Re: Impact of compressed=true attribute (in schema.xml) on Indexing/Query

[Chris Hostetter]

: Now the question is, how the compressed=true flag impacts the indexing 
: and Querying operations. I am sure that there will be CPU utilization 
: spikes as there will be operation of compressing(during indexing) and 
: uncompressing(during querying) of the indexed data. I am mainly looking 
: for any bench marks for the above scenario.

i don't have any hard numbers for you, but the stored data isn't 
uncompressed when executing aquery -- queries are executed against the 
indexed terms (which are never compressed) ... the only time the data will 
be uncompressed is when returning results to the client -- so if you set 
rows=17 in your request, only the values for the 17 docs returned  (or 
less if there were fewer then 17 matches) will be uncompressed.



-Hoss

Impact of compressed=true attribute (in schema.xml) on Indexing/Query

[Silent Surfer]

Hi,

We observed that when we use the setting compressed=true the index size is 
around 0.66 times the actual log file, where as if we do not use any 
compressed=true setting, the index size is almost as much as 2.6 times.

Our sample solr document size is approximately 1000 bytes. In addition to the 
text data we have around 9 metadata tags associated to it. 

We need to display all off the metadata values on the GUI, and hence we are 
setting stored=true in our schema.xml

Now the question is, how the compressed=true flag impacts the indexing and 
Querying operations. I am sure that there will be CPU utilization spikes as 
there will be operation of compressing(during indexing) and 
uncompressing(during querying) of the indexed data. I am mainly looking for any 
bench marks for the above scenario.

The expected volumes of the data coming in would be approximately 400 GB of 
data per day, so it is very important for us to evaluate the compressed=true, 
due to the file system utilization and index sizing issues.

Any help would be greatly appreciated..

Thanks,
sS