I was thinking of setting up similar experiments; nice idea hooking the MTD layer to count erase cycles..
I'm going to keep 1st implementation simple and just use sqlite direct to flash, no memory cache. Then I can measure the performance.. We should be using a flash controller that does wear levelling which of course will prolong life. - -Paul On Thu, 2008-10-02 at 17:44 -0400, Matthew L. Creech wrote: > On Thu, Oct 2, 2008 at 4:20 AM, Paul McMahon <[EMAIL PROTECTED]> wrote: > > What's the consensus on using sqlite with flash based storage such as sd > > cards on embedded devices? > > > > Has anyone done this successfully yet with a real product (that needs to > > last a few years in the field)? > > > > We've got several (Linux-based) embedded products using SQLite > databases. One has an M-Systems SoC with Ext3 on top of TFFS, another > uses CompactFlash with Ext3, and yet another uses a directly-addressed > NOR chip with JFFS2. None have had any problems related to SQLite. > > How long it lasts depends (obviously) on the amount of data you're > pushing through on average. For the NOR device, I implemented a test > system with SQLite, set up a simulator to model worst-case conditions > for the device so that it was writing to the database about as much as > it ever would, and hooked the MTD layer to count the number of erased > blocks over a pre-defined period of time. You can then use that rate > and the size of the partition to figure out how long it'll take to hit > e.g. 100k erase cycles over the whole device. For SD you don't know > how well it's doing wear-leveling behind the scenes, so throw in a > multiplier to account for that, and you'll at least get an > order-of-magnitude accurate idea of how long it'll last worst-case. > _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
