Aerospike’s multi-model real-time data platform is known in the industry for its small footprint and low total cost of ownership (TCO), thanks to our patented Hybrid Memory Architecture (HMA). Over the years, this compelling value proposition has been recognized for its combination of stable high performance, low cost, and massive scalability when it comes to telecom use cases.

Today, our telecom customers use Aerospike for a wide range of use cases – from billing/charging to policy control, signaling to subscriber data management, and even geospatial real-time microservices. We also support traditional telecom use cases through partners such as Neural Technologies. There is a growing interest in Aerospike for data-heavy multi-access edge compute (MEC) use cases, where the sub-millisecond to single-digit millisecond response of the Aerospike Real-time Data Platform can contribute to the reduction of end-to-end latency experienced by end users.

Aerospike Telecom Use Cases

Figure 1. A snapshot of Aerospike’s Major Telecom References

As of late, there are also inbound queries from providers of subscriber profile databases (HLR for 3G, HSS for 4G, UDM/UDR for 5G), especially the ones who focus on non-Tier one markets.

But we can do so much more.

In this blog post, I will highlight a few key functionalities/capabilities of the Aerospike data platform that make it stand out. Some of our peers have a subset of these capabilities, but none have all of them. So, let’s get into that in the context of telecom use cases.

Full-fledged document database

Aerospike Database 6 is the first expansion of our flagship NoSQL database beyond its initial key-value data model to include full document database capabilities, including the storage, indexing, and querying of JSON document data. Powered by the Aerospike Document API, Aerospike Document Database enables developers to create and store JSON document data as an Aerospike Collection Data Type (CDT) object that can be accessed, queried, and modified using a JSONPath syntax.

This recent blog post on leveraging a real-time document database articulates the trends and requirements around real-time unstructured data at scale. As telcos expand their customer interactions through a constant online/web presence and real-time applications, they have become gold mines of unstructured data at a massive scale, which is a sweet spot for a document database. In the blog, my colleague Apoorva Anupindi highlighted that 80% of today’s data is unstructured, growing at a crazy rate of 60%. Support for JSON, the de facto data model for the web, therefore offers a critical opportunity for modern data platform providers, and Aerospike is doing just that.

Implementation of graph database

A NoSQL graph database integrates heterogeneous data from a variety of sources and makes links between different datasets. It does this by focusing on the relationships between separate entities and then surmising new knowledge from the information on hand. Our Director of Product Management, Ishaan Biswas, recently introduced Aerospike’s graph database offering that combines our database, the most scalable real-time NoSQL data platform, with Apache TinkerPop™, the most widely adopted graph computing framework. As highlighted in the blog post, most of the well-known graph database vendors today have adopted read-replica architectures, which are not engineered from the ground up to handle large volumes of data. Their performance degrades as soon as the dataset grows beyond the memory capacity of a single instance, leaving the database practically unusable for large-volume distributed deployments (case in point: telecom applications).

Today, we already support a non-telco customer for their identity graph requirement, which is critical to their fraud detection and prevention use case. From the telco space, one existing customer and another key prospect are talking to us to test/explore our graph database capabilities in the context of segmentation and micro-targeting – the telecom use cases under consideration include what advertisement should be shown to whom based on that user’s social interaction.

Rich connector ecosystem for SQL access and full-text search

Hot off the press, Aerospike announced our implementation of Aerospike Connect for Elasticsearch to allow developers to incorporate full-text search capabilities for real-time data. This follows our earlier announcement of Aerospike SQL Powered by Starburst, which made it possible for our customers to run massively parallel, complex SQL queries on the data stored in the Aerospike Real-time Data Platform.

Aerospike Connect for Elasticsearch provides seamless low-latency integration and granular access between Aerospike and Elasticsearch. It eliminates the need to build complicated tech stacks for heavy ETL processes or integrate with Pub-Sub systems to move data from Aerospike to Elasticsearch to leverage those technologies. This full-text search capability on data stored within Aerospike complements the existing query capabilities of Aerospike Expressions and secondary index query enhancements. In the world of connectivity, this has popular use cases for the telecom industry that include the indexing and searching of real-time time series data, for example, from IoT sensors.

On the other hand, Aerospike SQL Powered by Starburst provides a tightly integrated, distributed SQL analytics engine to the Aerospike database. Our Chief Product Officer, Lenley Hensarling, articulated the two-fold value for customers:

  1. Use Starburst to enable SQL access to data in the Aerospike database, where you can use a myriad of tools for reporting, analysis, and information visualization.
  2. Bring Aerospike data into the overall data mesh realized by the Starburst model.

Why is this important in a telecom use case? Telecom data is stored and accessed in a highly distributed manner. Ideally, it requires access technology without moving the data from its place. Also, telcos have enthusiastically embraced a microservices architecture – the deconstruction of applications in parts, allowing for each component to evolve independently, decomposes the overall view of the underlying data. Our support for telecom microservices is helping one of our major customers transform from a Communication Service Provider (CSP) into a Digital Service Provider (DSP).

As highlighted by our Director of Product Management, the release of Aerospike Connect for Elasticsearch and Aerospike SQL Powered by Starburst is part of the transformation of Aerospike from a database to a data platform. It builds on our existing portfolio of connectors to essential data pipeline components such as Kafka, Pulsar, JMS, Spark, and Trino. It also complements our already powerful query capabilities realized by our highly parallelized secondary indexes and precise Expressions.

Cross Datacenter Replication (XDR) going strong

Any discussion on Aerospike’s real-time distributed data platform will remain incomplete if we don’t discuss the Cross Datacenter Replication (XDR) feature. It provides dynamic, fine-grained control for data replication across geographically separate clusters, for instance, in edge and core locations. XDR consistently comes up in our conversations with major telco customers as one of the most-liked attributes of the Aerospike Real-time Data Platform. They especially refer to XDR’s ease of use, bi-directional filtering capability, and, most importantly, its very reliable operation. Last year, I highlighted XDR’s role in real-time mission-critical data management in the adjacent vertical of IoT.

Although the XDR feature has been around for some time, the enhancements by our product team have continued. Last summer, we announced enhanced throughput for XDR as a key aspect of Aerospike Database 6.1. The improvement kicks in when XDR enters recovery mode to catch up after network disruptions or when a specified last-update-time (LUT) triggers a rewind of a namespace. This is particularly beneficial to hydrate new clusters or transfer data between clusters with considerable write activity.

Energy & cost savings in cloud deployments via support for Graviton

Last but not least, Aerospike’s total cost of ownership (TCO) advantage was boosted with the recent release of Aerospike 6.2, as Aerospike Database 6 became available on AWS Graviton. Aerospike Founder and CTO, Srini Srinivasan, states that Graviton support further strengthens our partnership with AWS and dramatically reduces infrastructure costs for real-time applications, which, with the explosion of streaming data, have increasingly become more common in telecom. Graviton offers a dual benefit of greater energy efficiencies for certain CPU-intensive in-memory workloads and better cost performance. In a benchmark we ran with the support of AWS, the Aerospike Database processed 25M transactions per second on Graviton instances versus 21.1M transactions per second on equivalent x86_64 instances (an 18% improvement). The Graviton deployment also came out 27% less costly than the x86_64 one on an annual basis. These improvements are materialized along with a significant reduction in carbon emissions when running workload on a 3-node Graviton2 cluster.

Aerospike for telecom use cases (Real-time data management)

Aerospike’s journey to establish its database as the most versatile real-time data platform in telecom (and other verticals, for that matter) does not end here. New improvements and features/capabilities are continuously added – a case in point is our recent Database-as-a-Service (DBaaS) offering. On the scale dimension, we aim to make any implementation future-proof, which is a critical requirement for any major telecom use case.

For more resources:

Join us in Bangkok from March 14-16 at TM Forum’s Digital Transformation World (DTW) Asia for our session with the Viettel Group on how a future-proof charging infrastructure is allowing Viettel to deliver a seamless digital customer experience