Table of Contents
- Executive Summary
- Edge Server Offerings
- Field Test Setup
- Field Test Results
- ROI Analysis
- Conclusion
- About Lumen, VMware, and Intel
- About William McKnight
- About GigaOm
- Copyright
1. Executive Summary
Edge computing has emerged as a critical enabler in many business tech stacks, reducing the cost of improving and managing the customer experience. Edge computing allows organizations to move the most essential portions of an application as close to the user as possible to improve interaction and delivery of data and services. Reductions in latency that might go unnoticed to a single end user can prove financially beneficial, and even critical, in machine transactions or high data volume applications.
Edge solutions are being used to enable a wide variety of applications, including:
- Smart manufacturing
- Edge POS for retail
- Vehicle telematics across industries
- Traffic analytics for urban planning
- Security analytics in retail, manufacturing, healthcare, and others
These types of use cases also allow application designs that keep datasets and processing localized with the user, making room for additional data validation and security, and potentially reducing blast radius and data leakage while delivering tighter controls on privacy.
To ensure that edge-enabled applications can provide value to the business, it is important to test the latency of proposed edge servers (bare metal and for private cloud) to determine if they can deliver the required levels of performance and responsiveness. Latency is the time it takes for data to travel between two points on a network. If latency is too high, an application or service will suffer from poor performance. Testing for latency ensures that your application or service is running smoothly and that users have a good experience while also identifying network bottlenecks and other issues that can affect performance.
Most businesses want to avoid the overhead of owning hardware in multiple data center facilities worldwide. Edge bare metal and edge private cloud services address this concern, providing better performance, higher security, and more control over the underlying hardware. It is a solution consumed as cloud OpEx, maintained by the vendor as another cloud consumption service, and utilized as infrastructure as a service (IaaS) for your custom workload.
When we look at “ideal” application performance, we want to see a 20 millisecond (ms) round trip time (RTT) for application response that is visible to the end user. If we look specifically at network traffic, that RTT equates to a distance of about 1,000 miles—any greater and the data physically cannot get there quickly enough. However, when you consider the queries and responses necessary for the user to “see” a difference, the distance to achieve this RTT may in fact become far shorter. Moving the application so it resolves within this 20ms limit can offer a significant, measurable improvement in customer satisfaction and all the ancillary value that improving customer experience affords the business, without a substantial redesign of the application or stack.
This report focuses on the latency of two top bare-metal server providers: Lumen Edge Cloud Services (Lumen) and Amazon Web Services (AWS). For our testing, we defined a high-performance model, mixing applications that can process 1,000 transactions per second and demand a maximum latency of 5ms or less, while the NGINX reverse proxy handles HTTP requests.
On the edge with bare metal, with a minimum latency of 4.3ms, Lumen’s transactions had a 100% success rate, zero drops, and minimal latency. AWS also had a 100% success rate. Its transaction rate was higher by 16ms, taking a minimum of 23ms overall. This suggests that Lumen is a better choice for edge bare metal.
As we set a desired threshold of 20ms in total for application responsiveness, and this is a simple application, it is easy to see how the difference of 11.7ms would make a significant difference to your digital and customer experiences.
Our findings: Lumen Edge Bare Metal servers offer faster response times and reduced operational costs, as well as improved scalability, security, reliability, revenue, competitive edge, and reputation. In our analysis, these findings make Lumen an ideal edge solution for IoT devices, video conferencing, surveillance systems, and artificial intelligence.
A big part of this performance advantage is Lumen’s density of edge locations compared to AWS. Lumen simply offers more edge locations due to its primary business value as a large internet service provider.