Joinutility seperatorLogin utility separator Infobright.com
   
 
Where is the Knowledge Grid stored? Can I interact with it directly?
Posted: 03 September 2008 08:54 AM   Ignore ]  
Sr. Member
Avatar
RankRankRankRank
Total Posts:  736
Joined  2008-08-18

This is a common question that we’ve encountered in the past during our PoC process.

Question:

Where is the Knowledge Grid stored?
Can I interact with it directly?

Answer :

KG (Knowledge Grid) files are stored in a place defined in *.ini file as KNFolder variable. By default it is /BH_RSI_Repository folder in the database main directory. All KG files are collected there.

There are three kinds of KG stored here:
HIST.xxx.yyy.rsi    - histograms
CMAP.xxx.yyy.rsi  - character maps
JPP.xxx.yyy.xxx.yyy.rsi  - pack-to-pack joining knowledge node

where xxx is a table internal number and yyy is a column number. There is also rsi_dir.rsd file which gathers the list of all KG present in the directory.

KG files are not intended to be human-readable. Changing them by hand may lead to wrong query results / server crash. They may be deleted completely (and query performance will be partially lost), but there is no way to recreate them, except JPP.* files.

Signature 
Profile
 
Posted: 13 January 2010 03:50 AM   Ignore ]   [ # 1 ]  
Newbie
Rank
Total Posts:  27
Joined  2008-10-01

Hi,

I am using Infobright (ICE 3.3) .  I have some queries regarding the creation of KN. My understanding is as follows,

There are 3 types of KNs,
1.Histogram
2.Charater MAP
3.Pack-by-pack

Histogram and Charater Map will be created during the load time. But the Pack by pack ,KN will be created during the
execution of queries involving joins.

I have loaded 10M of data using PDI (Pentaho data integration tool) ,IB loader . I checked BH_RSI_Repository folder,I can see the
Histograms and Charater Maps are created (files like HIST..rsi and CMAP..rsi respectively).

Now I have run some queries joining some tables. But I cant see the Pack-By-Pack KNs being created. I am expecting to see files like ‘JPP..’,
But its not created. So can you tell me how Pack-by-pack KNs to be created ?  Do we need to do some manual changes in configuration files to activate the creation of Pack by Pack? .My understanding is that Pack-By-Pack nodes will give speed up for the subsequent execution of queries involving joins.

Hoping for a positive reply,

Regards,

Dhanesh

Profile
 
Posted: 13 January 2010 05:00 AM   Ignore ]   [ # 2 ]  
Newbie
Rank
Total Posts:  13
Joined  2008-08-18

Hi Dhanesh!
There were some changes introduced since version 3.2.1 and P2P-files are no longer saved at disk.
Little more about this in Jakub’s post here:
http://www.infobright.org/Forums/viewthread/1186/

Greetings,
Marcin

Profile
 
Posted: 08 May 2010 07:18 AM   Ignore ]   [ # 3 ]  
Newbie
Avatar
Rank
Total Posts:  3
Joined  2010-05-08

Data Packs and the Knowledge Grid Data is stored in 65K Data Packs. Data Pack Nodes contain a set of statistics about the data that is stored and compressed

Profile
 
Posted: 23 November 2010 06:30 PM   Ignore ]   [ # 4 ]  
Newbie
Rank
Total Posts:  1
Joined  2010-11-23

Maria, your response has saved me plenty of time!

Profile