On Mon, 20 Dec 2010, Edward Ned Harvey wrote: >> From: [email protected] [mailto:[email protected]] >> On Behalf Of [email protected] >> >> having any cache drastically improves the write situation as the writes >> can be combined and written more efficantly. > > That's a partial truth. If it were ZFS, for example, then a whole bunch of > small writes get aggregated into a single sequential write in RAM buffer, > but in EXTn, no such thing is possible. What you get with EXT is ... the > controller buffers the write, but will not redirect to different disk > blocks. So you're able to benefit from increased size of the write buffer > queue, and more efficiently leverage the command ordering in the actual disk > hardware... But hardware raid isn't going to combine smaller writes into > larger sequential writes.
it will combine small writes into bigger ones if the small writes are to ajoining areas of disk. if you are doing lots of things in one directory for example, all the directory metadata writes are likely to be combined. on ext3 writes to the journal are likely to be combined. if you are running a database with a journal log file, all the writes (and their associated fsyncs) are likely to be combined. but random writes to different places on the disk will not be combined, which is what I think you were referring to. David Lang _______________________________________________ Tech mailing list [email protected] https://lists.lopsa.org/cgi-bin/mailman/listinfo/tech This list provided by the League of Professional System Administrators http://lopsa.org/
