-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 25/03/12 10:48, Tal Tabakman wrote: > I am evaluating a sqlite extension called zipvfs.
http://www.hwaci.com/sw/sqlite/zipvfs.html for anyone else interested. > I assumed that database writing will be faster ... in reality I see no > difference At the end of a transaction SQLite needs to do fsyncs which ensure that the data is firmly on the disk. If you are using spinning media this will require waiting one or more times for the disk to have rotated to a certain place. For example this is why you'll find you can't do more than 60 transactions per second on a 7200rpm drive. Since the source for zipvfs isn't public I can't tell the exact inner working. However it is reasonable to believe that it works on a page by page basis. If you use larger pages (the default is 1kb, max is 64kb) then there is more data to compress and more likely a reduction in size. SQLite's data encoding is quite compact so fairly random data won't compress that much. Current hard drives have internal buffers of 16 or 32MB. You'd have to write a heck of a lot of data to fill that and operating system buffers. Roger -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) iEYEARECAAYFAk9vbVAACgkQmOOfHg372QTzMwCcCplmO0R849xTBXrxbydTty15 fxUAn3A+AY+6xPxpaFQ9qvDLezn7gjBU =lHGv -----END PGP SIGNATURE----- _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
