Today's guest post comes from Amol Kekre, CTO and co-founder of DataTorrent.

Scaling and performance are some of the most critical aspects when processing Big Data in real-time. When we started on Google Compute Engine we wanted to explore how the performance of a virtualized cloud environment would match the needs of our platform for high-throughput, Big Data computations while maintaining sub-second latency.

DataTorrent is a real-time stream analytics platform designed to support today’s most demanding, high-throughput, Big Data applications. Many of our use cases (particularly around processing machine-generated data, such as from sensors, logs, etc.) see over half a billion events processed per second.

For businesses to instantly analyze any volume of data as it comes in and respond in real-time, a solution must ensure scalability along with consistent sub-second latency – even while processing a massive volume of events.

While testing our platform, Google Compute Engine provided the necessary layer of compute and enabled us to scale linearly, without compromising latency. In one test, using identical instance configurations, 10 instances were able to process over 87 million events per second on the DataTorrent platform. When we scaled this test to 45 instances, again using the same instance configurations, we achieved over 400 million events per second. This showcases Google Cloud Platform’s excellent performance and suitability for mission-critical, high-throughput, big data applications.

In addition, gcutil allowed us to easily automate instance provisioning and configuration of our cluster. The ability to manage virtualized networking and firewall settings enabled us to set up a secure cluster, protecting our source code and data.

The flexibility and ease of use of Compute Engine enabled us to quickly launch our solution and run stability and auto-scaling tests almost immediately - which made for a pleasant deployment experience.

Get DataTorrent on Google Compute Engine today
With as clear a focus on mission critical, massively scalable applications as Google - we are excited to offer Compute Engine users our platform for real-time computations on a massive scale.

  • Use it free: download and simply install on your Google Cloud Hadoop cluster.
  • Visit us to learn more and read up on our technology
  • Contact us at

-Contributed by Amol Kekre, CTO and Co-founder, DataTorrent