaction log
References: <4120EEA5.3F916EF7@davidsohn.com> <41237922$1@forums-1-dub> <41238737.12A73AFE@davidsohn.com> <41242c50@forums-1-dub> <4126646A.8C7CD141@davidsohn.com>
In-Reply-To: <4126646A.8C7CD141@davidsohn.com>
Content-Type: multipart/mixed; boundary="------------090907090809090909050001"
X-Original-NNTP-Posting-Host: d53-64-196-184.nap.wideopenwest.com
Message-ID: <4127d16d$1@forums-2-dub>
X-Original-Trace: 21 Aug 2004 15:49:17 -0700, d53-64-196-184.nap.wideopenwest.com
Lines: 167
X-Original-NNTP-Posting-Host: forums-2-dub.sybase.com
X-Original-Trace: 21 Aug 2004 15:33:40 -0700, forums-2-dub.sybase.com
NNTP-Posting-Host: forums-master.sybase.com
X-Original-NNTP-Posting-Host: forums-master.sybase.com
Date: 21 Aug 2004 15:43:24 -0700
X-Trace: forums-1-dub 1093128204 10.22.108.75 (21 Aug 2004 15:43:24 -0700)
X-Original-Trace: 21 Aug 2004 15:43:24 -0700, forums-master.sybase.com
X-Authenticated-User: ngsysop
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub team_fibonacci.technical:13140
Article PK: 1302500

Hi,

For the life of me I'm unable to find the white paper from IBM that
analyzes the type of ASE's transaction log over a versioning transaction
log (Oracle, Postgres, others).

does anyone know?

jason

Eugene Korolkov wrote:
> I think that transaction mechanism
> (more precisely ANSI isolation model locking implementation)
> is more than obsolete , it is just wrong .
>
> Without snapshot/isolation/multiversioning all your results
> from simple select stmt in multiuser env. might be unreliable
> and be dependable from access paths, indexes, CPU speed,
> other users activity (locks), query optimizer decisions and so on ,
> shortly "situation on the road" or artificially correct and consistent by
> blocking other users.
>
> With the absence of the snapshot concept (point of the time) you even
> could easy end-up with result rows never being presented in the table at the
> same moment. (BTW that is hiding on Walll Street by constant price fluctuation ,
> so nobody really knows what is correct result IMHO :-))
>
> Looks like talking to Sybase's TS engineers is useless,
> they (TS) responded that they have (eventually) that request as CR
> with very low priority and asking how our business will benefit on that feature
> (it is not a feature, it is core ! IMHO) instead of thinking how their business
> will benefit on that, though it might kill all data warehouse industry at all
> (in that part of separated OLTP and decision support/rpts system cause they interfere
> with each other otherwise )
>
> About "Lifting transaction above the connection" (maybe mine definition is not very good :-))
> we have something like below in VB/CRW code :
> BEGIN TRAN
> EXEC SP1
> ....
> CRW.CTL
> ....
> COMMIT TRAN
>
> Code above and below ... belongs to one ASE connection, but
> Crystal control generate another ASE connection.
> (which BTW linked to just one sql stmt as opposite to ORACLE where in one connection you
> can have many sql stmts from many tables and do ONLY select and
> implement master-detail relationship very easy which means
> that you do not need to create one big result table (DELETE/INSERT => locks/deadlocks),
> which was the case with Crystal before subreports arrived, but roots of that
> I think are in that stmt/connection 1:1 relationship )
>
> Because in ASE transaction could not be propagated via different connections your
> transaction mechanism could not handle 2 connections (imagine Internet which loves connect/disconnect)
> in the same transaction. Long time ago Sybase promised to put transaction above the connection
> as it opposite to what we have right now and did not make it.
>
> Regards,
> Eugene
>
>
> "Jason L. Froebe [TeamSybase]" wrote:
>
>
>>Eugene Korolkov wrote:
>>
>>
>>>Hi Jason,
>>>
>>>Already did (couple years ago, I guess, do not know the story cause
>>>I am not the member of ISUG since then) and constantly asking for years about that
>>>in this newsgroup (see for example MVC thread on 06/02/02 and
>>>particularly Sybase's representative responds and Sybase's gurus trying to defend obsolete
>>>moreover wrong IMHO ANSI isolation levels model lock implementation
>>>by Sybase, MSQL inherited from DB2 I guess, but in that article (1995!) which has
>>>been mentioned in that thread one of the the creator of this mess
>>>lock implementation looks like confessed that it was wrong )) and Peter Twaley personally on
>>>the seminar at 1998 or 1999 (when RLL was introduced) .
>>>
>>>BTW right now I am fighting with deadlock situations for the major client
>>>which also inspired as a result of one unimplemented feature
>>>which Sybase promised to implement already couple years ago:
>>>Lifting transaction above the connection.
>>>
>>>Regards,
>>>Eugene
>>
>>Hi Eugene,
>>
>>Well, I would say that the transaction mechanism the Sybase uses is
>>different than say Oracle but certainly not obsolete. That's just my
>>opinion which doesn't count for much ;-)
>>
>>I'm at TechWave and will bring up your request for the versioning
>>transcation system to the Sybase engineers. Keep in mind that I will be
>>relaying the message... I'm not an employee of Sybase so I won't be
>>privy to any internal meetings they may or may not have. I will pass on
>>any response that I receive.
>>
>>I'm travelling back to Chicago from Orlando tomorrow so I may not reply
>>until Friday depending how late I get in.
>>
>>The "Lifting transaction above the connection" ... I don't know what
>>you mean. Can you elaborate more on this?
>>
>>Deadlocking on index pages, self deadlocks, or other? Maybe we can help
>>you on this?
>>
>>jason
>>--
>>Jason L. Froebe
>>
>>"There is usually a balance between the left and the right... checks &
>>balances... the bane of the government but the boon of the people" -
>>Jason L. Froebe
>>
>>Bookcrossing (http://www.bookcrossing.com)
>>WebBlog http://www.livejournal.com/users/jfroebe
>>
>>TeamSybase (http://www.teamsybase.com)
>>ISUG member (http://www.isug.com)
>>Chicago Sybase Tools User Group (http://www.cpbug.com)
>
>

--
Jason L. Froebe

"There is usually a balance between the left and the right... checks &
balances... the bane of the government but the boon of the people" -
Jason L. Froebe

http://www.froebe.net
Bookcrossing (http://www.bookcrossing.com)
WebBlog http://www.livejournal.com/users/jfroebe

TeamSybase (http://www.teamsybase.com)
ISUG member (http://www.isug.com)
Chicago Sybase Tools User Group (http://www.cpbug.com)


Download VCard jason.vcf


Sethu Posted on 2004-09-18 15:01:51.0Z
From: "Sethu" <sethu@sybase.com>
Newsgroups: sybase.public.ase.product_futures_discussion
References: <4120EEA5.3F916EF7@davidsohn.com>
Subject: Re: Snapshot Isolation/MultiVersioning
Lines: 22
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1437
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1441
X-Original-NNTP-Posting-Host: vpn-dub-050.sybase.com
Message-ID: <414c4fa2$1@forums-2-dub>
X-Original-Trace: 18 Sep 2004 08:09:22 -0700, vpn-dub-050.sybase.com
X-Original-NNTP-Posting-Host: forums-2-dub.sybase.com
X-Original-Trace: 18 Sep 2004 07:49:25 -0700, forums-2-dub.sybase.com
NNTP-Posting-Host: forums-master.sybase.com
X-Original-NNTP-Posting-Host: forums-master.sybase.com
Date: 18 Sep 2004 08:01:51 -0700
X-Trace: forums-1-dub 1095519711 10.22.108.75 (18 Sep 2004 08:01:51 -0700)
X-Original-Trace: 18 Sep 2004 08:01:51 -0700, forums-master.sybase.com
X-Authenticated-User: ngsysop
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1622
Article PK: 96619

Hi Eugene,

You and I have discussed this a bit last year. This is definitely
on our plate. Right now our plate is pretty full with ASE 15.0 and ASE 15.0+
projects. We will do in earnest some research work next year.

Thanks,
Sethu

"Eugene Korolkov" <ekorolkov@davidsohn.com> wrote in message
news:4120EEA5.3F916EF7@davidsohn.com...
> Hi,
>
> Just read that it is implemented in SQL Server 2005.
> When Sybase is going to implement that absolutely necessary feature eventually
?
>
> Regards,
> Eugene
>


Andrew Schonberger Posted on 2004-10-31 10:19:43.0Z
From: Andrew Schonberger <andrewsc@iNOSPAM.nternode.on.net>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4) Gecko/20030624 Netscape/7.1 (ax)
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: sybase.public.ase.product_futures_discussion
Subject: Re: Snapshot Isolation/MultiVersioning
References: <4120EEA5.3F916EF7@davidsohn.com> <414c4fa2$1@forums-2-dub>
In-Reply-To: <414c4fa2$1@forums-2-dub>
Content-Type: multipart/alternative; boundary="------------070700090803000806010402"
X-Original-NNTP-Posting-Host: ppp217-200.lns1.syd3.internode.on.net
Message-ID: <4184ca4b@forums-2-dub>
X-Original-Trace: 31 Oct 2004 03:19:39 -0700, ppp217-200.lns1.syd3.internode.on.net
Lines: 118
X-Original-NNTP-Posting-Host: forums-2-dub.sybase.com
X-Original-Trace: 31 Oct 2004 03:19:41 -0700, forums-2-dub.sybase.com
NNTP-Posting-Host: forums-master.sybase.com
X-Original-NNTP-Posting-Host: forums-master.sybase.com
Date: 31 Oct 2004 03:19:43 -0700
X-Trace: forums-1-dub 1099221583 10.22.108.75 (31 Oct 2004 03:19:43 -0700)
X-Original-Trace: 31 Oct 2004 03:19:43 -0700, forums-master.sybase.com
X-Authenticated-User: ngsysop
Path: forums-1-dub!not-for-mail
Xref: forums-1-dub sybase.public.ase.product_futures_discussion:1645
Article PK: 96643

Sethu, Eugene,

I regret I missed this discussion, but I clearly remember the previous one. It is encouraging to hear Sybase is considering MV.   

I'm viewing this issue from the perspective of system theory.  An observable system always shows a consistent state to any observer at any time.   It may not reflect the latest changes being applied inside, but it is consistent and is always available.   When interogated, the Oracle system replies in this fashion. On contrast, Sybase says, in effect: "I cannot tell you right now, I must wait until some other process completes work".  

Andrew

Sethu wrote:

Hi Eugene, You and I have discussed this a bit last year. This is definitely on our plate. Right now our plate is pretty full with ASE 15.0 and ASE 15.0+ projects. We will do in earnest some research work next year. Thanks, Sethu "Eugene Korolkov" <ekorolkov@davidsohn.com> wrote in message news:4120EEA5.3F916EF7@davidsohn.com...
Hi, Just read that it is implemented in SQL Server 2005. When Sybase is going to implement that absolutely necessary feature eventually
?
Regards, Eugene