|
|
Posted by Erwin Moller on 10/27/49 11:20
0000=year wrote:
> Hi
>
> to add some more information what we need this for. Sorry, that I forgot
> to throw in this information right away:
>
> We need to write only about 5 lines to the file - or 1 record to the DB.
> It's only needed to tell the indexer (of TSEP) which reads the contents of
> the files that another instance of the indexer is already running. So, the
> first instance started will write to the file / db record when it has been
> started and that it is running at time X (every 30 seconds)
> The reading (from the file/record) is only done if a instance of the
> indexer is started and there IS a record already.
>
> Olaf
Use a database. :-)
You ARE accessing the data (files) again, so you better switch to a real
solution (database) right away.
Just think about it: You have 5.000.000 files in a directory on a filesystem
like NTFS. You need a file with the name xyzwtq.txt
How to find it?
The system has to scan through all filenames stored and compare it to your
filename.
If you have some smart indexing going on, this will seriously reduce the
seekspeed.
Maybe there are some fancy filesystems out there that can compete with a
database by using some smart indexing, I do not know to be honest.
Regards,
Erwin Moller
Navigation:
[Reply to this message]
|