Benchmarks from DigitalRoute for Achieving Cost Savings with Serverless and AWS Graviton

By Amanveer Singh, Sr. Solutions Architect – AWS
By Jonas Wallenius, Strategic Product Manager – DigitalRoute
By Jenny Sjöström, Industry Marketing Manager, Telecom – DigitalRoute

DigitalRoute

Telecom operators are increasingly migrating to Amazon Web Services (AWS) and cost savings is one of the primary drivers.

In this post, we will see how DigitalRoute’s benchmarks show that by migrating to AWS and taking advantage of services like AWS Fargate and AWS Graviton, telecom operators can retain their on-premises performance while seeing cost reductions when deploying the Usage Engine Private Edition on AWS.

DigitalRoute is an AWS Partner and AWS Marketplace Seller that’s collaborating with AWS to enable telecom and communication service providers (CSPs), as well as any company processing usage or financial data in their quote-to-cash process, to manage the flow of usage data between the core network and key business and analytics systems.

This ensures records are not lost, duplicated, or corrupted—specifically, for telecom operators DigitalRoute provides solutions for 3G, 4G, and 5G architectures.

Private Data Centers: From Differentiator to Liability

CSPs are in the middle of rolling out and upgrading their 5G networks. Flat revenues and ever-increasing data volumes have created financial headwinds, mandating the need to optimize costs and increase the speed of execution. Powering the 5G roll-out and all previous network generations are massive data centers.

Hardware

It’s no small task to operate a data center. The physical aspects range from electricity and cooling to physical security, cabling, networking, servers, storage, maintenance, upgrades, and replacements. On top of this comes things like redundancy, failovers, backups, and disaster recovery.

Virtualization

When it comes to virtualization, the telecom industry is moving on from using virtual network functions (VNFs) based on virtual machines into container network functions (CNFs), where Kubernetes has become a de-facto standard for how to operate and virtualize large data centers.

Supporting Services

Above this infrastructure layer of containers, compute, storage, and networking resides the application layer. Which services are available for containers executing in the data center? Which databases are there to choose from? Is there a messaging queue? An archiving service? A data warehouse? A container repository? Which versions are available? Who maintains and upgrades them?

A cloud-native application based on a microservices architecture requires a lot from the environment it’s deployed in, and the modern DevOps practice of continuous integration and deployment (CI/CD) even more so.

AWS as Cloud-Native Standard

AWS provides all of the above, with a rich set of backing services as a de-facto standard for how to build and run cloud-native applications. This is why vendors are providing AWS-ready solutions and why many telecom operators are planning to migrate their workloads to AWS. Some have already done it and are now reaping the benefits.

DigitalRoute in the Quote-to-Cash and Telecom Landscapes

DigitalRoute’s Usage Engine unlocks the value of usage data to optimize business outcomes. See Figure 1 for an industry-agnostic overview of where DigitalRoute’s Usage Engine fits into a customer’s quote-to-cash landscape.

Figure 1 – DigitalRoute’s Usage Engine in the quote-to-cash landscape.

Now, let’s map this landscape view to a telecom-specific quote-to-cash environment.

Figure 2 – DigitalRoute’s Usage Engine in the telecom landscape.

Running Usage Engine Private Edition on AWS Fargate

DigitalRoute’s Usage Engine Private Edition is elastic, taking advantage of Kubernetes for scaling in and out. It runs cloud natively on top of Amazon Elastic Kubernetes Service (Amazon EKS), which is a leap forward from the heterogenous private telecom cloud environments.

Although Amazon EKS offloads Kubernetes control plane installations and operations, in addition to providing high availability and scalability for the control plane, operating a large-scale telecom application on top of a large EKS cluster requires significant skills. You must still set up the underlying Amazon Elastic Compute Cloud (Amazon EC2) groups, perhaps with auto-scaling and min/max allowed compute and memory values, which can lead to over-provisioning of compute resources.

To simplify operations further, AWS Fargate is a serverless compute engine for containers and eliminates the need to provision and manage servers. DigitalRoute’s Usage Engine Private Edition can be run on top of AWS Fargate, and serverless is a step function in the simplicity of operations compared to all previous paradigms.

The following diagram shows a typical serverless Usage Engine Private Edition deployment on AWS Fargate.

Figure 3 – Usage Engine Private Edition on AWS Fargate.

Besides the simplicity of running Usage Engine Private Edition in serverless mode on AWS Fargate, we also note some details about this deployment.

Step-Wise Core Network Migration

In this example, the 3G/4G core network remains on-premises and is accessed via AWS Direct Connect, along with other on-premises systems.

The standalone 5G core network runs on AWS, as does the analytics and billing systems. Many telecom operators begin their AWS journey by migrating analytics applications, then billing, and then the core networks.

Redundancy and Archiving

For redundancy, all backing services run across multiple AWS Availability Zones (AZs) for regional redundancy and disaster recovery. This is always the case for Usage Engine Private Edition, whether using AWS Fargate or not.

Charging detail record (CDR) archiving is done cost-effectively using Amazon S3 Glacier.

Operational Concerns

For performance, if a use case has high I/O demands, AWS Fargate can be slower since storage must be handled using Amazon Elastic File System (Amazon EFS). Running EKS on top of EC2 with Amazon Elastic Block Store (Amazon EBS) volumes attached provides higher I/O performance at the trade-off cost of additional complexity to achieve redundancy since EBS is not regionally redundant.

For observability, using AWS-native tools like Amazon CloudWatch works well with AWS Fargate, but alerting capabilities are somewhat limited. For full alerting capabilities, DigitalRoute recommends using a tool like Amazon Managed Service for Prometheus.

Running Usage Engine Private Edition on Graviton Processors

A major benefit of running workloads on AWS is that operators get to partake in advances to the underlying infrastructure “for free” as AWS improves and adds services. In addition to the typical x86-based instances, AWS offers Arm-based instances which are based on the AWS Graviton processor and provide extra cost-performance benefits.

AWS Graviton processors are designed and optimized for cloud workloads. Together with the Nitro security chip, Nitro card, and Nitro Hypervisor, Graviton-powered instances will be the best within their instance families and can deliver up to 40% better price performance over comparable current generation x86-based instances.

A lot of AWS managed services also support Graviton natively, so the price-performance benefits can be passed to customers with little porting effort.

DigitalRoute’s Usage Engine Private Edition runs on Graviton processors in Amazon Elastic Container Service (Amazon ECS) and EKS. All that’s needed is choosing Graviton-based EC2 instances when setting up a new Usage Engine Private Edition deployment.

Benchmarks

The above architecture has been benchmarked on AWS Fargate and Amazon EKS with EC2 instances, using eight vCPU pods with 16GB, at 500k TPS for a real-time use case and 23k TPS for a batch use case with Amazon EFS for storage.

Note that DigitalRoute did not push Usage Engine Private Edition to its limits in this benchmark. They have production use cases running more than 10x these volumes.

All benchmarks have been performed in the AWS Ireland (eu-west-1) region.

The following use cases and traffic patterns have been used for benchmarking.

Business Benefits

There are several clear business benefits for telecom operators to run Usage Engine Private Edition on AWS Fargate and Graviton.

Cost Savings from Using AWS Graviton Processors

Running Usage Engine Private Edition on AWS Graviton2 or Graviton3 processors saves ~20% of compute cost compared to using Intel-based processors, according to DigitalRoute benchmarks, for all use cases and traffic patterns in our benchmark.

Note that AWS Graviton processors are currently not available for AWS Fargate on Amazon EKS. The cost savings are achieved running Usage Engine Private Edition on EKS with Graviton EC2 instances, compared to running Usage Engine Private Edition on EKS with Intel-based EC2 instances.

Cost Savings Using AWS Fargate for Variable Workloads

For workloads with significant variance between peak and off-peak traffic, even more cost savings can be achieved using AWS Fargate. When running serverless, there are no fixed compute costs, no over-provisioning, and no idle resources. Operators only pay for the actual compute usage at any given moment.

While circumstances depend on the use case, DigitalRoute’s benchmarks show the following cost savings for compute on AWS Fargate compared to running Usage Engine Private Edition on EKS on top of on-demand Intel-based EC2 instances without auto-scaling, mirroring the situation in a private data center.

As can be seen in the last traffic pattern, AWS Fargate is not suitable for constant-load, 24/7 workloads. They are better served using Amazon EKS with pre-committed EC2 instances matching the compute needs.

Note that if comparing using AWS Fargate with using EKS on top of on-demand EC2 instances with auto-scaling enabled, the cost savings are smaller; the main benefit in that case is the AWS Fargate resource granularity is 0.25 vCPU and 0.5GB, while the smallest EC2 granularity is 1 vCPU and 1 GB. This means AWS Fargate scaling can use compute resources more optimally than EC2 scaling can, with less over-provisioning needed.

The cost savings are based on on-demand EC2 instance pricing. Using pre-committed instances or Spot market instances would reduce these cost savings somewhat, but they would still be significant.

Elastic Scaling

AWS Fargate abstracts away the complexity of fine-grained horizontal scaling from the operator, providing access to virtually unlimited computing resources and allowing them to scale their operations to meet changing demand. There are no lead times for extra capacity.

This also simplifies operations, which can drive cost savings depending on the operator’s staffing.

Other Benefits

Beyond these benefits, operators can gain additional benefits from running Usage Engine Private Edition on AWS:

• Deploy containers using CI/CD, speeding up iterations and innovation speed.
• Get access to the full catalog of AWS services, from data lakes to machine learning to message queues and more.
• Achieve regional redundancy and disaster recovery by deploying across multiple Availability Zones and using a multi-AZ relational database instance for data storage.
• Security—AWS has strict security practices to protect your data both at rest and in motion. AWS Fargate abstracts and handles security for the Kubernetes control plane nodes and etcd database, and meets compliancy requirements for SOC1, SOC2, and ISO27001. AWS provides controls, access policies, and encryption for its services, and separates traffic and control planes in your virtual private cloud (VPC). See AWS Cloud Security to learn more.

Conclusion

For cost benefits and simplified elastic scaling on AWS for your telecom mediation workloads, deploy DigitalRoute’s Usage Engine Private Edition running in serverless mode on AWS Fargate.

Even if AWS Fargate is not right for you, you can still take advantage of cost savings by running Usage Engine Private Edition on AWS Graviton processors.

You can learn more about DigitalRoute and Usage Engine Private Edition in AWS Marketplace.

.

.

DigitalRoute – AWS Partner Spotlight

DigitalRoute is an AWS Partner that’s collaborating with AWS to enable telecom and communication service providers (CSPs) to manage the flow of usage data between the core network and key business and analytics systems.

Contact DigitalRoute | Partner Overview | AWS Marketplace