Preparing to the next generation of Multi-Model storage engine with OrientDB v3.0

Dublin (CA, USA), December 14, 2017.

After more than 1 year working to release a stable v3.0, we’re finally close to the 1st release candidate for v3.0. It’s just a matter of days. OrientDB v3.0 has many new cool features that will be described in the next days. Today I’d rather like to focus on the performance of the new storage engine. We compared the results against latest v2.2 GA by using the Yahoo YCSB benchmarks on an Intel(R) Core(TM) i7-6700 CPU @ 3.40GHz, 64 GB RAM, 512 GB SSD.

Version 2.2.20 throughput op/sec:

DB size       Initial load   50% reads/50% updates  95% reads/5% updates  100% reads
 25M records         7,480          24,788                83,051            177,822
 50M records         6,840           3,808                15,601             30,556
100M records         6,150           2,504                11,908             21,668

 

Version 3.0 throughput op/sec:

DB size       Initial load   50% reads/50% updates  95% reads/5% updates  100% reads
 25M records        26,624          55,194               118,175            113,606
 50M records        25,086           9,454                20,998             29,610
100M records        17,913           5,864                14,900             22,038

 

Those numbers show that OrientDB v3.0 is more than 3 times faster than v2.2 on writes (inserts and updates), while read performance remains pretty much the same. Write performance was one of the biggest complaints from our users, we finally figured out how to improve write performance without sacrificing read speed.

If you want to test your application with OrientDB v3.0, even before the GA is out, please check out the “develop” branch from GitHub:

https://github.com/orientechnologies/orientdb/tree/develop

And send us your results and feedback by writing to the Community Group.

 

Thanks.

Best Regards,
Luca Garulli
Founder OrientDB

Start using the world’s leading multi-model database today