Tokudb Compression not working

[A Moechsin]

Hi all,

I've been reading alot about Tokudb and have been particularly impressed by it's compression and performance claims. I decided to install Percona server 5.6 and Tokudb and enabled it on a virtual machine with Ubuntu 15.04 as OS and test the compression claim.

My primary machine though is running windows 7, and I have an Innodb table with compressed row at 4K, currently sized at 970MB. I created the same table on the virtual machine with Tokudb engine storage and ROW_FORMAT=tokudb_lzma . I spooled the content of the table from the main DB and loaded it to the one just created in the virtual machine and analyzed the table and optimized it, when I checked the size of the new table, there was barely any compression done.

Am I doing something wrong? Does compression not work in a VM or are there extra steps missing from my part? Also, when I optimized the table I would get the mesage 'Table does not support optimize, doing recreate + analyze instead' ...

Below are the screenshots of both tables (tokudb and Innodb):

Really appreciate if anyone can point to me what I'm doing wrong.

Best regards,

AM

[Mathieu Longtin]

Data length, for TokuDB, is an estimation of the decompressed size of the data.

To get the actual size, you need to run this:

SELECT table_schema, table_name, SUM(ROUND(bt_size_allocated / 1024 / 1024, 2)) AS table_size_mb
FROM information_schema.`TokuDB_fractal_tree_info`
GROUP BY table_schema, table_name;

Some other useful queries here:

http://www.youdidwhatwithtsql.com/tokudb-file-table-sizes-informationschema/2104/

--
You received this message because you are subscribed to the Google Groups "tokudb-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tokudb-user...@googlegroups.com.
To post to this group, send email to tokud...@googlegroups.com.
Visit this group at http://groups.google.com/group/tokudb-user.
For more options, visit https://groups.google.com/d/optout.

--

Mathieu Longtin

1-514-803-8977

[Reinis Rozitis]

> Really appreciate if anyone can point to me what I'm doing wrong.

To add to the previous reply you can read a bit more detailed explanation
here about the differences between InnoDB and TokuDB compression data
https://www.percona.com/blog/2014/10/10/mysql-compression-compressed-and-uncompressed-data-size/


rr

[Bohu TANG]

We can't compute TokuDB underlying data file length from 'show table status', need this patch:

https://tokutek.atlassian.net/browse/DB-851

Or do:

select * from information_schema.TokuDB_fractal_tree_info where table_name=your_table_name