Thread: Postgres vs commercial products
As long as postgres is open source and free, postgres never has to worry about the commercial competition (in my eyes). i have far more faith in the postgresql developers to implement features and standards than i do in the commercial vendors of similar systems. in commercial enterprises such as Oracle, INnformix and Sybase, whether to implement a feature or part of a standard gets boiled down to a business decision/question -- will adding this or that create more profit for the company? if the answer is no, the feature or standard spec dies. at postgres the story is different. the question is, will the feature be useful? if the answer is yes, unless it is beyond the scope of the developers, it will get implemented. whereas the commercial vendors claim to be customer focused, i don't see daily emails from larry ellison telling me whats new and what's being considered for change etc. etc. i'm sticking with the group that has time andd again proven itself to be the group which delivers, foreign keys will come soon enough. The glitches in vacuum will go away soon enough. Outer, left, right and combinations of these joins will come soon enough. And all of this i get for free. Postgres offered banner ads for a while. SPeaking of promoting it, is postgres willing to create a logo a la apache, powered by postgresql? it'll be up on my site in no time at all. steve doliov
Steve Doliov wrote:Postgres offered banner ads for a while. SPeaking of promoting it, is > postgres willing to create a logo a la apache, powered by postgresql? > it'll be up on my site in no time at all. Since I last made a batch of logo's to prompt comments, I've been meaning to make a "serious" attempt again but keep getting side tracked. View http://www.nettek-llc.com/postgresql/. Hopefully soon, I will spend a weekend on making something mo'better. Comments are welcome. -STEVEl
At 23:36 +0300 on 21/7/98, Steve Doliov wrote: > in commercial enterprises such as Oracle, INnformix and Sybase, whether to > implement a feature or part of a standard gets boiled down to a business > decision/question -- will adding this or that create more profit for the > company? if the answer is no, the feature or standard spec dies. > > at postgres the story is different. the question is, will the feature be > useful? if the answer is yes, unless it is beyond the scope of the > developers, it will get implemented. Not that I want to offend the Postgres developers, who are doing a great job. But in the same way that commercial enterprises think in terms of profit, volunteer workers think of it like this: Does it interest me to implement this feature? Do I have the time? Do I have the expertise? Commercial companies will make the time if there is profit. They will also train their workers if they don't have the expertise. And wheather or not the workers enjoy what they are doing is not an issue... Commercial enterprises *are* customer-focused. If they can offer a feature which will make a customer prefer them over the competition, they will implement it. There is a strong correlation between profit and client satisfaction - at least, in a competitive market (which, unlike the OS market, IS true for databases). Here in the university, my bosses seriously consider moving to a commercial product - either Oracle or Informix. There will be two reasons for it: * Compatibility with software. Most RADs and server-side whatever work with one of the mentioned commercial products. Some work with ODBC, but I think that at this moment, nobody actually tried them with the PostgreSQL ODBC driver. * Replication server etc. - very important for our sysadmin. Even a reliable backup facility. I've seen too many complaints on the lists, plus saw an actual glitch myself, to trust in pg_dump. Moreover, I have to invest extra time in making pg_dump into a solution with scheduled backups and alerts for tape changes and whatnot... * Features. The only real advantage Postgres has at the moment is its price... One last comment: We are based on Solaris/sparc. So as far as I'm concerned, commercial products are already here. Their intended invasion into the linux market will make the situation exactly similar to mine. Herouth -- Herouth Maoz, Internet developer. Open University of Israel - Telem project http://telem.openu.ac.il/~herutma
On Tue, 21 Jul 1998, Steve Logue wrote: > > postgres willing to create a logo a la apache, powered by postgresql? > > it'll be up on my site in no time at all. > > Since I last made a batch of logo's to prompt comments, I've been meaning to > make a "serious" attempt again but keep getting side tracked. View > http://www.nettek-llc.com/postgresql/. Hopefully soon, I will spend a > weekend on making something mo'better. Comments are welcome. I almost hate to ask, and really, it's not a big deal to me personally, but why is it called "PostgreSQL"? -- Amos Hayes Systems Architect ahayes@ingenia.com Ingenia Group - Software Kinetics Ltd. http://smurf.ingenia.com/~ahayes http://www.ingenia.com "Remember: No one can make you feel inferior without your consent." - ELEANOR ROOSEVELT
On Wed, 22 Jul 1998, Amos Hayes wrote:
> I almost hate to ask, and really, it's not a big deal to me personally,
> but why is it called "PostgreSQL"?
Originally it was just Postgres, and didn't support any SQL. When it was
redesigned with SQL support, that's when the SQL postfix came about.
Actually, it was known as Postgres95 originally to differentiate it from
the original Postgres, but as it moved towards SQL compliancy, the SQL
postfix becamse the norm. The full story is in the PostgreSQL
documentation.
Brett W. McCoy
http://www.lan2wan.com/~bmccoy
-----------------------------------------------------------------------
"The Number of UNIX installations has grown to 10, with more expected."
-- The UNIX Programmer's Manual, 2nd Edition, June, 1972
On Wed, 22 Jul 1998, Brett W. McCoy wrote: > On Wed, 22 Jul 1998, Amos Hayes wrote: > > > I almost hate to ask, and really, it's not a big deal to me personally, > > but why is it called "PostgreSQL"? > > Originally it was just Postgres, and didn't support any SQL. When it was > redesigned with SQL support, that's when the SQL postfix came about. > Actually, it was known as Postgres95 originally to differentiate it from > the original Postgres, but as it moved towards SQL compliancy, the SQL > postfix becamse the norm. The full story is in the PostgreSQL > documentation. Thanks for the reply Brett. Has there ever been any discussion about a new name? It doesn't matter to me but it would seem that with the current discussions about promotion and competition, it might help to have a more "public friendly" name. It is not obvious to me what a "postgres" is nor what it would do if I were to install it. Granted that "oracle", "informix", and "sybase" are all a little strange too, but they give hints (oracle, inform, base) about containing knowledge. They also seem to roll off the tongue a little easier. Of course, changing the name (again) would be a BIG pain in the butt and might cause current users a whole lot more confusion. But anyway, I'm happy the thing works, is supported by a keen bunch, and is free. Thank goodness for the RedHat installer blurbs on packages. It's the only reason I found out about PostgreSQL in the first place. -- Amos Hayes Systems Architect ahayes@ingenia.com Ingenia Group - Software Kinetics Ltd. http://smurf.ingenia.com/~ahayes http://www.ingenia.com "Remember: No one can make you feel inferior without your consent." - ELEANOR ROOSEVELT
On Wed, 22 Jul 1998, Amos Hayes wrote: > On Tue, 21 Jul 1998, Steve Logue wrote: > > > > postgres willing to create a logo a la apache, powered by postgresql? > > > it'll be up on my site in no time at all. > > > > Since I last made a batch of logo's to prompt comments, I've been meaning to > > make a "serious" attempt again but keep getting side tracked. View > > http://www.nettek-llc.com/postgresql/. Hopefully soon, I will spend a > > weekend on making something mo'better. Comments are welcome. > > I almost hate to ask, and really, it's not a big deal to me personally, > but why is it called "PostgreSQL"? It used to be called Postgres95, which was Jolly and Andrew's "joke/play" on Windows95...when it came time to rename it, we wanted to maintain the 'history' of it being a Postgres4.2 based system, while bringing forward the fact that it no longer maintained the old non-SQL based query language...so the name PostgreSQL arose.... Pronounced: Postgres-Q-L
On Wed, 22 Jul 1998, Amos Hayes wrote:
> Has there ever been any discussion about a new name? It doesn't matter to
> me but it would seem that with the current discussions about promotion and
> competition, it might help to have a more "public friendly" name. It is
> not obvious to me what a "postgres" is nor what it would do if I were to
> install it. Granted that "oracle", "informix", and "sybase" are all a
> little strange too, but they give hints (oracle, inform, base) about
> containing knowledge. They also seem to roll off the tongue a little
> easier.
Actually, I think Postgres originally came out of the Ingres family,
which is still around. It is an odd name, but sometimes odd names get
remembered. My company has a commercial database that is called
"Diogenes", and it gets remembered because it is so different from other
databases that are similar (like Medline or other healthcare related
online databases).
Brett W. McCoy
http://www.lan2wan.com/~bmccoy
-----------------------------------------------------------------------
"The Number of UNIX installations has grown to 10, with more expected."
-- The UNIX Programmer's Manual, 2nd Edition, June, 1972
On Wed, 22 Jul 1998, Brett W. McCoy wrote:
> On Wed, 22 Jul 1998, Amos Hayes wrote:
>
> > Has there ever been any discussion about a new name? It doesn't matter to
> > me but it would seem that with the current discussions about promotion and
> > competition, it might help to have a more "public friendly" name. It is
> > not obvious to me what a "postgres" is nor what it would do if I were to
> > install it. Granted that "oracle", "informix", and "sybase" are all a
> > little strange too, but they give hints (oracle, inform, base) about
> > containing knowledge. They also seem to roll off the tongue a little
> > easier.
>
> Actually, I think Postgres originally came out of the Ingres family,
Other way around, actually...Ingres is a branch off the original
Postgres source code...
> On Wed, 22 Jul 1998, Brett W. McCoy wrote: > > > On Wed, 22 Jul 1998, Amos Hayes wrote: > > > > > Has there ever been any discussion about a new name? It doesn't matter to > > > me but it would seem that with the current discussions about promotion and > > > competition, it might help to have a more "public friendly" name. It is > > > not obvious to me what a "postgres" is nor what it would do if I were to > > > install it. Granted that "oracle", "informix", and "sybase" are all a > > > little strange too, but they give hints (oracle, inform, base) about > > > containing knowledge. They also seem to roll off the tongue a little > > > easier. > > > > Actually, I think Postgres originally came out of the Ingres family, > > Other way around, actually...Ingres is a branch off the original > Postgres source code... University Ingres was first, then commercial Ingres was a split-off. PostgreSQL is based on University Ingres. -- Bruce Momjian | 830 Blythe Avenue maillist@candle.pha.pa.us | Drexel Hill, Pennsylvania 19026 + If your life is a hard drive, | (610) 353-9879(w) + Christ can be your backup. | (610) 853-3000(h)
On Wed, 22 Jul 1998, Bruce Momjian wrote:
> University Ingres was first, then commercial Ingres was a split-off.
> PostgreSQL is based on University Ingres.
That's what I thought...
Brett W. McCoy
http://www.lan2wan.com/~bmccoy
-----------------------------------------------------------------------
"The Number of UNIX installations has grown to 10, with more expected."
-- The UNIX Programmer's Manual, 2nd Edition, June, 1972
On Wed, 22 Jul 1998, Bruce Momjian wrote:
> > On Wed, 22 Jul 1998, Brett W. McCoy wrote:
> >
> > > On Wed, 22 Jul 1998, Amos Hayes wrote:
> > >
> > > > Has there ever been any discussion about a new name? It doesn't matter to
> > > > me but it would seem that with the current discussions about promotion and
> > > > competition, it might help to have a more "public friendly" name. It is
> > > > not obvious to me what a "postgres" is nor what it would do if I were to
> > > > install it. Granted that "oracle", "informix", and "sybase" are all a
> > > > little strange too, but they give hints (oracle, inform, base) about
> > > > containing knowledge. They also seem to roll off the tongue a little
> > > > easier.
> > >
> > > Actually, I think Postgres originally came out of the Ingres family,
> >
> > Other way around, actually...Ingres is a branch off the original
> > Postgres source code...
>
> University Ingres was first, then commercial Ingres was a split-off.
> PostgreSQL is based on University Ingres.
I stand corrected...I didn't even *know* there was a University
Ingres :(
On Wed, 22 Jul 1998, Herouth Maoz wrote:
> Not that I want to offend the Postgres developers, who are doing a great
> job. But in the same way that commercial enterprises think in terms of
> profit, volunteer workers think of it like this: Does it interest me to
> implement this feature? Do I have the time? Do I have the expertise?
I don't believe any offence could be taken by this...my question
back is are those features that are currently missing of such an import to
you that you'd be willing to pay one of the developers to take the time to
focus on what you require?
Marc G. Fournier
Systems Administrator @ hub.org
primary: scrappy@hub.org secondary: scrappy@{freebsd|postgresql}.org
At 21:12 +0300 on 22/7/98, Bruce Momjian wrote: > University Ingres was first, then commercial Ingres was a split-off. > PostgreSQL is based on University Ingres. Which also explains the name... Remember the old lesson in data structures about pre-order, in-order and post-order traversal of binary trees? Well, if you have an in-gres, then you can also have a pre-gres and a post-gres. Pre-gres would imply that it is older, less advanced than ingres. Hence Postgres. QED :-) Herouth -- Herouth Maoz, Internet developer. Open University of Israel - Telem project http://telem.openu.ac.il/~herutma
At 4:23 +0300 on 24/7/98, The Hermit Hacker wrote: > I don't believe any offence could be taken by this...my question > back is are those features that are currently missing of such an import to > you that you'd be willing to pay one of the developers to take the time to > focus on what you require? If I have to pay the authors for doing what I want, I am going to buy Informix or Oracle tomorrow... As I said, Postgres's main merit is that it's free. Herouth -- Herouth Maoz, Internet developer. Open University of Israel - Telem project http://telem.openu.ac.il/~herutma
On Sun, 26 Jul 1998, Herouth Maoz wrote:
> At 4:23 +0300 on 24/7/98, The Hermit Hacker wrote:
>
>
> > I don't believe any offence could be taken by this...my question
> > back is are those features that are currently missing of such an import to
> > you that you'd be willing to pay one of the developers to take the time to
> > focus on what you require?
>
> If I have to pay the authors for doing what I want, I am going to buy
> Informix or Oracle tomorrow... As I said, Postgres's main merit is that
> it's free.
Nobody said anything about *having* to pay for anything. The
question was whether or not ppl feel that a particular unsupported feature
was important enough to them, *right now* to cover the costs of paying a
contract programmer to do it *right now*.
Its a matter of some company saying "I need this feature right
now, and am willing to throw X dollars at it to get it moved to a higher
priority position to get it done" vs. "right now, this feature isn't
scheduale till v6.4 or v6.5 or...etc"...
Marc G. Fournier
Systems Administrator @ hub.org
primary: scrappy@hub.org secondary: scrappy@{freebsd|postgresql}.org
Um - let me get this straight... you want to go buy Oracle instead of kicking in a few bucks to pay someone to add it to PostgreSQL. OK then a quick call to Oracle would tell you that it's $295 per user, 5 user minimum. If you want to use it on the web for public use that's 20 minimum or about $6,000. Plus they suggest getting their application server for another $195 / user - pushing your web site up by another $4,000. Go ahead. I know if PostgreSQL did almost everything I needed I would rather: A) Write it myself B) Ask about it on the list and see if other people know how to do it C) Expense out $500 or $1000 to contribute to PostgreSQL and then let the company write it off. in that order ;-) Chris On Sun, 26 Jul 1998, Herouth Maoz wrote: > At 4:23 +0300 on 24/7/98, The Hermit Hacker wrote: > > > > I don't believe any offence could be taken by this...my question > > back is are those features that are currently missing of such an import to > > you that you'd be willing to pay one of the developers to take the time to > > focus on what you require? > > If I have to pay the authors for doing what I want, I am going to buy > Informix or Oracle tomorrow... As I said, Postgres's main merit is that > it's free. > > Herouth > > -- > Herouth Maoz, Internet developer. > Open University of Israel - Telem project > http://telem.openu.ac.il/~herutma > > >
At 19:28 +0300 on 27/7/98, Chris Johnson wrote: > Um - let me get this straight... you want to go buy Oracle instead of > kicking in a few bucks to pay someone to add it to PostgreSQL. > > OK then a quick call to Oracle would tell you that it's $295 per user, 5 > user minimum. If you want to use it on the web for public use that's 20 > minimum or about $6,000. Plus they suggest getting their application > server for another $195 / user - pushing your web site up by another > $4,000. I'll skip their application server. $6000 for Oracle? Sounds awfully cheap to me. You get the benefit of all those features for which I'd have to pay the Postgres creators, who in one year decide they want to take a vacation in Timbuktu, and their features will go with them... Won't happen? In the last couple of weeks I've seen a dozen questions pertaining to Postgres's object capabilities, such as how to cleanly insert values of a contained type and how to select them back. Up to this minute, nobody answered. To me, this indicates that the "O" in PostgreSQL's ORDBMS claim is no longer maintained. When you rely on an organization to maintain something, you know that even if someone gets married or dies in a car accident, your application will continue to be supported. If I pay an individual to do it, can you make the same claim? Besides, there's no way I could get away with paying an individual any sum of money. It's not my money - it's the university's. They will pay organizations, not individuals - unless the individual would like to sign a contract or something. And come the next day, I need another feature, I need to pay yet another individual. And yet another. Never mind having a Postgres version which nobody else has, meaning I won't be able to apply patches as they are posted for the main version - or should I pay *all* the Postgres developers so that they will all finish development, testing and beta to make everybody's version the same as mine? -- Herouth Maoz, Internet developer. Open University of Israel - Telem project http://telem.openu.ac.il/~herutma
> > Um - let me get this straight... you want to go buy Oracle instead of > > kicking in a few bucks to pay someone to add it to PostgreSQL. > > > > OK then a quick call to Oracle would tell you that it's $295 per user, 5 > > user minimum. If you want to use it on the web for public use that's 20 > > minimum or about $6,000. Plus they suggest getting their application > > server for another $195 / user - pushing your web site up by another > > $4,000. > > I'll skip their application server. $6000 for Oracle? Sounds awfully cheap > to me. You get the benefit of all those features for which I'd have to pay > the Postgres creators, who in one year decide they want to take a vacation > in Timbuktu, and their features will go with them... The application server increases performance by keeping connections open. They were very vague about exactly what it did, but I am sure that you could skip it. But the $6,000 I was quoted was for the workgroup server - not the Enterprise version that has all the advanced stuff. For example there is no Incremental backup, no parallel backup and recovery, no advanced replication. If you want those features you need the "Enterprise Edition" which is significantly more money. As for features becoming unsupported when someone takes 'a vacation in Timbuktu' - has that happened to Linux? Has it happened to perl? What about your great commercial program when the vendor goes belly up - don't the features go away then? I truly believe in Open Source software and I trust the authors of OSS more than their commercial counterparts, but that's just me. > Won't happen? In the last couple of weeks I've seen a dozen questions > pertaining to Postgres's object capabilities, such as how to cleanly insert > values of a contained type and how to select them back. Up to this minute, > nobody answered. To me, this indicates that the "O" in PostgreSQL's ORDBMS > claim is no longer maintained. Not necessarily - the reply might have gone to the user directly. > When you rely on an organization to maintain something, you know that even > if someone gets married or dies in a car accident, your application will > continue to be supported. If I pay an individual to do it, can you make the > same claim? Bullsh*t - Does MS still support DOS? Does MS still support Windows 3.X? How about other vendors... does Lotus still support 1-2-3? Does Ashton-Tate still support dBase? Also note that neither I nor the person that posed the original question to you suggested that a person individually be paid to develop the feature. I'm suggesting that sending a small amount of money to someone could be used to motivate having them put off other "for pay" work to do the work on the feature for Postgres. Putting the money aside for a moment I believe someone else already asked you what features you thought were missing. Maybe the features you want are probably already scheduled for development. I suspect that many if not most are indeed somewhere in the pipe. > Besides, there's no way I could get away with paying an individual any sum > of money. It's not my money - it's the university's. They will pay > organizations, not individuals - unless the individual would like to sign a > contract or something. And come the next day, I need another feature, I > need to pay yet another individual. And yet another. OK - get the university to donate $1,000 to the 'PostgreSQL Global Development Group' and let them write a check for the same amount to a developer as I mentioned above. Or make it part of a software development class. I know I would have loved to take part in helping develop something like Postgres as part of my database systems class. > Never mind having a Postgres version which nobody else has, meaning I won't > be able to apply patches as they are posted for the main version - or > should I pay *all* the Postgres developers so that they will all finish > development, testing and beta to make everybody's version the same as mine? Come on - be reasonable... The person that asked if you would be willing to pay some money to get the development of features you want was not suggesting that you would have a special version of PostgreSQL. Any additions made would wind up back in Postgres itself for everyone to use. Now since you have avoided the question posed by that other person I will ask again. What feature or features were you looking for? Chris (not a Postgres developer, but annoyed enough to reply)
> > should I pay *all* the Postgres developers so that they will all finish > > development, testing and beta to make everybody's version the same as mine? > > Come on - be reasonable... The person that asked if you would be willing > to pay some money to get the development of features you want was not > suggesting that you would have a special version of PostgreSQL. Any > additions made would wind up back in Postgres itself for everyone to use. > Actually, that is standard operating procedures for most software developers whethor or not they do it in the Open Source paradigm. Its just too much of a hassle for the developer to keep track of multiple source streams. It can be done, but it is a monumental waste of time and money. In the end no one is happy. A few times in my career there have been opportunities to get certain features added to a product, an d it was always understood that in the next release the features would be there anyway. The key thing was that the company I was working for knew hot change the software companies priorities. Though, PostgreSQL is a Open Source organization, it is still an organization inhabited by living breathing individuals who require finances to survive. So, all the getlemen was asking was to help him change some of his developers priorities, because in most cases its awefully hard to turn down a pay for job, in place of free one. ...james > Now since you have avoided the question posed by that other person I will > ask again. What feature or features were you looking for? > > Chris > (not a Postgres developer, but annoyed enough to reply)
At 20:20 +0300 on 27/7/98, Chris Johnson wrote: > Come on - be reasonable... The person that asked if you would be willing > to pay some money to get the development of features you want was not > suggesting that you would have a special version of PostgreSQL. Any > additions made would wind up back in Postgres itself for everyone to use. > > Now since you have avoided the question posed by that other person I will > ask again. What feature or features were you looking for? He didn't ask me what the features were. I'm quite willing to specify them. Yes, I'm quite aware that many of them are in the pipeline... But hey, if I install Oracle/Informix, I'd have them all *now*, tested and debugged by many users before me. Functionality: - Row-level locking. - Transaction isolation levels. - PL/SQL - triggers not in C - inner/outer joins - standard referetial integrity - BLOBs with less hassle Administration: - Convenient tools for backing up, including scheduling of backups. Same for vacuum and any other periodical maintenance. - Backups include BLOBs. - Replication - Support for raw devices (my sysadmin prefers it). Those are just off the top of my head (I didn't have a chance to ask my sysadmin at length). Herouth -- Herouth Maoz, Internet developer. Open University of Israel - Telem project http://telem.openu.ac.il/~herutma
On Tue, 28 Jul 1998, Herouth Maoz wrote:
> He didn't ask me what the features were. I'm quite willing to specify them.
> Yes, I'm quite aware that many of them are in the pipeline... But hey, if I
> install Oracle/Informix, I'd have them all *now*, tested and debugged by
> many users before me.
>
> - Convenient tools for backing up, including scheduling of
> backups. Same for vacuum and any other periodical maintenance.
What to give me pointers to these? We use Oracle here, and the
SOP from before I got here was to shutdown Oracle, backup the system and
then restart Oracle up again. We're moving our backups to a centralized
system, and using Oracle modules for doing this, but from our DBA's
perusal of the Oracle documentation, there is nothing "convienent" about
setting it up...
> - Support for raw devices (my sysadmin prefers it).
This one I just checked about, and Oracle still appears to
recommend using raw devices, as they claim it can be up to 50%
faster...but, how would one implement this in PostgreSQL? So far, I
believe you are the only one that is asking for it, so don't hold your
breath on it ever getting done, but I'm curious, unless you wanted to
implement it yourself...
Under Oracle, its easy, since you 'pre-size' your tables when you
create your databases, and the tables will not grow beyond that size.
With PostgreSQL, there is no bounds, except hard disk, to how large your
tables will grow.
I don't know how Oracle implements this, as we don't use it
here...
The other problem with trying to implement RAW devices, and,
granted, I could be over cmplicating it, but how do you implement it
across X operating systems running Y platforms? Doesn't each of them
access drives differently? And, in some cases, multiply that by two for
IDE vs SCSI...or...vs...?
At 14:46 +0300 on 28/7/98, The Hermit Hacker wrote: > On Tue, 28 Jul 1998, Herouth Maoz wrote: > > > He didn't ask me what the features were. I'm quite willing to specify them. > > Yes, I'm quite aware that many of them are in the pipeline... But hey, if I > > install Oracle/Informix, I'd have them all *now*, tested and debugged by > > many users before me. > > > > - Convenient tools for backing up, including scheduling of > > backups. Same for vacuum and any other periodical maintenance. > > What to give me pointers to these? We use Oracle here, and the > SOP from before I got here was to shutdown Oracle, backup the system and > then restart Oracle up again. We're moving our backups to a centralized > system, and using Oracle modules for doing this, but from our DBA's > perusal of the Oracle documentation, there is nothing "convienent" about > setting it up... Personally, I don't know. All I know is that our computer center backs up anything, even if it doesn't need backing up. Since we don't use Oracle as yet, I can only give you the hearsay. That is, that most people seem to be satisfied by Oracle's backup facilities. > > - Support for raw devices (my sysadmin prefers it). > > This one I just checked about, and Oracle still appears to > recommend using raw devices, as they claim it can be up to 50% > faster...but, how would one implement this in PostgreSQL? So far, I > believe you are the only one that is asking for it, so don't hold your > breath on it ever getting done, but I'm curious, unless you wanted to > implement it yourself... I didn't. That was the whole point, wasn't it? I expect nothing from a free database. Whatever is given, is great. PostgreSQL is far better featured than other freebies, and that's why I use it. Functionality is missing? Tough baby for me. But if there is functionality that I miss, and the commercial product offers, then Postgres is no match. The only thing that stands for it is the price. So, eventually, you weigh functionality, reliability, cross-product compatibility, on-site support, localization and training - against price. Oracle is a much better, more mature RDBMS. Informix, too. Years will pass before Postgres catches up - if ever. For its price, it's great. Trying to compete against the commercial products by merit of functionality is not serious. On the other issue - about raw devices: the reason why it is supposed to work better is because it cuts on the overhead of the system library calls and implementation. That makes sense. There was a claim that the system's library calls are optimized for the given hardware, and therefore make a better solution. The truth is, however, that these calls and data structures are optimized to handle relatively small files. This makes sense, if I recall correctly how INodes work (three levels of indirection for large files, wasn't it?) True, raw devices require that the programmer will be able to anticipate which devices will be used. But Oracle runs on almost every platform, does it not? So it's possible to do it. Herouth -- Herouth Maoz, Internet developer. Open University of Israel - Telem project http://telem.openu.ac.il/~herutma
> The other problem with trying to implement RAW devices, and, > granted, I could be over cmplicating it, but how do you implement it > across X operating systems running Y platforms? Doesn't each of them > access drives differently? And, in some cases, multiply that by two for > IDE vs SCSI...or...vs...? That is the real problem. I have to assume whenever you start implementing raw devices, you go far far away from DBMS design to OS design. Perhaps you do not have to write the process control part of an OS, but (I mean this jokingly) in one's arrogance one must think you can access the HD's more efficiently than the OS can. Hey, maybe some can do this, and have enough knowledge about HD's and controllers to do this, but making this portable is got to a _lot_ of work. Not that it couldn't be done, or that it wouldn't be eventually a good idea (*though I am not certain about that)...james
Hi,
let me add one thing here ... (if this comes a second time, of missed
the first one ...)
Wouldn't it be possible to get some backing from one of the
commercial Linux distribution makers? Like Redhat supports Gnome or
S.u.S.e supports XFree? What about GNU? Anyone listening?
This would be major leap for PostgreSQL if someone financed two
months of Vadim for example ...
Ciao
Ulrich
Ciao
Das Boersenspielteam.
---------------------------------------------------------------------------
http://www.boersenspiel.de
Das Boersenspiel im Internet
*Realitaetsnah* *Kostenlos* *Ueber 20.000 Spieler*
---------------------------------------------------------------------------
On Tue, 28 Jul 1998, Boersenspielteam wrote:
> Hi,
>
> let me add one thing here ... (if this comes a second time, of missed
> the first one ...)
>
> Wouldn't it be possible to get some backing from one of the
> commercial Linux distribution makers? Like Redhat supports Gnome or
> S.u.S.e supports XFree? What about GNU? Anyone listening?
RedHat supports Gnome vs KDE because Gnome is GPLd. GNU would
require a GPLd source code. We fall under Berkeley's CopyRight, not GPLs
CopyLeft, and we're quite happy with that...
> This would be major leap for PostgreSQL if someone financed two
> months of Vadim for example ...
As for '2 months' of Vadim...its more difficult for 1
person/company to pay for '2 months' of his time (or Bruce's or Thomas' or
Byron's, or...) then it would be for 10 ppl to share that financing...
I'm working on something that might make that a reality, hopefully
have something to "announce" early next week...
> That is the real problem. I have to assume whenever you start implementing raw > devices, you go far far away from DBMS design to OS design. Perhaps you do not > have to write the process control part of an OS, but (I mean this jokingly) in > one's arrogance one must think you can access the HD's more efficiently than the > OS can. Hey, maybe some can do this, and have enough knowledge about HD's and > controllers to do this, but making this portable is got to a _lot_ of work. > > Not that it couldn't be done, or that it wouldn't be eventually a good idea > (*though I am not certain about that)...james this reminds me so much of the multi threading issue Sybase faced. When they were adding support for multi threading, every Unix has different level of support that they end up added their own threading engine. I for one, don't see RAW device support in Postgres has a "critial" missing feature, in my thinking, there are other things that are more important. Though for something like RDBMS, where each person's need could be so different, it truly is YMMV.
> Hi, > > let me add one thing here ... (if this comes a second time, of missed > the first one ...) > > Wouldn't it be possible to get some backing from one of the > commercial Linux distribution makers? Like Redhat supports Gnome or > S.u.S.e supports XFree? What about GNU? Anyone listening? > > This would be major leap for PostgreSQL if someone financed two > months of Vadim for example ... That would be great. I feel like I am standing still when Vadim is working. -- Bruce Momjian | 830 Blythe Avenue maillist@candle.pha.pa.us | Drexel Hill, Pennsylvania 19026 + If your life is a hard drive, | (610) 353-9879(w) + Christ can be your backup. | (610) 853-3000(h)
=>From: Bruce Momjian <maillist@candle.pha.pa.us> =>Date: Wed, 29 Jul 1998 01:01:05 -0400 (EDT) =>... =>> This would be major leap for PostgreSQL if someone financed two =>> months of Vadim for example ... => =>That would be great. I feel like I am standing still when Vadim is =>working. Hi. We're currently using Oracle for our Web stuff, but their licensing program is byzantine, and the price (if you can figure it out) starts at outrageous and goes up. We'd be very interested in seeing PostgreSQL become a viable alternative for our applications, and would be willing to put up some money for it. What's the best way to set something like this up? There's a few features we'd need to be assured of getting for our investment, but other than that, fronting money for free software development is not a problem (depending on the amount, of course :-) regards, d.
On Wed, 29 Jul 1998, dave madden wrote:
> =>From: Bruce Momjian <maillist@candle.pha.pa.us>
> =>Date: Wed, 29 Jul 1998 01:01:05 -0400 (EDT)
> =>...
> =>> This would be major leap for PostgreSQL if someone financed two
> =>> months of Vadim for example ...
> =>
> =>That would be great. I feel like I am standing still when Vadim is
> =>working.
>
> Hi. We're currently using Oracle for our Web stuff, but their
> licensing program is byzantine, and the price (if you can figure it
> out) starts at outrageous and goes up. We'd be very interested in
> seeing PostgreSQL become a viable alternative for our applications,
> and would be willing to put up some money for it.
>
> What's the best way to set something like this up? There's a few
> features we'd need to be assured of getting for our investment, but
> other than that, fronting money for free software development is not a
> problem (depending on the amount, of course :-)
By the end of this weekend, I will have an announcement up on the
WWW and here concerning this...I want to build the pages and get the -core
group to review and comment on it before I elaborate further in public...
Patience :)