David Busby ha scritto:
Dave Nebinger wrote:
Search for "unmaintainable ridiculous software architecture" and you'll probably find a match to your product.
It doesn't matter what filesystem you choose as this architecture is going
to abuse them significantly.
As an application developer I would never create software that met the
requirements that you've listed above; it's obviously a poor solution to a
problem that is better handled by a database.
You really should contact whoever created the software and ask them to take
a programming course or two.
You should watch what you say. I know this will abuse the system, which will it abuse the least?
A database would not work because I wouldn't be able to #1 make this easy tree structure to scan #2 it will be smaller storage wise in files and in a DB with all that overhead #3 digging through the tree is faster than an index scan in mysql, postgresq or DB2 with as many records as I'll have. The projected database structure would be >=3Tb, but by using directories as my index I can eleminate much of that overhead and shring my storage requirement to a maintainable 750G. You don't learn that in programming class, you learn that building financial software for the last 10 years.
You #1 didn't have enough information to tell me how to arch. my software, #2 assumed the wrong info and made the wrong suggestion and #3 didn't provide any helpful information #4 attempted an insult on a list that is supposed to assist. Why even send your post? (don't bother with answer)
/djb
#1
I've never used them but this seem a problem that can be solved by ldap databases,
Anyway I've implemented for my website a tree structure for users, based on traditional SQL databases (MySQL).
It uses a varchar to simulate the tree, now it's used for only few thousand users, with a max depth level of 64 but it can be extended to support more.
Quite all operation are done by single query, using recursion is totally avoided (It's the only thing that I can think doing it slower than a filesystem).
One year ago I've thinked about opensourcing it and write a howto, but with a low priority, if you want to translate your application in a database driven one I can give you the software in 2 or 3 days, leave me the time to clean and comment it a bit more.
#2
the only one that permit you to save disk space is reiserfs *avoiding* notail mount option, the overhead for a database structure is minimal
#3
using unique indexes should really be faster, one disk access to the index, one to the data, but only you can evaluate that.
Francesco
-- [email protected] mailing list
