Aaron J. Seigo said: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > On Thursday 05 June 2003 04:47, Trevor Lauder wrote: >> A carefully >> designed implementation should be more reliable then our current >> systems, >> and offer less overhead. > > how so? assuming that the database approach means an extra layer between > the > inodes on disk and requests for those files, how is this of any benefit > to, > say, apache?
I believe some of these implementations were going to get rid of files & directories completely. Everything, including the data would be in the DB.... not just the pointer information. So the DB wouldn't be an extra layer on top of the inodes in this case, it would just replace the file/directory structure we currently have. > >> Being able to search and index a global database >> will be a great advantage to people trying to harness the power of >> things >> like PHP/SQL. > > again, how so? if you are doing full text indexing of a large number of > documents, sure, i can see that. or if you are providing a song > downloading > service, sure. it let's you cut out the middle man (the SQL database) > which > is a good thing. but if you are already using a RDBMs for your data, or > your > data isn't structured or doesn't have any need for further structuring, i > don't see the point. True, and this is probably where the breaking point is between using a DB filesystem and not using one. > > note that i'm obviously not saying that approaches like what ReiserFS > wants to > become are useless, just that i don't know if they are a great general > purpose solution. actually, i should say that Reiser's concept of > filesystem > plugins is an excellent concept since it allows one to scale the FS up or > down the scale of "metadata enabled". > >> > i think there is a large place for both and those saying otherwise >> have a >> > vested interest. >> >> I'm not sure what you are accusing me of here, but I have no vested >> interest in database filesystems. > > i was actually refering to Reiser, Oracle and Microsoft, not you =) Ahh, that clears up some confusion on my part :) Microsoft's vested interest in this would be that they can further increase the strangle-hold they have on customers with proprietary file formats. It's hard enough now with a proprietary file format on a normal filesystem, I don't want to imagine how hard it would be with a DB filesystem if it was implemented by Microsoft and they weren't forced to publish the specs. > > - -- > Aaron J. Seigo > GPG Fingerprint: 8B8B 2209 0C6F 7C47 B1EA EE75 D6B7 2EB1 A7F1 DB43 > > KDE: The 'K' is for 'kick ass' > http://www.kde.org http://promo.kde.org/3.1/feature_guide.php > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.2.2-rc1-SuSE (GNU/Linux) > > iD8DBQE+39vx1rcusafx20MRAhmiAJ4oPLVIn9DJqjWVnYuW//YwWoUt0wCfRKfa > y0PnuBZtSqQ0E9wH4u8M3D0= > =jXLn > -----END PGP SIGNATURE----- > >
