Chat with us

WHEN MILLISECONDS MATTER:

LEVERAGING THE POWER OF EDGE COMPUTING TO DRIVE YOUR BUSINESS FORWARD

WHITE PAPER

EXECUTIVE SUMMARY

While cloud computing has achieved widespread adoption, many organizations are bumping up against the limitations of its centralized nature. New technologies that demand sub-second response times, an explosion of Internet of Things (IoT) devices and data, and user expectations for consistent high-quality online experiences are pushing the industry toward a new model: edge computing.

 

Certainly, the proven advantages of traditional cloud computing mean it will continue to play an important role in the technology ecosystem. But for some applications, moving processing power from a central location to the edge of the last-mile networks is critical to success. For these applications, milliseconds matter. A centralized cloud architecture, for all its power, can’t keep up with the real-time responsiveness demanded by today’s connected devices. The edge computing model reduces the latency that arises when data travels significant distances between the user or the source of data and the compute resources (or application) in the data center.

 

But setting up an edge computing infrastructure creates new challenges for organizations. The process is time consuming, costly, and demands considerable expertise while diverting valuable time and resources from core business activities. This white paper explores the factors driving the growth of edge computing and what it takes to implement an effective solution. It also takes a closer look at how Limelight Networks’ edge cloud services can help you quickly and cost-effectively leverage the benefits of edge computing.

WHEN CENTRALIZED CLOUD INFRASTRUCTURES ARE NO LONGER ENOUGH

Organizations have flocked to Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS) solutions because they provide fully managed services in an attractive OpEx model. Organizations no longer have the significant up-front costs and hassles of buying, installing, configuring, monitoring, maintaining and upgrading hardware and software in the data center. However, depending on data usage patterns, transfer and transaction costs can add up to a significant percentage of the monthly invoice — to the surprise of many organizations. In addition, because modern cloud computing centralizes processing for applications in cloud-based mega data centers, the distance between the data center and the user or data source can create significant problems, depending on the application’s latency requirements and the user expectations.

IoT APPLICATIONS

The installed base of IoT devices is forecast to grow to almost 31 billion worldwide by 2020. 1

 

 

Drones, security cameras, cars, healthcare monitors, industrial robots, and a host of other devices incorporating IoT sensors generate vast amounts of data that is tremendously valuable, especially at the moment of its creation. The challenge is figuring out how to process the data quickly enough to derive maximum value from all of the information that is generated. While the cloud excels at performing complex computations, sending the data from each device to the cloud for processing is simply too slow. When milliseconds matter, taking too long to process the data makes it useless.

A NEW ARCHITECTURE MEETS THESE DEMANDS—EDGE COMPUTING

Edge computing is a new architecture that addresses these challenges. It dramatically improves the user experience by moving processing power from a central point to the network “edge,” closer to the source, or more specifically, next to the last-mile internet connection. Data no longer has to wind its way through the internet back to a central location that, in most cases, is not close to the device or user.

 

Figure 2: Leveraging the Power of Edge Computing

 

As with any emerging technology, the edge computing model is still evolving. Currently, three approaches are being discussed in the market:

  • Some suggest edge computing will return to housing server hardware in on-premise data centers. While on-premise data centers may provide the low latency, near real-time compute experience required by many applications, organizations have become accustomed to the advantages of low CapEx and outsourced management that are offered by cloud deployments.
  • A second model brings cloud-based data centers closer to end-users worldwide, combining the advantages of edge computing with those of a cloud model. Organizations can leverage the benefits of a managed data center and an efficient OpEx model without sacrificing the low latency required by the application, device, or end-user.
  • Since edge computing is not appropriate for every scenario, a third model suggests combining the advantages of edge computing with those of a cloud model. With this deployment methodology, time-sensitive functions such as the collection and processing of real-time data can be handled at the edge, while elements of the application that benefit from being centralized, such as back-end services, centralized command and control, and analysis of aggregated data-sets can be managed in traditional cloud data centers.

 

Since most organizations have limited internal resources available to deploy and maintain their own on-premise data centers, we do not expect the first edge computing approach to gain traction with anyone other than the few largest global IT service providers. Limelight is focused on enabling organizations to easily leverage readily available globally distributed edge computing resources that can be interconnected and/or integrated with centralized computing capabilities as described in the approaches above.

EDGE COMPUTING IN ACTION

IOT APPLICATIONS

IoT devices produce vast amounts of data and transfer the data to a local data center where it is processed instantly. For example, a healthcare facility may have thousands of IoT devices such as patient monitors that constantly collect patient data. Edge computing quickly processes that data as it comes off the sensors to filter and analyze it in real time. It can then send out alerts for things that require immediate attention, filter out the “junk” data to reduce the amount of bandwidth used and data that needs to be stored, or identify patterns quickly, allowing faster response to issues.

 

Once the data center at the edge performs these functions, all or a portion of the data is sent to the central processing or storage repository in the corporate data center, colocation facility or IaaS cloud. The cloud-based data center, as a centralized repository, can provide the necessary data sets to derive patterns and conclusions about the data collected. By deploying edge computing, the healthcare facility can share the analytical resources across multiple facilities.

REAL-WORLD EDGE COMPUTING USE CASES

Here are several examples of how leading companies are putting the power of edge compute to use.

  • IOT INDUSTRIAL INSPECTION AND MONITORING

  • A company uses drones and robots to inspect infrastructure such as pipelines and power lines. It must ingest and process large amounts of data in real time to identify anomalies that require further investigation. Edge computing resources geographically located near the inspection sites provide real-time processing and analysis that reduce costs of manual inspection and minimize downtime.

  • VIDEO SURVEILLANCE AND LIFE SAFETY

  • A company performs video surveillance and incidence detection across many locations. It must ingest and process large amounts of data instantly to detect potential security events as quickly as possible. A comprehensive edge computing solution gives the company the necessary compute capability and responsiveness plus extensive last-mile connectivity across multiple ISPs, without a large CapEx investment.

  • UNIFIED COMMUNICATIONS

  • An online conferencing vendor serves users across the globe. Long load times, rebuffering or delayed audio caused by network latency can damage the user experience, leading to expensive support tickets and potentially high customer churn rates. An edge computing solution can dramatically reduce network latency, enabling consistently high performance resulting in positive user experiences worldwide.

  • ONLINE GAMING

  • An online gaming company needs to provide the fastest possible online response times to its interactive gaming customers to maximize user experience. By utilizing distributed edge compute resources located close to its users that are tied together with a private backbone and connected to all the major ISPs and end-user networks, gamers enjoy the fastest real-time and interactive online gaming experiences that keep them coming back for more.

CHALLENGES WITH DEPLOYING AN EDGE COMPUTING INFRASTRUCTURE

While the benefits of edge computing are tremendous, developing, deploying, and operating your own edge computing environment is a major undertaking that requires a significant financial investment as well as deep technical expertise. Challenges include:

  • Establishing last-mile connections. Users, and by extension their devices, are always on the move and are constantly connecting to different Internet Service Providers (ISPs). Building robust, reliable connections in a global market that enables a service to connect with all of the ISPs in a region is time-consuming, expensive, and can be a major distraction for an organization that needs to focus on other business-critical issues.
  • Building a backbone. The network backbone allows edge sites to connect to cloud instances of compute and to each other for failover/business continuity or to share datasets and unify distributed edge computing assets. Implementing the backbone requires finding a telco provider and provisioning a physical circuit, a process that can take several months. Again, this forces the organization to shift resources from developing and marketing its product to building out an infrastructure.
  • Building the data center. Colocation facilities simplify data center build outs, but the process still requires significant planning, budget, and effort to establish a production environment. Elapsed time from contract to ‘go-live’ is measured in weeks or months - even for a relatively simple deployment.
  • Staffing. Organizations also need to have knowledgeable staff in place to manage the setup and ongoing operations of all edge compute processes and resources. Thorough analysis is required to determine whether the cost and effort of engineering, building, and managing a global infrastructure is the best use of their capital.

LIMELIGHT NETWORKS’ EDGE CLOUD SERVICES

Limelight Networks’ edge cloud services allow organizations to spend their time, effort, and money scaling their businesses instead of scaling the underlying infrastructure. We apply our experience in delivering latency-sensitive digital media to edge compute applications that demand the highest performance, regardless of locations or device type. We also collaborate with our customers to deploy their application close to the user or edge device, while maintaining compliance for country data sovereignty.

 

Limelight provides the four most critical elements necessary in a successful edge compute solution:

  • A private, secure backbone network that connects the central cloud resources to all the edge compute sites and all the edge computing sites to each other.
  • An established and comprehensive network of last-mile service providers to efficiently service end users.
  • Flexible compute options—the ability to either rent compute resources as with cloud-based models or to use your own hardware as with co-location models.
  • More than a decade of experience building out and operating one of the world’s largest global distributed networks, including network operations, monitoring and 24/7/365 live global support.

PRIVATE MANAGED NETWORK BACKBONE

Limelight’s private network - one of the world’s largest - allows data to bypass the public internet and travel at high speed across the globe securely and reliably. Our ongoing investment in the development and optimization in all components of our network ensures industry-leading performance consistently, everywhere in the world.

CONNECTIONS TO ALL THE LAST-MILE PROVIDERS

Limelight provides businesses with one-hop connectivity to 900+ global ISPs by leveraging over 80+ Points of Presence (PoPs). Whether your users and devices are utilizing fixed broadband, wireless, or even emerging 5G connections, Limelight brings computing as close as possible with sub-second response times that reduce latency.

DATA CENTERS AT THE EDGE

With Limelight’s flexible compute options, customers can either rent compute resources in our data centers or deploy their own hardware. In addition to managing all aspects of data center operations for customers, our experts can help ensure compliance with regional data sovereignty regulations.

CONCLUSION

Edge computing offers a powerful new solution for applications that demand sub-second response times. By putting the compute power at the edge of the network, organizations can dramatically reduce latency and significantly improve user experiences. By turning to Limelight’s edge cloud services, organizations can easily leverage the benefits of a private, high performance edge compute solution and focus on building their business rather than building out a network.

 

For more information about Limelight’s IoT and Edge Cloud Solution visit: https://www.limelight.com/iot-and-edge-cloud/

1https://www.statista.com/statistics/471264/iot-number-of-connected-devices-worldwide/