Re: sleepycat db VS MySQL or postgres
On 01/07/2013 22:28, Jim Pazarena wrote: I have a rather extensive series of databases created and in use all with the very old sleepycat db3. I believe in the addage don't fix what ain't broken, but in the case of db3, it IS broken and my db files get corrupted on occasion. I could move to db5 or db6 OR MySQL, or even postgres. I use simple primary key files, most entries are added from a CLI or termcap/curses screen. Some programatically. With about the same number of sequential dumps vs indexed random reads. I have no experience with the c interface for postgres or mysql, but also, do not know how much the c interface has changed for sleepycat 5/6 compared to the c interface for db3, which I understand quite well. So I am prepared for a learning curve irrespective of which platform I select. Records do not exceed much more than 10-20,000, with key sizes not much wider than 16 bytes (ipv4), 13 (mac), 32 (ipv6). And various smaller key sizes. Suggestions would be very much appreciated. Well, this is essentially a bikeshed thread... so why not chip in :) I'd say it depends on what is your priority or what do you want to achieve by switching databases. If you want it to be as easy as possible, switch to DB5 and you'll be ok. If you want to learn something interesting, try one of the recent NoSQL databases, such as Redis, MongoDB or CouchDB - they're like DBx but with significantly more powerful query capabilities. If you want to get a feel of how SQL databases work, go with PostgreSQL, but be aware that to really use a SQL database the way they're made to be used, you'll need to properly design a relational schema. Using them to store 20 KiB blobs indexed by a single key is way too simplistic and probably much slower than what you could get with a simple DBx engine. Also, SQL databases usually work with SQL queries, which are text, so you'll have a non-trivial task of fitting C structs in their text/blob field types - it's best to avoid it. Also, you'll need to learn how to tune and maintain proper database servers. If you want to just try SQL but without bothering with tuning and maintainance, try SQLite, but beware it is basically limited to a single writer (and inifinite reader) clients in the best case (with WAL journalling). signature.asc Description: OpenPGP digital signature
Re: sleepycat db VS MySQL or postgres
On Tue, 02 Jul 2013 05:45:47 -0500, Ivan Voras ivo...@freebsd.org wrote: Well, this is essentially a bikeshed thread... so why not chip in I disagree; all of these databases have distinctly different uses. MySQL/PostgreSQL: pick your poison. Relational databases. Will you have multiple users connecting to the database? Will there be lots of updates to the data? These are what you want. If you care about data integrity, I'd choose Postgres. SQLite: Do you want a relational database without needing a daemon to be running and will only have a single user/process accessing the database at one time? This is what you want. NoSQL: Do you want to dabble with the mess that is NoSQL so you can build your cloud? Don't care if other nodes aren't guaranteed to get the latest copy of the data? This is what you want. SleepyCat/BerkleyDB: Is your data WORM? (Write Once Read Many) If so, this is *ABSOLUTELY* what you want. If twitter was built upon a WORM database instead of MySQL they could host the entirety of twitter on a handful of servers instead of the gross MySQL+Cassandra mess they're fighting with today. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: sleepycat db VS MySQL or postgres
On 02/07/2013 13:55, Mark Felder wrote: If twitter was built upon a WORM database instead of MySQL they could host the entirety of twitter on a handful of servers instead of the gross MySQL+Cassandra mess they're fighting with today. I'd say their problem is not exactly solvable by only choosing a database :D They, like Facebook, have the problem of fanout, where a single piece of data goes into thousands of different user pages. Whatever they save in the raw data access operations will probably be relatively small compared to the horsepower needed to combine pages from all these fleeting data pieces. But yes, obviously a database designed specifically for one thing will be optimized for that thing. signature.asc Description: OpenPGP digital signature
Re: sleepycat db VS MySQL or postgres
On Tue, 02 Jul 2013 07:12:37 -0500, Ivan Voras ivo...@freebsd.org wrote: On 02/07/2013 13:55, Mark Felder wrote: If twitter was built upon a WORM database instead of MySQL they could host the entirety of twitter on a handful of servers instead of the gross MySQL+Cassandra mess they're fighting with today. I'd say their problem is not exactly solvable by only choosing a database :D They, like Facebook, have the problem of fanout, where a single piece of data goes into thousands of different user pages. Whatever they save in the raw data access operations will probably be relatively small compared to the horsepower needed to combine pages from all these fleeting data pieces. Good point :) I'm still sure it would work much, much better though. However, I'm just glad that's not *my* problem to fix. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
sleepycat db VS MySQL or postgres
I have a rather extensive series of databases created and in use all with the very old sleepycat db3. I believe in the addage don't fix what ain't broken, but in the case of db3, it IS broken and my db files get corrupted on occasion. I could move to db5 or db6 OR MySQL, or even postgres. I use simple primary key files, most entries are added from a CLI or termcap/curses screen. Some programatically. With about the same number of sequential dumps vs indexed random reads. I have no experience with the c interface for postgres or mysql, but also, do not know how much the c interface has changed for sleepycat 5/6 compared to the c interface for db3, which I understand quite well. So I am prepared for a learning curve irrespective of which platform I select. Records do not exceed much more than 10-20,000, with key sizes not much wider than 16 bytes (ipv4), 13 (mac), 32 (ipv6). And various smaller key sizes. Suggestions would be very much appreciated. -- Jim Pazarena fqu...@paz.bz ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: sleepycat db VS MySQL or postgres
On 1 July 2013 16:28, Jim Pazarena fqu...@paz.bz wrote: I could move to db5 or db6 OR MySQL, or even postgres. snip I have no experience with the c interface for postgres or mysql, but also, do not know how much the c interface has changed for sleepycat 5/6 compared to the c interface for db3, which I understand quite well. So I am prepared for a learning curve irrespective of which platform I select. Records do not exceed much more than 10-20,000, with key sizes not much wider than 16 bytes (ipv4), 13 (mac), 32 (ipv6). And various smaller key sizes. Suggestions would be very much appreciated. Jim - ultimately I'd recommend deciding which of the three you WANT to learn and then use it. I know nothing about db* but I use both MySQL and PostGreSQL on a regular basis. I like them both. I prefer the licence used by PostGreSQL, I prefer the PostGreSQL replication but I use whichever suits my needs. Some people are fanatical about one or the other, I say pick whichever you're more interested in and learn how to use it. kmw ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: sleepycat db VS MySQL or postgres
On 07/01/2013 01:28 PM, Jim Pazarena wrote: I have a rather extensive series of databases created and in use all with the very old sleepycat db3. I believe in the addage don't fix what ain't broken, but in the case of db3, it IS broken and my db files get corrupted on occasion. I could move to db5 or db6 OR MySQL, or even postgres. I use simple primary key files, most entries are added from a CLI or termcap/curses screen. Some programatically. With about the same number of sequential dumps vs indexed random reads. I have no experience with the c interface for postgres or mysql, but also, do not know how much the c interface has changed for sleepycat 5/6 compared to the c interface for db3, which I understand quite well. So I am prepared for a learning curve irrespective of which platform I select. Records do not exceed much more than 10-20,000, with key sizes not much wider than 16 bytes (ipv4), 13 (mac), 32 (ipv6). And various smaller key sizes. Suggestions would be very much appreciated. Jim, I'm a lazy bugger and what I'd do is knock together a small perl program using DBI and the DBD for Berkeley DB (sleepcat) and either MySQL or Postgres (pick your religion). You could grap the records out of one table and insert them into another as though they are the same db... LIke I say, I'm lazy. No, it's not shiney and new (may EVEN be deprecated) If you'd like assistance, I'll be happy to hold hands Bruce Ferrell ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: sleepycat db VS MySQL or postgres
On Mon, 01 Jul 2013 13:28:59 -0700, Jim Pazarena wrote: I have a rather extensive series of databases created and in use all with the very old sleepycat db3. I believe in the addage don't fix what ain't broken, but in the case of db3, it IS broken and my db files get corrupted on occasion. I could move to db5 or db6 OR MySQL, or even postgres. I use simple primary key files, most entries are added from a CLI or termcap/curses screen. Some programatically. With about the same number of sequential dumps vs indexed random reads. I have no experience with the c interface for postgres or mysql, but also, do not know how much the c interface has changed for sleepycat 5/6 compared to the c interface for db3, which I understand quite well. So I am prepared for a learning curve irrespective of which platform I select. Records do not exceed much more than 10-20,000, with key sizes not much wider than 16 bytes (ipv4), 13 (mac), 32 (ipv6). And various smaller key sizes. Suggestions would be very much appreciated. If you have to make a change, I'd recommend switching to Postgres. Worth the effort in the long run, and Postgres is not at all difficult if you have experience with any RDB. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: sleepycat db VS MySQL or postgres
On Mon, Jul 1, 2013 at 4:28 PM, Jim Pazarena fqu...@paz.bz wrote: I have a rather extensive series of databases created and in use all with the very old sleepycat db3. I believe in the addage don't fix what ain't broken, but in the case of db3, it IS broken and my db files get corrupted on occasion. I could move to db5 or db6 OR MySQL, or even postgres. I use simple primary key files, most entries are added from a CLI or termcap/curses screen. Some programatically. With about the same number of sequential dumps vs indexed random reads. I have no experience with the c interface for postgres or mysql, but also, do not know how much the c interface has changed for sleepycat 5/6 compared to the c interface for db3, which I understand quite well. So I am prepared for a learning curve irrespective of which platform I select. Records do not exceed much more than 10-20,000, with key sizes not much wider than 16 bytes (ipv4), 13 (mac), 32 (ipv6). And various smaller key sizes. Suggestions would be very much appreciated. -- Jim Pazarena fqu...@paz.bz The following page may be useful : http://en.wikipedia.org/wiki/Comparison_of_relational_database_management_systems MySQL is dropped from some Linux distributions due to its restrictions and replaced by MariaDB : http://en.wikipedia.org/wiki/MariaDB Dropping of MySQL is continuing among Linux distributions for new releases . The PostgreSQL would be best choice : http://en.wikipedia.org/wiki/PostgreSQL I am not using any one of them , but my PhD ( in 1996 ) subject is A Multimedia Information Management System . Thank you very much . Mehmet Erol Sanlituk ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org