Hello Dan, Yes, it is true that insufficient memory (or SSD for the L2ARC) can slow writting down, but with ZFS deduplication + ZFS compression both turned on, in production environments we are seeing deduplication ratios of 12 to 1. Backup speeds with such deduplication ratios are rather spectacular.
Best regards, Kern On 10/04/2013 04:12 PM, Dan Langille wrote: > On 2013-10-04 09:33, Radosław Korzeniewski wrote: >> Hello, >> >> 2013/10/4 Dan Langille <d...@langille.org> >> >> On 2013-09-27 14:17, Radosław Korzeniewski wrote: >> Hello, >> >> 2013/9/26 BOURGAULT Antoine <antoine.bourga...@sib.fr> >> >> Thanks for your answer ... but can you explain what is " aligned >> volumes" ? >> >> http://www.google.com/search?q=bacula+aligned+volumes [1] [1] >> >> I don't know, but I think this refers to ZFS, not Bacula. >> >> Bacula aligned volumes works well with ZFS or other deduplication >> solutions like DataDomain, etc. > Some issues regarding ZFS dedup to keep in mind: > > * Dedup will go slowly unless you have enough RAM to hold every hash for > every block you are deduping > * Compression, instead of dedup, can result in higher throughput > (compression and decompression is faster than writing/reading the larger > amount of data to/from HDD) > * The OpenZFS (http://open-zfs.org/) project holds good potential for > ZFS on Linux > ------------------------------------------------------------------------------ October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
