Обсуждение: RE: [GENERAL] info on people running postgre
Hi, We have tryed PostgreSQL as the backend of a Java application used to collect orders and other information. Everything would have been fine but for the lack of row-level locking, and some problems with processes that go hanged when locked for to long. Also we have experienced a couple of corruptions on a table that suffers the deletion and insertion of 46000 records every day, and some sporadic and misterious backend crashes. Table-locking has forced us to quit from PostgreSQL and move to Oracle. So if you need a read-only (or single user) database with moderate performance (any comercial one is faster), PostgreSQL will probably give you what you need in this other aspects: good SQL, ease of use and administration, really thin JDBC type 4 driver; and you may find to need more of: Export/import tools Backend stability Crash recovery Programming interfaces doc. Best luck, Juan Alvarez Ferrando
On Thu, 25 Mar 1999, Juan Alvarez Ferrando wrote: > > collect orders and other information. Everything would have been fine but > for the lack of row-level locking, and some problems with processes that go > I'm not yet doing anything in a production environment yet, but the next version of postgresql does has something more advanced than row level locking, MVCC. I've been running the CVS version during development and it seems quite a bit faster than 6.4.2 as well. I'm seriously thinking about scraping my Oracle DB at work and replacing it with postgresql. Why? One...Oracle is on site license so I'm not out any money if I switch :-) Two...Oracle is an absolute hog when it comes to memory and disk space. Here is a look at top viewing an idle Oracle DB on a Solaris system PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND 6174 oradba 23 0 33M 12M sleep 0:00 0.18% 0.12% oracle 6166 oradba 33 0 35M 12M sleep 0:00 0.16% 0.11% oracle 6168 oradba 33 0 33M 10M sleep 0:00 0.13% 0.08% oracle 6170 oradba 33 0 33M 10M sleep 0:00 0.13% 0.08% oracle 6176 oradba 27 0 33M 10M sleep 0:00 0.08% 0.05% oracle 6172 oradba 33 0 33M 10M sleep 0:00 0.08% 0.05% oracle 351 oradba 33 0 11M 2944K sleep 0:00 0.03% 0.02% tnslsnr Three...By adopting postgresql as my primary db I hope to help the free software community by dealing with problems as they come up and submitting problem reports. The people on the dev team have always been helpful so I'm not too worried that I'll get burned. On a related note. I'm currently working with a local company to automate their entire business process and GPL the resulting code. Postgresql is currently the data storage system of the backend. We are betting that a freeware solution with (hopefully) free software community support will provide a better solution in the long run than a propritary one. So you will not be alone in depending upon postgresql. ->->->->->->->->->->->->->->->->->->---<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-< James Thompson 138 Cardwell Hall Manhattan, Ks 66506 785-532-0561 Kansas State University Department of Mathematics ->->->->->->->->->->->->->->->->->->---<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<
On Thu, 25 Mar 1999, Juan Alvarez Ferrando wrote: > Hi, > > We have tryed PostgreSQL as the backend of a Java application used to > collect orders and other information. Everything would have been fine but > for the lack of row-level locking, and some problems with processes that go > hanged when locked for to long. Also we have experienced a couple of > corruptions on a table that suffers the deletion and insertion of 46000 > records every day, and some sporadic and misterious backend crashes. > > Table-locking has forced us to quit from PostgreSQL and move to Oracle. So > if you need a read-only (or single user) database with moderate performance > (any comercial one is faster), PostgreSQL will probably give you what you > need in this other aspects: good SQL, ease of use and administration, really > thin JDBC type 4 driver; and you may find to need more of: > > Export/import tools pg_dump and psql > Backend stability would love to see more information on this .. > Crash recovery > Programming interfaces doc. huh? Marc G. Fournier ICQ#7615664 IRC Nick: Scrappy Systems Administrator @ hub.org primary: scrappy@hub.org secondary: scrappy@{freebsd|postgresql}.org
The criticisms were pretty accurate overall, and to the point. The Hermit Hacker wrote: > > On Thu, 25 Mar 1999, Juan Alvarez Ferrando wrote: > > > Hi, > > > > We have tryed PostgreSQL as the backend of a Java application used to > > collect orders and other information. Everything would have been fine but > > for the lack of row-level locking, and some problems with processes that go > > hanged when locked for to long. Also we have experienced a couple of > > corruptions on a table that suffers the deletion and insertion of 46000 > > records every day, and some sporadic and misterious backend crashes. > > > > Table-locking has forced us to quit from PostgreSQL and move to Oracle. So > > if you need a read-only (or single user) database with moderate performance > > (any comercial one is faster), PostgreSQL will probably give you what you > > need in this other aspects: good SQL, ease of use and administration, really > > thin JDBC type 4 driver; and you may find to need more of: > > > > Export/import tools > > pg_dump and psql Insufficient. Does not dump and restore everything in the DB. I see constant complaints about this in these mailing lists. > > > Backend stability > > would love to see more information on this .. > Run a load test and longevity test. Enter the scenario with the mindset of "I want to see what I can do to break this." I'll bet that if I started from scratch, I could crash any PGSQL implemenation in minutes, unless you folks have done SERIOUS improvements since 6.3.2. (PS - a longevity test is something that should take more than 5 minutes ;-)) > > Crash recovery Consider transaction logging... > > Programming interfaces doc. > > huh? > Well - how good is your overall documentation? API documentation does exist, but lots of other stuff is missing. Explain to me where I can find an explanation of server error log entries, such as: Failed Assertion("!(hctl->nkeys > 0):", File: "dynahash.c", Line: 564) !(hctl->nkeys > 0) (0) > Marc G. Fournier ICQ#7615664 IRC Nick: Scrappy > Systems Administrator @ hub.org > primary: scrappy@hub.org secondary: scrappy@{freebsd|postgresql}.org -- ------------------------------------------------------------ Thomas Reinke Tel: (416) 460-7021 Director of Technology Fax: (416) 598-2319 E-Soft Inc. http://www.e-softinc.com