<p>Small investigation of EOSUSER namespace</p>

<p>Out of curiosity (using a dump from Yolanda) I checked the effect of deduplication (file level). Please note deduplication is *not* a prioritiy IMO.</p>

<p>Input: 396 M files (Early October)</p>

<p>Consider only files &gt;10MB</p>

<p>Use AD32 (as recorded in the catalogue). With "large&nbsp; files", AD32 collisions are not too many collisions.</p>

<p>Dedup saving ~15%&nbsp; (188 TB out of 1184). I am shocked but I do not find any loophole</p>

<p>Anyway: top files being "repeated":</p>

<p>- cernbox/smashbox testing (e.g. file "c1857a3c" has 20k replicas for a total of 1.1 TB). They are characterised by a flat time distribution (test executed every x hours).</p>

<p>- File 05f0347e is an output of a job (2.9 GB x 374 copies). Suboutputs of single job (all equals...).&nbsp;&nbsp;&nbsp; Time distr concentrated well within an hour.</p>

<p>- Similar cases exists where 1 file is in the user dir and all the others are in the trash (again, rather peaked time distribution.</p>

<p>&nbsp;</p>

<p>Q: de-duplication effect on number of files? Not looked.</p>

<hr>
<p>Q; is SWAN now using the new unified principals (since will play with instances)? will check (Enrico: looks OK)</p>