Deploying a real-time data platform to deliver new revenues and experiences

Multi-Access Edge Computing (MEC), AI/machine learning and customer experience enhancement are key areas of focus for the telecoms industry, but how can they help communications service providers (CSPs) deliver new revenues and services? Mark Newman, chief analyst, TM Forum, discusses the question with Shahed Mazumder, Global Director, Telecom Industry Solutions at Aerospike.

We know that you (Aerospike) recently became a member of TM Forum. Please tell us about who you are and what you do.

Aerospike is a massively scalable and real-time NoSQL data platform. The company was founded in 2009. We are headquartered in Mountain View, California and have more than 200 customers worldwide. Our telecoms customers include Indian telcos Airtel and Jio, Vietnamese telecoms group Viettel and Globe in the Philippines. On the vendor side we work with Nokia (four products and ~90 deployments) and Amdocs (one product and eight deployments).

We have been supporting telco use cases such as customer 360, backend database for HSS, billing/charging, policy control, and geospatial & real-time microservices. In recent months, many of our conversations have been around data-heavy multi-access edge compute (MEC) use cases and AI/ML capabilities .

What do you see as the key technology and business trends over the next 12 months?

The key trends in telecoms will continue to be Open RAN, private Networks, MEC, AI/ML and customer experiences through hyper personalization. We also expect to see new use cases from adjacent sectors such as interactive online gaming and image/video analytics helping to drive greater consumption of connectivity. However, if telecoms operators are to derive new revenues from traffic growth in the 5G era they will need to provide some of the value-added applications themselves instead of acting as a simple pipe. We believe that real-time data platforms can play a crucial role here.

So, let’s drill down into some of the things that you mentioned. When it comes to MEC where do you feel that we are at as an industry? And where can Aerospike help to add value?

MEC enables many new services which were not possible until now because they need guaranteed low latency connectivity. I first wrote a blog post on MEC back in the summer of 2020, right after the very first commercial deployment of AWS wavelength by Verizon. Since then, the momentum has only grown stronger across the telco ecosystem.

This low latency can be enabled by the network (by bringing network functions closer to the end users) or through faster computation (by introducing accelerators such as GPU, FPGA or special ASICs). But there is also a third dimension which tends not to get as much attention. This is lowering the latency of data processing at the edge. And this is where high performance at scale data platforms such as Aerospike can add value.

One of the other capabilities that we offer is the ability to minimize the footprint of edge computing deployments by housing data in solid state drives (SSDs) rather than in DRAM, as is the case with other in-memory data platforms.

OK, Great. Let’s switch gears and talk a little bit about another topic you raised – telco AI/ML. What value can data platform providers like you bring here?

In a nutshell, machine learning models run better with more data and fast data. The more iterations and the more training, tuning and validation you can do within the target time window, the better the results. Across the data science community, the consensus is that the key challenges lie in data preparation and model creation and tuning (combinedly referred to as “plumbing”) as models are constantly evolving. This relates to the classic “garbage in, garbage out” argument and highlights the role of “high-quality data” in AI/ML operations.

Aerospike’s real-time data platform is designed to ingest large amounts of data in real-time for parallel processing, while also connecting to compute platforms and notebooks and ML packages. In terms of data volume, our “Petabyte Scale Benchmark” paper summarizes how, in collaboration with Intel and AWS, we ran operational workloads at petabyte scale in AWS on just 20 nodes. On the speed dimension, in the same benchmarking exercise, we realized >5M TPS (<1 ms latency in 100% cases) for 100% read workload and a combined ~4M TPS for 80/20 read/write workload.

I recently authored a blog post on this very topic. There is also a technical blog that my colleague Kiran Matty published on how to build a performant feature store with Aerospike to power ML applications.

Okay, let’s move on to another area that you mentioned – customer experience enhancement. How do you do this and where have you done this?

We focus on customer experience enhancement leveraging the strength of our data platform in multiple areas. Let me give you couple of examples-

The marquee reference use case is Airtel’s Customer 360. Airtel talks publicly about how it reinvented its customer engagement model by creating an engine, Customer 360, which understands each customer intelligently and contextually and powers a “Digital Brain” that supports a completely personalized customer engagement model effectively creating a segment of one. To make this happen, Airtel needed a very dynamic way to store and serve this information with over 25K transactions per second run rate and sub-millisecond latency. It accomplished this in partnership with Aerospike and our SI partner Hoonartek, where the Aerospike data platform forms the backbone of this Customer 360. We recently announced a more generic solution named “DigiTwin”, which fits into the premium customer experience bucket by leveraging real-time data.

There is another reference use case from a slightly different angle- Viettel’s Online Charging System (vOCS). In a recent TM Forum event (APAC Leadership Summit Series), together with our customer Viettel, we presented on how a future-proof and versatile online charging solution built on top of Aerospike’s real-time data platform helps Viettel deliver digital customer experience. In the context of charging, customer experience relates to accuracy, fast processing, scalability, and consolidation across services and networks. That’s what Aerospike helped deliver in the form of a highly capable data platform as the backbone of the vOCS solution.