Path: csiph.com!x330-a1.tempe.blueboxinc.net!newsfeed.hal-mli.net!feeder1.hal-mli.net!weretis.net!feeder4.news.weretis.net!newsfeed.utanet.at!newscore.univie.ac.at!aconews-feed.univie.ac.at!aconews.univie.ac.at!not-for-mail From: "Laurenz Albe" Newsgroups: comp.databases.postgresql References: <1308640710.210659@proxy.dienste.wien.at><1308738638.232318@proxy.dienste.wien.at><1308911383.819034@proxy.dienste.wien.at><1309166263.561072@proxy.dienste.wien.at><1309250071.981133@proxy.dienste.wien.at> Subject: Re: Is PostgreSQL good? Date: Wed, 29 Jun 2011 09:50:32 +0200 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 6.00.2900.5931 X-RFC2646: Format=Flowed; Original X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.6109 Organization: dienste.wien.at ISP Message-ID: <1309333854.777629@proxy.dienste.wien.at> X-Cache: nntpcache 2.3.3 (see http://www.nntpcache.org/) Lines: 61 NNTP-Posting-Host: 141.203.254.23 X-Trace: 1309333858 aconews.univie.ac.at 73248 141.203.254.23 X-Complaints-To: abuse@univie.ac.at Xref: x330-a1.tempe.blueboxinc.net comp.databases.postgresql:162 Mladen Gogala wrote: >> PostgreSQL has some of these things. >> Even if Oracle has some performance related features that PostgreSQL >> does not, that does not prove that it will always perform better. I'm >> sure that Oracle will sometimes be faster than PostgreSQL, but I don't >> accept a general statement like that. > Look no further than the Postgresql performance mailing list. At least > once a month, there is a question why does Oracle do this or that faster > than Postgres. If you asked users with the experience in both worlds, > what do you think that the answer would be, which one is faster? That argument is flawed. Only people who have a problem will write a question to the list. > This was supposed to be a discussion on the > technical merits of hints, yet there is no answer to the main points I've > made. Let me repeat them once again: > - Humans are smarter than the optimizer and have better knowledge of > the data. Therefore, they should have the possibility to override the > decisions made by the optimizer. > - Hints are needed, as the last resort, but they are. When there is no > other way of getting the desired performance, hints must be available > as the method of the last resort. > - Database must offer a tuning methodology, things that user can do to > improve performance. You start with a SQL, you run explain analyze and > then there must be a series of steps to take. In PostgreSQL, there > isn't such a methodology. > - Such refusal to provide a method to override the optimizer decisions > reflects deep mistrust of the database developers toward their users. All but the third of these points are not about technical merits. #1 is philosophy. I'd agree that humans are smarter than machines in principle, but I wouldn't enter a number crunching contest with one. Nor do I personally want to enter a query optimizing contest with PostgreSQL's optimizer. #2 is an opinion. What should I answer? #4 sounds like a conspiracy theory. I agree with #3 as far that a database must offer tools for tuning. I have named some that PostgreSQL has. Query tuning is always a bit of an art, don't pretend that there's a step-by-step foolproof method in Oracle. I may look like that to you because you know it well. >> That's ok, share your experiences. >> But if you want your opinion to be heard, refrain from insult and >> sweeping generalizations with no more proof than your authority. > I don't think that I am getting my message across here. I don't care > about being heard by the Postgres community, that's way behind me. No, I meant be heard by people who post here to get advice and opinions on migrating from Oracle to PostgreSQL. Yours, Laurenz Albe