The Fifth Elephant 2013

An Event on Big Data and Cloud Computing

The database cannot be better than the underlying datastructure

Submitted by Abhishek Kona (@sheki) on Wednesday, 1 May 2013

videocam_off

Technical level

Intermediate

Section

Storage and Databases

Status

Submitted

Vote on this proposal

Login to vote

Total votes:  +34

Objective

Understand the common underlying datastructures in current storage engines, the trade offs and why this should drive decide which database to use for your next app.

Description

  • look at the different datastrucutres behind different storage engines, give examples.
  • an introduction to the three datastrucutres in use today - Binary Trees - InnoDB, LSM Trees - LevelDB, BigTable, HBase, Cassandra . Fractal Trees - TokuDB, TokyoTyrant.
  • understand the trade-offs when you decide to pick the next database for you application.

Previously, I chose databases for my applications by treating databases as a black box. I believe a lot of Engineers start with this approach. Knowing about the underlying datastrucutres and the optimization parameters will make the decision more scientific.

I will also briefly explain how Facebook benchmarks databases and makes the decision of investing in a database technology.

Requirements

  • Basic understanding of the Binary Tree.
  • Understand basic performance terminology Throughput and Latency.

Speaker bio

I currently work on the Database Engineering team at Facebook. We here are hacking on the next generation storage engine.

I have worked on data storage problems for 2+ years now. I have been on both sides of the coin, a database user, and now a database Engineer.

Links

Comments

Login with Twitter or Google to leave a comment