Experimentation Platforms

Deliver delightful products and drive growth with real-time A/B experiments

Real-Time Experiments

Scale the number of experiments run and uncover new insights in real-time so your team can build better user experiences faster. In this talk, Rockset CEO and co-founder Venkat Venkataramani shares the requirements for real-time experimentation platforms from his past experience building and scaling data management systems at Facebook.

Why Rockset

Why A/B Testing Is Hard

How Rockset Makes It Easy

Other
Limited by the number of experiments run: Data delays result in decisions on experiments being made in weeks and not days. If your data is going through a data lake or warehouse, you’ll see your experimentation process drag out and fewer features go live.
Check
Increase the rate of experimentation: Rockset ingests data in real-time and makes it immediately queryable with an end-to-end data latency of 1 second. Reach decisions faster by quickly accessing data.
Other
Need to join data from multiple sources: : A/B testing requires the ability to access real-time event streams and join them with user profles, product, device data and more. But traditional data tools require denormalization, resulting in data explosion that is prohibitively expensive.
Check
Join data across multiple sources: Join data from different collections and data sources effortlessly. Create a self-service experimentation platform where teams can add new datasets as needed, without requiring denormalization.
Other
Need interactive slicing and dicing: Interactive slicing and dicing enables teams to build new hypotheses, construct better ideas and design follow up experiments. Teams that can’t drill down lack full insight into their A/B experiments, limiting their value.
Check
Interactive slicing and dicing: Run ad-hoc SQL queries to interrogate your data. Rockset is optimized to serve millisecond latency, complex SQL queries so you can get instant results.

Why A/B Testing Is Hard

Other
Limited by the number of experiments run: Data delays result in decisions on experiments being made in weeks and not days. If your data is going through a data lake or warehouse, you’ll see your experimentation process drag out and fewer features go live.
Other
Need to join data from multiple sources: : A/B testing requires the ability to access real-time event streams and join them with user profles, product, device data and more. But traditional data tools require denormalization, resulting in data explosion that is prohibitively expensive.
Other
Need interactive slicing and dicing: Interactive slicing and dicing enables teams to build new hypotheses, construct better ideas and design follow up experiments. Teams that can’t drill down lack full insight into their A/B experiments, limiting their value.

How Rockset Makes It Easy

Check
Increase the rate of experimentation: Rockset ingests data in real-time and makes it immediately queryable with an end-to-end data latency of 1 second. Reach decisions faster by quickly accessing data.
Check
Join data across multiple sources: Join data from different collections and data sources effortlessly. Create a self-service experimentation platform where teams can add new datasets as needed, without requiring denormalization.
Check
Interactive slicing and dicing: Run ad-hoc SQL queries to interrogate your data. Rockset is optimized to serve millisecond latency, complex SQL queries so you can get instant results.
Core Features for Experimentation Platforms
  • Built-In Data Connectors

    Set up continuous ingestion from common data sources, like MongoDB, DynamoDB, Kafka and S3, in a plug-and-play manner. New data is reflected in Rockset in seconds.
  • SQL with JOINs

    Join multiple data sets for analysis. Rockset supports full-featured SQL, enabling complex search, aggregation and join queries.
  • Converged Index™

    Rockset automatically builds search, columnar and row indexes on your data to deliver fast queries at scale.
  • Serverless Autoscaling

    Rockset scales storage, ingest and query layers automatically and independently in response to growth and spikes in usage.

How It Works

Rockset serves as a real-time sink that continuously ingests and indexes event data from Kafka or Kinesis. Rockset serves fast ad-hoc SQL queries and powers web interfaces with A/B testing results.

Resources



Related BlogRelated Blog

Rapid Experimentation and Growth Using Real-Time Analytics

This post explains how to build for the requirements of a massive-scale A/B experiments platform.

Read more->
Related WebinarRelated Webinar

Running Real-time A/B Experiments at Massive Scale

In this talk, Rockset Co-founder and CEO Venkat Venkataramani will cover the changing face of real-time analytics.

Read more->
Related WebinarRelated Webinar

How We Scaled It: Facebook’s Online Data Infrastructure to 1B+ Users

An inside look into building massively scalable online infrastructure at Facebook.

Read more->
Related BlogRelated Blog

Converged Index™: The Secret Sauce Behind Rockset's Fast Queries

Learn how Rockset delivers low-latency SQL for search and analytics using a combination of row, column, and search indexes.

Read more->
Related BlogRelated Blog

Building Real-Time Data Architectures to Foster Innovation

Lessons on building real-time data architectures based on experiences growing Facebook users 30x.

Read more->
Related BlogRelated Blog

Building a Real-Time Customer 360 on Kafka, MongoDB and Rockset

A step-by-step guide to building a real-time customer 360 using purchase data from MongoDB and marketing data from Kafka.

Read more->

Customers

Real-Time Analytics At Lightning Speed

See Rockset in action