Handling Real-Time Use Cases at Scale Using a Hybrid Memory Architecture
In this talk, we will describe key real-time use cases in the areas of real-time advertising, real-time billing in Telco, fraud detection and risk management in payment systems. We will then present a brief overview of a database platform that has proven to be well suited for handling such use cases.
Real-time advertising systems have some of the highest throughput and uptime requirements while also needing the most stringent SLA in terms of low latency. We will describe use cases in deployment in several companies close to the scale of Google, Facebook, etc.
Telecommunication systems require the system to provide accurate real-time billing and charging on network data 10-100 times the scale of that which was required just a few years ago.
In payment systems, detecting fraud as it happens is critical to maintaining the integrity of the business. This could mean that for each payment transaction, several components in the payment chain have to be both monitored in real time and analysed for anomalies based on recent historical data to enable a decision to be made in 200-500 milliseconds.
In trading systems, there is a need to constantly calculate risk on a database of changing customer positions (due to both trades and fluctuations in stock values). These calculations are critical in enabling a better user experience by allowing faster, higher quality decisions such as whether to grant margin loans.
We have found that the best way to handle the demanding workloads of these types of real-time decisioning use cases is to use a hybrid memory database platform that uniquely leverages DRAM with flash storage. This platform needs to smoothly scale out processing across multiple nodes and optimize the usage of CPUs, DRAM, SSDs and networks to efficiently scale up performance on a single node. Over the past six years, this technology has been continuously used in over a hundred successful production deployments, as many enterprises have discovered that it can substantially improve their business capabilities and enhance their users’ experience.
When it comes to databases, Srini is one of the recognized pioneers of Silicon Valley. He has two decades of experience designing, developing and operating high scale infrastructures. He has over 25 patents in database, web, mobile and distributed systems technologies. His career includes high-profile technical and executive management roles at IBM, Liberate and Yahoo!.
Srini co-founded Aerospike to solve the scaling problems he experienced with Oracle databases at Yahoo! where, as Senior Director of Engineering, he had responsibility for the development, deployment and 24×7 operations of Yahoo!’s mobile products. Srini is an M.S./ Ph.D. holder from the University of Wisconsin-Madison and is an alumnus of IIT Madras.