At PhotoWorks circa 2003, we had approximately 500,000,000 jpegs of various sizes sitting in Unix directories on one IBM RS6000.
We got around the OS limitations by storing image 12345678a.jpg as /12/34/12345678a.jpg, and the top level directories were on 50 different filesystems as I recall. All the image metadata was stored in UniVerse on a separate server. The only software running on the image server was Apache and a couple of shell scripts (and the embedded Tivoli HSM software - yes all the images were backed up off site). I understand PhotoWorks has moved off their IBM archive in the last few years. On 8/23/07, Martin Phillips <[EMAIL PROTECTED]> wrote: > > Hi Harold, > > > The simplest solution - that of simply copying the JPG files to one type > > 19 directory - is working, to my amazement. There are 171,000 mugshots > > in one directory and they are being handled correctly. > > Remember that the process of searching an operating system directory is > essentially a linear scan. This will need to examine, on average, half the > entries before finding the one you want when reading. A write may have to > scan the whole directory. > > There probably isn't much you can do about this aside from moving to a > hashed file which might be a tuning nightmare. Also, if this is Unix, each > file takes an inode so you need to keep an eye on the file system limits. > > > Martin Phillips, Ladybridge Systems Ltd > ------- > u2-users mailing list > u2-users@listserver.u2ug.org > To unsubscribe please visit http://listserver.u2ug.org/ ------- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/