Max. no of database in a server? - Database Discussions
This is a discussion on Max. no of database in a server? - Database Discussions ; >what i think is : > >if there is NO LIMIT, I think the manual should state clearly there is >really NO LIMIT, just as the others If there is NO LIMIT on the number of 'a' characters appearing in ...
![]() |
| | LinkBack | Thread Tools | Display Modes |
|
#31
| |||
| |||
| > >if there is NO LIMIT, I think the manual should state clearly there is >really NO LIMIT, just as the others If there is NO LIMIT on the number of 'a' characters appearing in table names (all tables in all databases combined), then the manual should state clearly there is really NO LIMIT. (But we all know this can't easily exceed 256TB because of limits on disk size). If there is NO LIMIT on the number of ALTER TABLE statements you can execute on a given server in your lifetime, then the manual should state clearly that there is really NO LIMIT. (But as a practical matter, your life expectancy expressed in microseconds is likely to be way over what you can actually do. If there is NO LIMIT on the number of people that you can safely carry in an elevator, how come elevators can get overloaded? (The limit isn't on the number, it's on the weight.) For MySQL, the question of "how many databases can you have on a server?" roughly comes down to "how many subdirectories can you have in a directory?", which is an OS issue. For FreeBSD, I think that limit is 65533, assuming you don't use symlinks. If you do use symlinks (to directories elsewhere), it may come down to "how many files (symlinks) can you have in a filesystem?" For FreeBSD, this is about 2**32 minus a half a dozen or so. Now, how many databases can you have on a server before looking up the database name in the directory kills performance? (I recall a version of Xenix where UUCP having more than 100,000 spool files in the spool directory (no UUCP subdirectories in this version) started causing communication timeouts. This was on mid-1980's hardware, and CPUs and disks were pretty slow, and memory scarce. I think you could have 65,533 subdirectories (databases) without too much degradation. And how much are you going to USE these databases? At one query a minute for all databases combined, you could probably handle a billion databases with no problem. You might run into troubles with the number of SIMULTANEOUSLY OPEN tables allowed, as many OS don't allow a process (MySQL server) to have more than a few thousand open files simultaneously. You can run into this problem with ONE database and a few thousand tables, also. MySQL may try to cache the user tables in memory, and if you've got complex permission tables with a different user for each database of a couple million of them, you might start feeling the limits of a 4G address space on x86 hardware. |
|
#32
| |||
| |||
|
I have no idea about it will you please help me by answering my question How do I build Workbench in Linux/FreeBSD/Solaris/Whatever? please reply me thanks in advance.
__________________ datawarehousing definition article |
![]() |
« Previous Thread
|
Next Thread »
| Thread Tools | |
| Display Modes | |
| |
All times are GMT -4. The time now is 09:45 AM.




Linear Mode