Tips and Tricks for Maximizing Accumulo Ingest Performance

Back to Schedule



Internals Performance


Many organizations choose to use Accumulo for its high ingest rates. When use in a well architected application, Accumulo performs better than any of the other databases in its class. However, there are several concepts of distributed computing that can lead an under engineered application to perform poorly. In this talk, we will present Accumulo's ingest architecture, discuss the most critical areas that determine application performance, and detail a highly optimized document store application. We will then show how the design of our document store achieves millions of key/value writes per second per node in a sustainable and scalable operational environment.


Adamn Fuchs
Chief Technology Officer, Sqrrl Data, Inc
Adam Fuchs is one of the original Accumulo developers, and after 8 years he continues to participate in the management and development of Accumulo. He architected a number of Accumulo's core systems and played a critical role in bringing Accumulo to the Apache Software Foundation (ASF). As the chief technology officer of Sqrrl Data, Inc., Adam uses his skills in distributed computing and data science to enable analysis of behavior graphs for indications of cybersecurity threats. As an ASF member, Adam also mentors and shepherds new open source projects into ASF. When he's not developing new technologies, Adam enjoys exploring the wilderness near his home in Seattle, WA.