The future of Java
in [Java Programming]
|
From: Arne Vajhøj on 25 Nov 2009 21:41 Lew wrote: > Arne Vajhøj wrote: >> scalablity - largest public known MySQL customers are bigger >> than largest known PostgreSQL customers > > As of May, 2008, supposedly the largest database installation in the > world was on Postgres. > <http://it.toolbox.com/blogs/oracle-guide/worlds-largest-database-runs-on-postgres-24979> Read the fine print. That database is not really a PostgreSQL database. It it is a custom database backend. They only use PostgreSQL as frontend. So if we can agree that the scalability is in getting the data on and off the disk and not in parsing SQL statements, then that does not say anything about PostgreSQL's scalability. Well it does say that PostgreSQL backend is not suited for PB data sizes. But then neither are any other relational databases. Various big table & map reduce concept exists to handle those data sizes. Interesting from a Java perspective is that HDFS & Hadoop are by far the most widely used technologies for this. BTW, accidentally Yahoo is a big MySQL customer. One can wonder whether they picked PostgreSQL for the above just to get the SQL frontend or because even though MySQL comes with multiple storage engines, then they really did consider the API for those to be good. Arne
From: Arved Sandstrom on 26 Nov 2009 06:20
Arne Vajhøj wrote: > Arved Sandstrom wrote: >> To me a high-end database concentrates on features like data >> integrity, scalability, reliability, solid procedural language >> support, constraints, serious transactional support etc etc, all of >> which are things that PostgreSQL has been working on since the >> beginning, and all of which are MySQL after-thoughts. > [ SNIP ] > Based on what yoy say it seems as if the distrust of MySQL is > based on 3 things: > - what MySQL were missing 10 years ago > - the ability to use Tcl, Perl and Python for USP's/UDF's/triggers > - lack of understanding of where MySQL is used > > Not convincing. I daresay the distrust of MySQL doesn't go back as far as 10 years. There are simply too many reports of problems, right across the board, with releases prior to 5.0 (2005). In fact 5.0 is probably the MySQL high point; 5.1 last year under the auspices of Sun was accompanied by reports of poor quality control and these reports have not gone away. >> MySQL has storage engines out the ying-yang, and PostgreSQL has *one*. >> This allows the Postgres developers to concentrate on quality, IMHO. > > That does not make any sense. Focus on one thing and do it well. Makes sense to me. > It is not the same people that does MyISAM and InnODB. It is not > even the same company. > > (not at least until EU approves Oravle buying SUN) > >> Ultimately it boils down to benchmarking your own application in >> realistic field conditions. To me high-end also means that you've got >> a person or people who qualify as DBAs - it's their #1 job to >> administer the database of choice. I'll bet a lot of the benchmarks >> that highlight MySQL over PostgreSQL are set up by folks who don't >> class as DBAs for either, or just for MySQL. In any case I'm willing >> to bet that for a "real" high-end app in realistic conditions that >> Postgres will win out. > > The CIO's of the big internet companies seems to bet differently. > > Arne MySQL took off and PostgreSQL and Firebird (to highlight 2 competitors) did not. There is no rhyme or reason as to why the market, especially users of free stuff, and especially hundreds and thousands and tens of thousands of non-professional DBA users, tends to pick one database over another. The LAMP notion however had a great deal to do with it in this case - this was marketing genius, and ten or so years ago that stack was something that every hacker in a single room could get a website up and running with...right around the time of the bubble. MySQL has never looked back - they got lifted right up on that rising tide. Once you generate that kind of user base, a large percentage of your "DBAs" and "system architects" and "CIOs" are people who grew up with (and I mean that literally) MySQL. They for the most part haven't seriously used other databases. They are so used to the workarounds for MySQL warts, of which there are still many, that they forget how many there are. There are so many MySQL users out there that there are also more MySQL experts than there are, say, PostgreSQL or Firebird experts [1]; this in and of itself is a selling point, because even if MySQL does need a lot of help in order to develop a large installation, there's no shortage of relatively inexpensive people to help you do it. I can fully understand why CIOs of big Internet companies go with MySQL. And few of the reasons for why they do so necessarily have anything to do with MySQL being the best database for the job (except for human resourcing). AHS [1[ For all I know there are more MySQL experts than there are Postgres users. |