During this Meetup we will have two talks:
First talk: Cassandra and Multi Data Centers: How to Achieve Reliability at Scale
Abstract : When it comes to making an application available to the public, or to a private group within an organization, consumers of those applications expect the same two things: availability at any time and immediate responsiveness.
It’s no wonder half of the population falls in tears when their favorite application such as Instagram or Facebook become unavailable for a short moment, as seen recently.
Building an IT infrastructure that provides the kind of availability we now come to expect is not a simple task. To achieve that, you need to rely on physical redundant infrastructure blocks, and on technologies that can leverage this redundancy. One of those technologies is Cassandra.
This talk will focus on the key mechanism that makes Cassandra so reliable.
The Presentation will cover
The different level of availability you might want to reach
How to achieve it and the pros and cons of each solution.
At the end of this talk, you will have a better understanding of the Cassandra build in multi data center feature and be able to make the right Cassandra architecture decision to balance cost vs availability.
Second Talk: Solving Optimal Data Placement for Instagram’s Global Scale
Abstract – The deployment of Cassandra at Instagram is a challenging problem, especially considering the global placement of our data centers. This talk will cover the motivation and details behind partitioning our data across regional groupings of Cassandra clusters. We integrated with a framework developed internally called Akkio, which allows us to split data into logical units, with strong locality for each unit. We will dive into the details of the placement, access, and migration of data in a geographically partitioned data infrastructure.
Speaker: Xiangzhou Xia, Software Engineer, Instagram