Powering the World's Data Centers with DSE

Equinix is the leading interconnection platform and colocation provider with more than 100 data centers across the globe. Powering millions of business applications with higher performance and reduced latency, Equinix provides its customers' network monitoring solutions and data center infrastructures for various purposes including monitoring, troubleshooting, and customer billing. To optimize customer experience and to provide more network traffic visibility to its customers, Equinix started to build applications and was looking for a database replacement to realize these goals.

The Challenge

Equinix has over 100 data centers across the globe and provides services to share Internet network exchanges. To guarantee optimal data center performance and network traffic health to its customers, Equinix aggregates streaming data from infrastructures for monitoring, troubleshooting, and customer billing. In 2013, approximately 90% of the U.S. Internet traffic flew through Equinix’s platform.

This mindboggling volume and velocity of time-series data required the corresponding scale, real-time performance, and continuous availability. It also raised Equinix’s unique needs for a superior database capable of handling heavy write workloads, with cross-referencing and data aggregation.

Equinix concluded that reliance on relational database management systems (RDBMS) is limiting. The inflexibility of relational databases made it difficult to accommodate varying and growing data sets. Further, the inability to efficiently process real-time streaming data without constraining storage as well as compute resources created latency issues.

The Solution

With the requirements of scalability, continuous availability, and performance in mind, Equinix started its search for new a database solution. Hadoop integration was important in order to provide deeper network health and data center performance analysis for their customers. After extensive total cost of ownership evaluation, Equinix selected DataStax Enterprise and Apache Cassandra™ over HBase for cost-effective scale, continuous availability architecture, superior write and read performance and seamless integration with Hadoop.

The greatest thing for us with Cassandra is the ease of operation, which goes hand-in-hand with Cassandra’s always-on node architecture. Cassandra’s peer-to-peer architecture guarantees no single point of failure to collect and store streaming data from infrastructure instruments and enables us to write and read from any node with minimal latency.

Praveen Kumar
Sr. Manager of Emerging Technologies & Platform at Equinix

OpsCenter, DataStax’s visual management, and monitoring tool eliminates the complexity and provides Equinix visibility into its write-intensive operations. DataStax expert support assures Equinix that its mission-critical applications are always covered.

The Results with DataStax

With DataStax, Equinix has been able to unlock the value of streaming data that was impossible to even gather with relational systems. As a result, Equinix has been able to maximize the value delivered to customers and their own return on services. In addition, as a heavily instrumented data center provider, it is imperative for Equinix to plan for power capacity, optimize resource consumption, and failures.

Applications built on Cassandra deliver insights on power consumption which allow administrators to plan infrastructure and resource allocation. As an example, during peak trading hours for Equinix’s financial service customers, it is now possible to proactively monitor and improve operational efficiency by ingesting real-time data to a Cassandra cluster. Previously, this process was manual and time-consuming with relational systems.

Today, 90 million records of network traffic data are collected and stored in Cassandra on a daily basis. Equinix leverages this information for better customer engagement by providing insights on network health and traffic analysis. “Cassandra simplifies and provides meaningful insight into operations,” said Kumar, “more insights for our customers means they can proactively plan data center usage for peak times and during disasters like a hurricane.”