How Intel® FPGAs and rENIAC’s* Distributed Data Engine Can Turbo-Charge Your NoSQL Databases

Financial services can benefit hugely by exploiting the power of Intel’s field programmable gate arrays.

As data and transaction volumes increase, it’s necessary to find a solution that tightly integrates hardware and software for predictable performance, high throughput, and low latency.

By Steve Bradley, Technology Writer

You’ve probably heard the phrase ‘data is the new oil’. It’s credited to mathematician Clive Humby in 2006 but has been repeated or paraphrased countless times since.

Dig a little deeper and it doesn’t entirely work as an analogy, but there’s still an element of truth to it. Just like taking oil from the ground, you have to be able to extract insights from data before it can be of any value.

For banks and financial institutions this is still very much the case, with use cases including capturing market data, fraud detection, and getting a 360-degree view of the customer.

These are all things NoSQL databases are used for and as data and transaction volumes increase, NoSQL databases enable deployment of various applications and services to get set-up quickly, scale-out fast and enable creation of additional data nodes when needed. While scaling for data and transaction volumes, customers don’t want to compromise on performance and that requires significant tuning to achieve high throughput and predictable low latency to meet business’s service level agreement (SLA).

Large data companies can build their own proprietary solutions in house, but enterprises adopting open source technologies need to extract more performance to meet the SLA and scale from open source databases but don’t have the required skill set to do this, which is where a business like rENIAC* comes in. As data and transaction volumes increase, it’s necessary to find a solution that tightly integrates hardware and software for predictable performance, high throughput, and low latency.

According to rENIAC*, its business solution can integrate hardware and software to deliver up to 10x improvement in performance and deliver 99th percentile latency of sub 1 millisecond.

rENIAC’s* distributed data engine is a software acceleration layer between an open-source database server and client, which runs on Intel® FPGAs (field-programmable gate arrays), saddling the raw horsepower and helping clients to get more out of NoSQL databases such as Apache Cassandra*.

“We help customers get more performance out of their databases at a reduced footprint of the system without disrupting their existing software stack,” says Prasanna Sundararajan, rENIAC’s* co-founder and CEO. “A smaller or reduced footprint also means it’s easier to manage as you scale, leading to overall reduced total cost of ownership.”

There are various things that make Intel® hardware attractive to companies like rENIAC*. “FPGAs in general are very good as a hardware platform,” explains Sundararajan. “You can customise the design that is running on the FPGAs so you can deliver predictable performance, but Intel’s acceleration stack also reduces development time by taking care of all the system-level IO interfaces. It also has go-to market channels with vendors to get FPGAs adopted on industry standard servers more easily.” Sundararajan also says that one of the customers chose Intel® Platform because of their support for virtualisation, which will become increasingly in demand.

So what else does the future hold? Large-scale applications are moving towards a microservice architecture, which means they’re divided into smaller services. For the data structure to adapt and scale without expansion of the footprint or compromising the business metrics and SLA, performance is key.

Expansion of workload beyond NoSQL is another one that Sundararajan sees Intel and rENIAC* delivering together, with a lot of customers requesting support for open-source SQL databases as well. The other commonly requested feature is the ability to run AI algorithms close to the data, or where the data is being accessed. “I do see use cases in the future where there is closer integration of AI and databases on Intel® Platform,” says Sundararajan.

To hear more from rENIAC* click here.

For More Information: