By using this site, you agree to our updated Privacy Policy and our Terms of Use. Manage your Cookies Settings.
424,652 Members | 1,487 Online
Bytes IT Community
+ Ask a Question
Need help? Post your question and get tips & solutions from a community of 424,652 IT Pros & Developers. It's quick & easy.

No future for DB2

P: n/a
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp

Nov 12 '05 #1
Share this Question
Share on Google+
375 Replies


P: n/a
On 2005-07-25, rkusenet <rk******@hotmail.com> wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp


Predictions are hard. Especially if they're about the future.

Or so goes the saying.

hth
Rene

fup to comp.databases.ibm-db2
--
Rene Nyffenegger
http://www.adp-gmbh.ch/
Nov 12 '05 #2

P: n/a
rkusenet wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp


I am not familiar with the author.

What information the article does contain certainly appears to be on
target and seems fairly accurate in my opinion.

Nov 12 '05 #3

P: n/a
rkusenet wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp


This guy sounds like a magazine troll.

The more I work with DB2, the more I appreciate its power.

Maybe the world is just starting to catch up with DB2.

--
Texeme
http://texeme.com
Nov 12 '05 #4

P: n/a
"rkusenet" <rk******@hotmail.com> wrote in message
news:3k************@individual.net...
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp


There are several inaccuracies and illogical statements made in the article.
For example he says that Oracle has won the Linux market because "Oracle
that was willing to make the bold move by announcing at Oracle World its
intention to convert its entire back office infrastructure to run Linux."
AFAIK, all relevant IBM software runs on Linux, so I don't know what he is
talking about. Oracle may being doing very well on Linux, but that is mostly
at the expense of Oracle on other platforms.

His other basic premise is that the market is not big enough for 3 premium
RDBMS vendors, and that DB2 will be odd man out. I don't see that happening
anytime soon, if ever. The market is plenty big for all three, plus MySQL
for some time to come.

Both IBM and Microsoft have one big advantage over Oracle in that RDBMS's
only represent a small part of their overall revenue, and they can afford to
cut prices, unlike Oracle who depends heavily on database revenues. So IBM
will be able to undercut Oracle on pricing, and that will allow them to have
a respectable market share.

Most people on this forum are more concerned about job prospects than which
RDBMS is better or the actual number of database licenses for a particular
vendor. The real key is supply vs. demand. If there are more Oracle DBA's
than jobs for Oracle DBA's, then the job prospects may be dimmer than a less
popular database with fewer available DBA's. Since their is plenty of room
to make Oracle a lot easier to manage with fewer DBA's (DB2 and MS SQL
Server are already easy to use) then the job prospects for Oracle DBA's may
actually get worse even if the installed base gets larger.
Nov 12 '05 #5

P: n/a

"John Bailo" <ja*****@texeme.com> wrote in message
news:54********************@speakeasy.net...
rkusenet wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp


This guy sounds like a magazine troll.

The more I work with DB2, the more I appreciate its power.

Maybe the world is just starting to catch up with DB2.


Take it from comp.databases.informix, John, that the quality of the product
is more-or-less immaterial.

My first love is Informix. I've administered DB2, SQL Server and Oracle
and - no flames please, it's just an honest opinion - nothing comes close to
Informix for its elegance, ease of administration and astonishing,
leave-for-years, reliability. As a Certified DB2 UDB Specialist I happen to
agree with your warm thoughts about DB2 to an extent.

All this counts for nothing. The best product does not necessarily win out.
The best-positioned, and best-marketed product does.

So you may be right, and the magazine author may be a troll and he may be
completely wrong. But if he is wrong it isn't because of any of DB2's (or
Oracle's, or SQL Server's) inherent qualities.
Nov 12 '05 #6

P: n/a
Mark A wrote:
"rkusenet" <rk******@hotmail.com> wrote in message
news:3k************@individual.net...
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp
There are several inaccuracies and illogical statements made in the article.
For example he says that Oracle has won the Linux market because "Oracle
that was willing to make the bold move by announcing at Oracle World its
intention to convert its entire back office infrastructure to run Linux."
AFAIK, all relevant IBM software runs on Linux, so I don't know what he is
talking about.


When he says 'entire back office infrastructure' he is talking about
Oracle's internal business apps (i.e General Ledger, AR, Payroll, eMail,
Document Management, Support, CRM etc). AFAIK, IBM does not have
equivalent software.

Oracle may being doing very well on Linux, but that is mostly at the expense of Oracle on other platforms.
Oracle is also showing good growth on Windows (at least according to
Gartner). Any supposed decline in Unix platforms is so negligble to be
hardly measurable.
His other basic premise is that the market is not big enough for 3 premium
RDBMS vendors, and that DB2 will be odd man out. I don't see that happening
anytime soon, if ever. The market is plenty big for all three, plus MySQL
for some time to come.
Probably very true.
Both IBM and Microsoft have one big advantage over Oracle in that RDBMS's
only represent a small part of their overall revenue, and they can afford to
cut prices, unlike Oracle who depends heavily on database revenues. So IBM
will be able to undercut Oracle on pricing, and that will allow them to have
a respectable market share.
Yes IBM can undercut Oracle on pricing and in fact can bundle the
database with hardware and/or global services. However, it doesn't seem
to actually be helping (at least in the deals I have been involved in)
Most people on this forum are more concerned about job prospects than which
RDBMS is better or the actual number of database licenses for a particular
vendor. The real key is supply vs. demand. If there are more Oracle DBA's
than jobs for Oracle DBA's, then the job prospects may be dimmer than a less
popular database with fewer available DBA's. Since their is plenty of room
to make Oracle a lot easier to manage with fewer DBA's (DB2 and MS SQL
Server are already easy to use) then the job prospects for Oracle DBA's may
actually get worse even if the installed base gets larger.


Some how I doubt that. There are a couple of key 'speed of light' issues
that mean that DBAs will be needed for many, many years. The first is
that we are managing more data than ever before, and being asked to
manage it for longer periods of time. The second is that the number of
DBAs overall is declining, not growing, as the baby boomers start to
enter enmasse into retirement.

Not sure how a supply and demand argument for DB2 (less popular database
with fewer available DBA's) strengthens the total TCO argument either.
Buy a cheaper database and pay more to have someone look after it ?
Nov 12 '05 #7

P: n/a
"Mark Townsend" <ma***********@comcast.net> wrote in message
When he says 'entire back office infrastructure' he is talking about
Oracle's internal business apps (i.e General Ledger, AR, Payroll, eMail,
Document Management, Support, CRM etc). AFAIK, IBM does not have
equivalent software.

That is exactly why the statement is illogical. IBM does not have such
backend software, so how can IBM be faulted for not converting software it
does not have?

AFAIK, IBM has made all of it's database and related software run on Linux,
including Linux running on IBM mainframes.

I certainly am not discounting that Oracle has a larger market share than
DB2 on Linux, but the article made it seem like IBM did something wrong by
not supporting Linux for their own software.
Nov 12 '05 #8

P: n/a
rkusenet wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp


DB2 takes a bit of getting used to, but it IS easy to use compared
to Oracle. DB2 also takes a rather logical path in its hierarchy
of its architecture, which translates into an easy-to-understand
engine. It has its own quirks, like any product, but it will work
well in shops looking to make a move from SQL-Server, or other
engines.

As far as Oracle, it really is Oracle-vs-Everyone-Else, on the merits
that Oracle does not live by pessimistic locking--it is an optimistic
locking engine, the only one worth mentioning. The important point
here is that application development for Oracle is different than
say for Everyone-Else on that alone. Having said that it means that
if anyone really stands to be "odd-man-out" it would be Oracle, not
DB2. It is far easier to get DB2 working in SQL-Server shops, and
get people to actually using it. DB2 has a very easy to understand
architecture, like Informix, and has a lot of features Informix should
have, like table-level memory-management.

All the big vendors will do the leap-frog of who-is-better but none
of them are judged on a day. Besides in many IT shops Oracle is an
application requiring support, not an application development environment.
This is important to note, because it means Oracle is only as important
as its applications. Outside of Oracle applications, Oracle has a lot
more to worry about than DB2. It is also important to note that DB2
is also being used in a utilitarian way on a lot of IBM hardware, and
in Websphere, so its hard to say that DB2 is in "trouble". It runs
well, and takes a lot of abuse, similar to SQL-Server. While it might
be more favorable to work on your pet DB, most shops are using more
than just one DB anyway, so it's safe to say DB2 will be around for
a while.
Nov 12 '05 #9

P: n/a
Mark A wrote:

I certainly am not discounting that Oracle has a larger market share than
DB2 on Linux, but the article made it seem like IBM did something wrong by
not supporting Linux for their own software.


We have to be careful with 'market share'.

Oracle, for instance, during its heydey, went around selling Oracle to
almost anyone they could browbeat into buying a copy. Whether those
databases amounted to anything useful or not .. who knows.

Same with SQL Server. Everyone and his brother has a SQL Server instance
set up...but how many of those are being used for a single table of
contacts?

I maintain, and I have no data just experience, that IBM, the iSeries
( which is a DB2 UDB ), p and z Series DB2 installations are almost
entirely being used for /real/ applications...transactional
applications...live business applications.

And, while the iSeries has declined in revenue somewhat since inception,
it's staging a comeback...which means a comeback for DB2. What is more,
the architecture of OS400 (everything an object in a database) may have
been radical for 1989...but now, with Longhorn (or whatever they call it)
aping the OS400, one has to say, "yeah, IBM has been there all along".

So, let's get real. DB2 runs the world. Other RDBMS run vanity systems
and developer projects that never went anywhere.
--
Texeme
http://texeme.com
Nov 12 '05 #10

P: n/a
> AFAIK, IBM has made all of it's database and related software run on
Linux, including Linux running on IBM mainframes.


IMO, the only thing sounds more ludicrous than Linux on mainframe is Java on
mainframe. Has anybody at IBM done the math on price/performance?
Nov 12 '05 #11

P: n/a
"Bob Jones" <em***@me.not> wrote in message
news:C0***************@newssvr13.news.prodigy.com. ..
AFAIK, IBM has made all of it's database and related software run on
Linux, including Linux running on IBM mainframes.


IMO, the only thing sounds more ludicrous than Linux on mainframe is Java
on mainframe. Has anybody at IBM done the math on price/performance?


Price-performance ain't what it used to be. Mainframes are air-cooled and
some PC's (including some Mac servers) are water cooled these days.

Many mainframes use the same enterprise storage systems as most UNIX
servers.
Nov 12 '05 #12

P: n/a
rkusenet wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp


in addition to what's been remarked on: the data for the last
5 or so years make one thing quite clear -- DB2 revenue and installs
growth are mainframe driven. factor that out, and there is little to
negative growth.

locking: oracle isn't the only multiversion concurrency engine, so
are postgres and ingres (same roots, not oracle's). the other
commercial db's are lockers. this does have significance going
forward; web style databasing pretty much requires optimistic locking.
oracle should do nicely in that environment. lockers won't. this
observation has no relation (pun??, you decide) to the technical
superiority of one or the other. it's the main reason oracle is
pilloried as a resource hog.

there is *no* DB2: there are three unrelated codebases. whether or if
ibm will merge them, is opinion. mine is, not in anyone's lifetime. as
i understand it, oracle is one codebase. on the other hand, oracle is
reknowned for collapsing on z/OS. gee, ya think? will ibm be willing
and able to carry three codebases?? probably not. to the extent that
the *nix database future is web related, oracle wins. see above.

eventually, may be sooner than armonk will admit, all of those hoary
COBOL/VSAM apps will be migrated to DB2. once that happens, the
growth engine dies. cough, cough.

BobTheDataBaseBoy
Nov 12 '05 #13

P: n/a
Data Goob wrote
As far as Oracle, it really is Oracle-vs-Everyone-Else, on the merits
that Oracle does not live by pessimistic locking--it is an optimistic
locking engine, the only one worth mentioning. The important point
here is that application development for Oracle is different than
say for Everyone-Else on that alone. Having said that it means that
if anyone really stands to be "odd-man-out" it would be Oracle, not
DB2. It is far easier to get DB2 working in SQL-Server shops, and
get people to actually using it.


This is a funny way of looking at. Obviously Oracle's none locking
engine is perfectly suited to scaling multi user applications,
particularly when most people are developing for stateless clients.
The fact that developers have to go through the same gyrations for
most other databses, copying data to private sessions to avoid locks,
can hardly be a long term benefit for those databases. Yes developers
hit problems moving to Oracle because of this but it is mostly
because they want to do things that are no longer needed, see all the
posts about creating temp tables in Oracle. You very rarely need to,
just run the query, this is a database not a file system. The whole
idea is you say what you want and let the database work out how to
get it. This is alien to many developers, but is is also in part alien
to databases that don't let you just run a query, or insert what
you want to, without first working out how to go about it and not
tread on other users of the system.

Obviously without these benefits competition is easier between
locking databases, so it if it is easier to move between DB2 and
SQL Server, as the technical merits are closer, it becomes more
of a choice of OS or hardware and the database can become secondary.
I don't see how this protects DB2 share as IBM does not have a good
track record when it comes up against Microsoft in the software
market. IBM's software business plays third fiddle behind hardware
and services. It is a huge assumption that ease of migration
will result in movement mostly from SQL Server to DB2.

Of course all this might change if reports of Microsoft's efforts
to turn SQL Server into a none locking database are true. Although,
again I don't see how DB2 being the only big, locking database along
with Informix, Sybase and Ingres (maybe?) is going to help it much
either.

--
MJB

Nov 12 '05 #14

P: n/a
BobTheDataBaseBoy wrote:
rkusenet wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp

in addition to what's been remarked on: the data for the last
5 or so years make one thing quite clear -- DB2 revenue and installs
growth are mainframe driven. factor that out, and there is little to
negative growth.

locking: oracle isn't the only multiversion concurrency engine, so
are postgres and ingres (same roots, not oracle's). the other
commercial db's are lockers. this does have significance going
forward; web style databasing pretty much requires optimistic locking.
oracle should do nicely in that environment. lockers won't. this
observation has no relation (pun??, you decide) to the technical
superiority of one or the other. it's the main reason oracle is
pilloried as a resource hog.

Not to start a flame war, but optimistic-vs-pessimistic locking is
simply an architecture to deal with, it doesn't mean that one is
better suited for the web or not. If your assertion was true we'd
all be forced to use Oracle, and the rest would have all died off.
MySQL would never have made it, and Larry would be buying out Bill.

If anything is true, low cost drives the web market, and when you
begin to actually bundle prices for products, like in a spreadsheet,
and actually compare products that YOU will be responsible for their
purchase, you will sing a different tune. You will also notice that
when actually comparing prices and looking at what vertical markets
are choosing, the price point for run times is almost identical
among the big vendors--except Oracle is the priciest. Workgroup versions
too. Oracle also requires a lot of other stuff, the add-ons are too
numerous to mention, so the cost is not just the DB alone. DB2 on
this alone makes it cheaper. Utilization costs, overhead of getting
people trained, etc should also be factored in. Most people can
actually start using DB2 without training, compare that with the
beast Oracle, which has one of the most complicated engines on the
market.
there is *no* DB2: there are three unrelated codebases. whether or if
ibm will merge them, is opinion. mine is, not in anyone's lifetime. as
i understand it, oracle is one codebase. on the other hand, oracle is
reknowned for collapsing on z/OS. gee, ya think? will ibm be willing
and able to carry three codebases?? probably not. to the extent that
the *nix database future is web related, oracle wins. see above.
Yawn. If you actually knew the market outside Oracle you would not
say such blatant flame bait.
eventually, may be sooner than armonk will admit, all of those hoary
COBOL/VSAM apps will be migrated to DB2. once that happens, the
growth engine dies. cough, cough.

BobTheDataBaseBoy


Nov 12 '05 #15

P: n/a
rkusenet wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp


The author forgets a couple of things:
DB2 is the only RDBMS that runs on mainframe, midrange and PC servers.
(Oracle is non-existent on the mainframe)
There is more data stored in DB2 than in any other DBMS product.

The article is pure FUD.
--
Anton Versteeg
IBM Netherlands
Nov 12 '05 #16

P: n/a
When you really get down to it both AIX and Linux are just different
versions of UNIX. Our AIX machines are every bit as powerful as the
mainframe we migrated off of a decade ago. So Linux on a mainframe
makes perfect sense if you want to run a UNIX style system and the cost
of Linux is less than other available operating systems.

IBM is now trying to sell by service units and Java is expensive to run
under this pricing method; however, IBM sells for a very reasonable sum
Java co-processing units that remove the Java processing from the
billing. Since the underlying concept of Java is write once run
anywhere the ability to run Java on a mainframe makes perfect sense.
Whether it is the best tool for the job at hand is a different question
since most code written on a mainframe will ever run anywhere but on
the mainframe. But then again if your shop has mainframes, midrange
servers, and PC's to support the ability to have the programmers write
in Java on all three platforms instead of Visual Basic on the PC, C on
the Unix servers, and COBOL on the mainframe can be a real benefit.

IMHO -- Mark D Powell --

Nov 12 '05 #17

P: n/a
On 2005-07-26, Anton Versteeg <An************@nnll.ibm.com> wrote:
rkusenet wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp


The author forgets a couple of things:
DB2 is the only RDBMS that runs on mainframe, midrange and PC servers.
(Oracle is non-existent on the mainframe)
There is more data stored in DB2 than in any other DBMS product.

The article is pure FUD.


Oracle does run on the host.

Mike
Nov 12 '05 #18

P: n/a
bka
The statement below is untrue. Bob, databaseboy, when will you become
Bob the databaseman, and get your facts straight?

DB2 revenue has grown on zOS since 1999. But DB2 has also grown revenue
consistently on Linux/UNIX/Windows every year since at least 1997.

BobTheDataBaseBoy wrote:
in addition to what's been remarked on: the data for the last
5 or so years make one thing quite clear -- DB2 revenue and installs
growth are mainframe driven. factor that out, and there is little to
negative growth.


Nov 12 '05 #19

P: n/a
bka
How credible is the author? This calls for a lexical analysis!

He mentions DB2 by name over 20 times, but he also mentions himself
more than a dozen times. So he is promoting himself whilst disparaging
DB2 at a 3:5 ratio. He also uses the word "irregardless", so eWeek may
not have thought it worthwhile to have an editor cleanse his prose.

Merriam-Webster provides this helpful advice on the word
"irregardless":

"Etymology: probably blend of irrespective and regardless

usage Irregardless originated in dialectal American speech in the early
20th century. Its fairly widespread use in speech called it to the
attention of usage commentators as early as 1927. The most frequently
repeated remark about it is that "there is no such word." There is such
a word, however. It is still used primarily in speech, although it can
be found from time to time in edited prose. Its reputation has not
risen over the years, and it is still a long way from general
acceptance. Use regardless instead."

Nov 12 '05 #20

P: n/a
"bka" <ba*******@yahoo.com> wrote in message
news:11**********************@f14g2000cwb.googlegr oups.com...
He also uses the word "irregardless", so eWeek may
not have thought it worthwhile to have an editor cleanse his prose.


Or he might just be American.
Nov 12 '05 #21

P: n/a
bka
Strunk and White were American:

Strunk and White's The Elements of Style (4th ed.):

"Irregardless. Should be regardless. The error results from failure
to see the negative in -less and from a desire to get it in as a
prefix, suggested by such words as irregular, irresponsible, and
perhaps especially, irrespective."

Nov 12 '05 #22

P: n/a
BobTheDataBaseBoy wrote:
rkusenet wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp

in addition to what's been remarked on: the data for the last
5 or so years make one thing quite clear -- DB2 revenue and installs
growth are mainframe driven. factor that out, and there is little to
negative growth.

5 to 10 years ago there was little to no DB2 UDB for LUW revenue.
How could it then be negative.
there is *no* DB2: there are three unrelated codebases. whether or if
ibm will merge them, is opinion.

No opinion, fact: No merge. You stated below that Oracle falls appart on
zOS. Guess why? Same codebase on on *ix.
Aside.. Why do you think Oracle ships on different platforms at
different times?
I hardly call that one codebase when you need hordes of developers to port.

Cheers
Serge

PS: Just migrating Oracle RDB to DB2 .. one codebase: Hah, hah!

--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab
Nov 12 '05 #23

P: n/a
Serge Rielau wrote:
BobTheDataBaseBoy wrote:
rkusenet wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp

in addition to what's been remarked on: the data for the last
5 or so years make one thing quite clear -- DB2 revenue and installs
growth are mainframe driven. factor that out, and there is little to
negative growth.


5 to 10 years ago there was little to no DB2 UDB for LUW revenue.
How could it then be negative.


total revenue for all database could well be negative without that
COBOL/VSAM hoary system migration. and i said, sort of, "from" *5*
years ago to now. 10 years ago LUW was either an infant or still
unborn. no argument about that.
> there is *no* DB2: there are three unrelated codebases. whether or if
ibm will merge them, is opinion.


No opinion, fact: No merge. You stated below that Oracle falls appart on
zOS. Guess why? Same codebase on on *ix.


no offense to you, but i'm more inclined to think it's the armonk
version of Uncle Bill: "DOS ain't done til 1-2-3 won't run". and it
could be that Larry decided that it wasn't worth the effort to compete
where the OS was not agnostic (iSeries and zSeries; although different
histories). that's not argueable about S/38 integrated database. by
the way, PICK did it earlier. *nix is a level-ish playing field.
Aside.. Why do you think Oracle ships on different platforms at
different times?
I hardly call that one codebase when you need hordes of developers to port.
well, check out the o'reilly site: either letters or ask tim, don't
recall. q: why don't you publish platform specific versions of your
books. a: oracle is oracle, don't matter.

and so far as that goes, it's only been recently that IBM even had texts
available in bookstores. not a good move if you want mindshare.

Cheers
Serge

PS: Just migrating Oracle RDB to DB2 .. one codebase: Hah, hah!


my?!?! i just know what i read in the papers and see and hear with my
own two of each.

2005 (Colleen Graham) -- "Much of IBM's growth was generated by its DB2
on the zSeries"

2004 (Gartner, no name) -- IBM's growth was driven primarily by strong
sales on DB2 on the company's iSeries and zSeries

2002 (Gartner, no name Oracle refuting) -- only 37 percent of IBM's
database business is for UNIX and Windows NT

2002 (Gartner, no name) -- Windows... IBM 18.5 percent... Oracle 37.3
percent. UNIX... IBM (14.4 percent)... Oracle 66.2 percent

if any of these players really thought it was a slam dunk, they'd
publish per sector installs, revenue, and renewals audited figures. do
they??? i doubt it (to quote the Other Bob).

and for what it's worth: some of my colleagues get to go to the DB2
soirees (alas i must stay in CubeLand). either prior to or just back
from a recent one, one of said colleagues allowed as how we (CubeLand
management) and "other" Big Iron DB2 shops were on the verge of
threatening a defection, en masse, to Oracle. i allowed as how Big Iron
ex-COBOL/VSAM hoary systems don't get along all too well with mvcc type
engines, and it wouldn't be smart. we're still a Blue shop. perhaps
others know of this cabal??

by the way: i prefer DB2. if i could stay just on LUW, i'd be happy.
but i can't.

BobTheDataBaseBoy
Nov 12 '05 #24

P: n/a
Data Goob wrote:
Not to start a flame war, but optimistic-vs-pessimistic locking is
simply an architecture to deal with, it doesn't mean that one is
better suited for the web or not.
Could you explain how you might implement pessimistic locking in
the context of a web application which is stateless and typically
uses connection pooling?

For example, User A gets the data with connection 1, user B gets
the same data with connection 2, user A updates the data with
connection 2 (or might not). Anything else is simply client server
running in a web browser, technically achievable but pointless.
Can user B not get the data that user A has? Or does each get a
different copy of the data and the application decides who wins?
If your assertion was true we'd all be forced to use Oracle,
and the rest would have all died off.


No, it just means that somehow optimistic locking must be
implemented. It can be implemented in the application, or in the
database. It makes sense for the database to implement it as that
will provide a consistent transaction control mechanism, as opposed
to an ad hoc application dependent one. Also the cost benefits
go up with the number of applications and users that share the
data as the transaction control does not need to be re-implemented
with every new application.

I have seen developers needlessly implement their own transaction
and concurrency control in Oracle, either intentionally for database
indpendence, or just because they are used to doing it that way.

The result is always slower, less reliable and requires more code
and maintainance.

--
MJB

Nov 12 '05 #25

P: n/a
> PS: Just migrating Oracle RDB to DB2 .. one codebase: Hah, hah!

Oracle <=> RDB
DB2 <=> Informix

Does this mean Informix will be assimilated?

--
MJB

Nov 12 '05 #26

P: n/a
When you really get down to it both AIX and Linux are just different
versions of UNIX. Our AIX machines are every bit as powerful as the
mainframe we migrated off of a decade ago. So Linux on a mainframe
makes perfect sense if you want to run a UNIX style system and the cost
of Linux is less than other available operating systems.

The issue is more of the cost of mainframe rather than the cost of Linux.
Would anyone want to run Linux on mainframe when they can run it on x86 or
RISC? I am not talking about those who are unfortunately locked into
mainframe. The price/performance difference is just out of this world.
IBM is now trying to sell by service units and Java is expensive to run
under this pricing method; however, IBM sells for a very reasonable sum
Java co-processing units that remove the Java processing from the
billing. Since the underlying concept of Java is write once run
anywhere the ability to run Java on a mainframe makes perfect sense.
Whether it is the best tool for the job at hand is a different question
since most code written on a mainframe will ever run anywhere but on
the mainframe. But then again if your shop has mainframes, midrange
servers, and PC's to support the ability to have the programmers write
in Java on all three platforms instead of Visual Basic on the PC, C on
the Unix servers, and COBOL on the mainframe can be a real benefit.


I am only talking in terms of price/performance. Have you ever priced that
Java co-processor? I bet most people will get sticker shocks.
Java is "supposely" write once run anywhere, but that doesn't mean we have
to run it everywhere. It makes no sense to run it on mainframe when you can
run it much faster on x86 or RISC at a fraction of the price.
Nov 12 '05 #27

P: n/a
po******@bebub.com wrote:
Data Goob wrote:

Not to start a flame war, but optimistic-vs-pessimistic locking is
simply an architecture to deal with, it doesn't mean that one is
better suited for the web or not.

Could you explain how you might implement pessimistic locking in
the context of a web application which is stateless and typically
uses connection pooling?

For example, User A gets the data with connection 1, user B gets
the same data with connection 2, user A updates the data with
connection 2 (or might not). Anything else is simply client server
running in a web browser, technically achievable but pointless.
Can user B not get the data that user A has? Or does each get a
different copy of the data and the application decides who wins?

If your assertion was true we'd all be forced to use Oracle,
and the rest would have all died off.

No, it just means that somehow optimistic locking must be
implemented. It can be implemented in the application, or in the
database. It makes sense for the database to implement it as that
will provide a consistent transaction control mechanism, as opposed
to an ad hoc application dependent one. Also the cost benefits
go up with the number of applications and users that share the
data as the transaction control does not need to be re-implemented
with every new application.


A good argument for application logic in the database server. But
it also is a good arguement against using a web browser as the means
to run an application that would allow shared access to records that
can be updated. In other words, browser-based, "stateless" applications
are not well-suited to an environment where a record may be picked by
more than one user to be updated. It is also something to think about
when picking your next database engine, and think about whether or
not the high cost of a "built-in" record-protector in Oracle is worth
the cost, or maybe your application doesn't need that kind of feature,
and you can save a ton of money building your application with a simple
solution on a non-Oracle database, using a web browser.

I have seen developers needlessly implement their own transaction
and concurrency control in Oracle, either intentionally for database
indpendence, or just because they are used to doing it that way.

The result is always slower, less reliable and requires more code
and maintainance.

Building systems that allow several users the ability to update the
same record at any one time will always require some kind of application
logic no matter what database you use. For the money I'd want that logic
in the application rather than in the database engine, but then again
I'm always open to learning another way to do things if it will get the
job done.
Nov 12 '05 #28

P: n/a
> This is a funny way of looking at. Obviously Oracle's none locking
engine is perfectly suited to scaling multi user applications,
particularly when most people are developing for stateless clients.


ahum does the above explain why informix was faster on a 5 times
smaller machine then obstacle...????

Superboer.

Nov 12 '05 #29

P: n/a
bka wrote:
factor that out, and there is little to
negative growth.
That's always a funny one: "negative growth".


--
Knut Stolze
Information Integration Development
IBM Germany / University of Jena
Nov 12 '05 #30

P: n/a
Mike wrote:
On 2005-07-26, Anton Versteeg <An************@nnll.ibm.com> wrote:
rkusenet wrote:
This article is very bleak about future of DB2. How credible is the
author. http://www.eweek.com/article2/0,1895,1839681,00.asp


The author forgets a couple of things:
DB2 is the only RDBMS that runs on mainframe, midrange and PC servers.
(Oracle is non-existent on the mainframe)
There is more data stored in DB2 than in any other DBMS product.

The article is pure FUD.

Oracle does run on the host.

Mike

I said it is non-existent, meaning that i haven't seen a customer using it.

--
Anton Versteeg
IBM Netherlands
Nov 12 '05 #31

P: n/a
Superboer wrote:
This is a funny way of looking at. Obviously Oracle's none locking
engine is perfectly suited to scaling multi user applications,
particularly when most people are developing for stateless clients.

My interpretation of the article if that is what it was is that DB2 is
not penetrating very far in the unix world.

It obviously will not EVER go away too many big shops running it on the
mainframe.

And it has certainly gotten at least limited traction in the unix
world, some of my good friends and former colleagues are supporting SAP
at a big chemical manufacturer running DB2 on unix.

It's a good product but so much depends on momentum and marketing.

Just my opinion.

Nov 12 '05 #32

P: n/a
Superboer wrote:
This is a funny way of looking at. Obviously Oracle's none locking
engine is perfectly suited to scaling multi user applications,
particularly when most people are developing for stateless clients.

ahum does the above explain why informix was faster on a 5 times
smaller machine then obstacle...????

Superboer.

Changed the subject lines and following up on what Knut started.

How does Oracles snapshot isolation help with stateless clients.
To the best of my knowledge snapshot semantics only operate on either a
statement or a transaction level. In a stateless scenario I'd assume
that teh application transaction covers at least two database
transactions. A read phase wher the resultset is displayed at the client
and a separate write phase where the modified data is written back.
How does snapshot isolation help here?
Informix supports versioning columns which can be used by the app to
prevent overwriting other users changes across DB transaction boundaries.
MS SQL server has a somewhat similar approach and even buried optimistic
locking into the cursor logic (not applicable in a stateless enviroment
(no cursor open).
I see tha value of snapshot isolation for certain purposes. I don't see
it for a 3 tier web application....

Thoughts?
Cheers
Serge
--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab
Nov 12 '05 #33

P: n/a
bka
Short term, useful Informix features (onstat, HDR, range partitioning)
are being added (or have been added) to DB2. But there's no reason to
rush and merge the codebases as long as customers continue to buy
Informix products like XPS, IDS, RedBrick, Cloudscape and U2. IBM
never merged IMS and DB2 on zOS, as both have healthy revenue streams.
And we never distcontinued DB2 on VSE/VM, as customers continue to pay
their license fees. There are many ways for a product to be viable,
successful, profitable, and do useful work for customers without being
the number 1 database on platform x.

Nov 12 '05 #34

P: n/a
"hpuxrac" <jo*********@sbcglobal.net> wrote in message
news:11*********************@f14g2000cwb.googlegro ups.com...
Superboer wrote:
> This is a funny way of looking at. Obviously Oracle's none locking
> engine is perfectly suited to scaling multi user applications,
> particularly when most people are developing for stateless clients.

My interpretation of the article if that is what it was is that DB2 is
not penetrating very far in the unix world.

It obviously will not EVER go away too many big shops running it on the
mainframe.

And it has certainly gotten at least limited traction in the unix
world, some of my good friends and former colleagues are supporting SAP
at a big chemical manufacturer running DB2 on unix.


If only IBM already had a database product hugely respected in the UNIX
market, and in fact designed specifically *for* that market, eh?


Nov 12 '05 #35

P: n/a
Anton Versteeg wrote:
The author forgets a couple of things:
DB2 is the only RDBMS that runs on mainframe, midrange and PC servers.
False as can be...
(Oracle is non-existent on the mainframe)
Nevertheless, it RUNS on mainframes. Which makes your
claim above totally false.
There is more data stored in DB2 than in any other DBMS product.


Size matters now? :)

Face it: outside of the mainframe environment, DB2
has got NO CHANCE of long term survival.

Nov 12 '05 #36

P: n/a
bka wrote:
Strunk and White were American:

Strunk and White's The Elements of Style (4th ed.):

"Irregardless. Should be regardless. The error results from failure
to see the negative in -less and from a desire to get it in as a
prefix, suggested by such words as irregular, irresponsible, and
perhaps especially, irrespective."

immaterial

Nov 12 '05 #37

P: n/a
Serge Rielau wrote:
I see tha value of snapshot isolation for certain purposes. I don't see
it for a 3 tier web application....

Thoughts?


I'm with you here. But we'll be enlightened, I'm sure.

Nov 12 '05 #38

P: n/a
Captain Pedantic wrote:

If only IBM already had a database product hugely respected in the UNIX
market, and in fact designed specifically *for* that market, eh?


You know: from the above, to IBM buying MySQL and re-labeling
it as "common code base DB2", there is only a very little step...

Nov 12 '05 #39

P: n/a
Serge Rielau wrote:
Superboer wrote:
This is a funny way of looking at. Obviously Oracle's none locking
engine is perfectly suited to scaling multi user applications,
particularly when most people are developing for stateless clients.
ahum does the above explain why informix was faster on a 5 times
smaller machine then obstacle...????

Superboer.


Changed the subject lines and following up on what Knut started.

How does Oracles snapshot isolation help with stateless clients.
To the best of my knowledge snapshot semantics only operate on either a
statement or a transaction level. In a stateless scenario I'd assume
that teh application transaction covers at least two database
transactions. A read phase wher the resultset is displayed at the client
and a separate write phase where the modified data is written back.
How does snapshot isolation help here?

I did not mean that to read the way it does, it is sloppy and
not the point I was trying to make. My main point was contained
in the full quote.
This is a funny way of looking at. Obviously Oracle's none locking
engine is perfectly suited to scaling multi user applications,
particularly when most people are developing for stateless clients.
The fact that developers have to go through the same gyrations for
most other databases, copying data to private sessions to avoid locks,


What I was trying to get across was that it appears to me that
copying data to a private area in a session, usually a temporary
table, is a technique that can be employed in a locking database
by an application to provide a read consistent view of the data
when needed. I think this adds complexity to the application and
not having to do it because the database does it for me is a real
benefit. The fact that I imagine this would be more complex when
state is not maintained was an aside that in retrospect was not
necessary and in fact detracted from the point I was trying to
make.

Sorry.

I also apologize in advance if I have misrepresented the purpose
of temporary tables, as I have not used them in this way myself.
A separate response to my earlier post suggested they were
being used to break a query down into sub components for
performance reasons.

--
MJB

Nov 12 '05 #40

P: n/a
bka
>> Nevertheless, it (Oracle) RUNS on mainframes. Which makes your
claim above totally false.


runs, stumbles or crawls?

Nov 12 '05 #41

P: n/a
bka
>> If only IBM already had a database product hugely respected in the UNIX
market, and in fact designed specifically *for* that market, eh?


4 in the top 9, including #1:

http://www.tpc.org/tpcc/results/tpcc...5&currencyID=0

Cheapest Linux result:

http://www.tpc.org/tpcc/results/tpcc...5&currencyID=1

17 of 42 results are DB2, including 4 of 5 first place results:

http://www.tpc.org/tpch/results/tpch...ype=&version=2

Seems like the DB2 design gives decent performance on UNIX, Linix and
Windows.

Nov 12 '05 #42

P: n/a
Noons wrote:
Captain Pedantic wrote:
If only IBM already had a database product hugely respected in the UNIX
market, and in fact designed specifically *for* that market, eh?

You know: from the above, to IBM buying MySQL and re-labeling
it as "common code base DB2", there is only a very little step...

Informix Standard Engine **IS** the equivalent product to compete directly
with MySQL, but apparently nobody at Informix or IBM ever figured that one
out--despite the known clues, such as MyISAM tables, etc etc. I was told
personally by an Informix marketing exec that they "couldn't compete with
MySQL", despite at the time ( 1999-2000 ) Informix was going down in
flames. They opted to buy the racing team with their marketing budget, instead
of promoting any of their products. The racing team really helped get the
message across. Dumb fucks. Now it's too late, MySQL is the king in the
ISAM space, period, on Linux. But of course IBM is more interested in
the Java space, thusly this is why Cloudscape is given so much attention
despite the fact that there are hardly any Java applications outside
corporate environments, or installation programs. :-) When MySQL becomes
enough of a threat IBM might pull its head out and rethink SE as an open
source alternative to MySQL. But then again I doubt that will happen.
Nov 12 '05 #43

P: n/a
On 27 Jul 2005 18:45:45 -0700, "bka" <ba*******@yahoo.com> wrote:
If only IBM already had a database product hugely respected in the UNIX
market, and in fact designed specifically *for* that market, eh?


4 in the top 9, including #1:

http://www.tpc.org/tpcc/results/tpcc...5&currencyID=0

Cheapest Linux result:

http://www.tpc.org/tpcc/results/tpcc...5&currencyID=1

17 of 42 results are DB2, including 4 of 5 first place results:

http://www.tpc.org/tpch/results/tpch...ype=&version=2

Seems like the DB2 design gives decent performance on UNIX, Linix and
Windows.


Which brings up a sore point for me . . . .

WHERE IN THE WORLD ARE THE INFORMIX TPC BENCHMARKS? I get really sick
and tired of hearing "it costs too much to do that". Right, wrong, or
indifferent, if would be good to hear how Informix 10.00 stacks up
against the 'big boys'.

JWC

Nov 12 '05 #44

P: n/a
On 27 Jul 2005 16:17:57 -0700, "Noons" <wi*******@yahoo.com.au> wrote:
Anton Versteeg wrote:
The author forgets a couple of things:
DB2 is the only RDBMS that runs on mainframe, midrange and PC servers.


False as can be...
(Oracle is non-existent on the mainframe)


Nevertheless, it RUNS on mainframes. Which makes your
claim above totally false.


It takes a mainframe to run Oracle? I'd believe it . . . .8-)

JWC
Nov 12 '05 #45

P: n/a
po******@bebub.com wrote:
What I was trying to get across was that it appears to me that
copying data to a private area in a session, usually a temporary
table, is a technique that can be employed in a locking database
by an application to provide a read consistent view of the data
when needed. I think this adds complexity to the application and
not having to do it because the database does it for me is a real
benefit. The fact that I imagine this would be more complex when
state is not maintained was an aside that in retrospect was not
necessary and in fact detracted from the point I was trying to
make.

Undeserved credit to temp tables, actually....
Either way, glad we straightened that one out. I've heard too many
misconceptions about "optimistic locking" vs read consistency of late
and was starting to wonder whether I'm the one having it wrong. :-)

Cheers
Serge
--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab
Nov 12 '05 #46

P: n/a
bka wrote:
Nevertheless, it (Oracle) RUNS on mainframes. Which makes your
claim above totally false.


runs, stumbles or crawls?


It doesn't matter: it's there. The same can be said
for udb on Windows, yet it's there.

Nov 12 '05 #47

P: n/a
John Carlson wrote:

It takes a mainframe to run Oracle? I'd believe it . . . .8-)


dunno. But one thing I know for certain:
it takes a mainframe to run DB2.

Nov 12 '05 #48

P: n/a
Data Goob wrote:
flames. They opted to buy the racing team with their marketing budget, instead
of promoting any of their products. The racing team really helped get the


Oh puh-leaze: the ONLY reason IBM bought Informix
was that it was a cheap way of buying another 5 or so %
points in the db size of user base race so they could
claim to be ahead of Oracle!

They NEVER had any plan to do anything else with Informix
other than improperly claiming their share of the market
as DB2's or IBm's.

And that's a fact.

Nov 12 '05 #49

P: n/a
"Noons" <wi*******@yahoo.com.au> wrote in message

Oh puh-leaze: the ONLY reason IBM bought Informix
was that it was a cheap way of buying another 5 or so %
points in the db size of user base race so they could
claim to be ahead of Oracle!

They NEVER had any plan to do anything else with Informix
other than improperly claiming their share of the market
as DB2's or IBm's.

And that's a fact.

Kind of like Oracle buying Peoplesoft.
Nov 12 '05 #50

375 Replies

This discussion thread is closed

Replies have been disabled for this discussion.