stored only attributes

Feature Requests
1

hi,
Suppose we have attributes in a table that we only want to store them and we don’t want to use them for sorting or aggregating. In my opinion, similar to fields, it should also be possible for attributes to store only them.
thanks,

2

what is the difference between stored attributes and stored fields if you only fetch that data?

3

i want to store number not string.

4

I sure there there is no difference

5

for example, we have a table with integer attribute. when we define this attribute as rt_attr_uint, the total size of index is: 670.4 MB, and when define this attribute as filed and set this attribute as store only, the total size of index is: 684.6 MB. (because we store number as string). i want to use this attribute only for using in select statement not sorting and aggregating. i want to reduce total size of index but when we using store only filed, the size is increased.

6

could you use stored_only_field option? that makes data to be only stored on disk and retrieved to client

7

I tested both options. i have a table that i configured it with two option:


option 1

table translation
{
    type = rt
    path = c:/manticore/var/data/translation/translation

    rt_field = Text
    rt_attr_uint = BookSectionId
    stored_only_fields = Text
}

option 2

table translation
{
    type = rt
    path = c:/manticore/var/data/translation/translation

    rt_field = Text
    rt_field = BookSectionId
    stored_only_fields = Text,BookSectionId
}

when i configure my table with option 1, total size of this index is: 670.4 MB, and
when i configure my table with option 2, total size of this index is: 684.6 MB.

the size is important for me and i want to reduce this size. but with using stored_only_fields, Its size increases!

8

could you provide ls -lh of the index files for both options 1 and 2 ?

9

you also could try to use columnar attribute instead of the plain one - it also keep data on disk and use compression with the converted data (not the source text)

10

it also could be better to issue FLUSH RAMCHUNK statement to convert all collected data from the .ram into disk chunk as RAM part used as raw storage and do not use any kind of compression.