Re: [Sedna-discussion] Disk Space Usage

[Ivan S. <shc...@gm...>]

Dave, try the following hints:

1. Create one file per device id. Use collection to store these files.
However, if you have reasonable devices amount it may be better to load each
file separately. It also depends on queries you want to run. For example, do
you have queries which search text within whole data?

2. Simplify your XML in the following way:

*<device_event event_id='1' datetime='2009-01-27T09:37:29+'
category='DISTORTION' numerical_severity='801'>Module #245 has been
distorted.</device_event>*

Attributes take less space to store them and are easier to process.

3. Use value indexes to find events with specified attribute value(s):

http://modis.ispras.ru/sedna/progguide/ProgGuidesu8.html#x14-480002.5.3
http://modis.ispras.ru/sedna/progguide/ProgGuidesu5.html#x9-310002.2.2

4. It's not a very good idea to use fn:contains(). Use full text search
indexes (we can give you instructions how to get Sedna with dtSearch trial).


I will later need to periodically add new log files in. Let's say... every
> 10 minutes add another 10 MB of log lines (parsed to 20MB xml) -- add more
> events to a device id.
>

Are you going to load new data infinitely? Do you plan to remove old data?

Ivan Shcheklein,
Sedna Team