By Sunay Tripathi, CTO, EVP Engineering, MobiledgeX

2020 - The Year of the Mobile Edge

Enterprise demand for edge computing already exists, as evidenced by a growing number of documented use cases across industries.  The conversation on how we respond to that demand must begin today.  This coincides with accelerating investments in 5G and the needed telecom network upgrades, thus creating a perfect storm for telecom and mobile to generate new revenues and new relevance that is greatly needed for the telecom industry.  While all edges are important, we believe 2020 is the year of the mobile edge for these reasons.  The scale revenues will not appear in 2020, but those that plant the seeds of the new business during that time will yield the largest crops in 2022 and onwards. 

Background

The transformation of information technology has been fueled by the parallel and synergistic development of massive numbers of increasingly mobile devices, public cloud and broadband cellular connectivity.

The next decade will bring ubiquitous XR experiences, the rise of the autonomous vehicle, data crunching IoT deployments, and AI and machine learning advancements that precisely tune every element of the world around us. It will be fueled not just by next-gen mobile broadband but by systematically moving cloud services closer to where they are being accessed - to the edge. This will require those services to be dynamically deployed when needed since the edge is a place of limited resources, unlike the cloud, which is perceived to be infinite.

Edge will be increasingly defined by the growing needs of more and more mobile devices delivering more immediate experiences, rather than more distribution of the cloud. Rather than thinking about an edge as the last bit of the infrastructure you see from the cloud, it is useful to think of edge as first part of the infrastructure that you see from the device.

It is important to understand that edge computing as a whole is not a new phenomena. Since the invention of the transistor, resources have either been centralized ( first with the mainframe) and/or distributed (first with PCs). As a natural evolution companie also invested in on-premise infrastructure and then co-location facilities were born, for sharing of common infrastructure.

Telecom has always had edge computing due to its need to deliver real time communication services that have had strict real time performance requirements that have needed to scale to total population and deliver service quality that is regulated by governments. The Internet was built on the distributed communication infrastructure provided by telecom, which then enabled cloud computing to create the largest scale sharing of computing resources, due the addition accessibility and agility it provided. To date, the cloud discussion has been dominated by the positive -- the myriad applications and services where cloud performance was adequate.

Increasingly the focus is on what additional solutions you can address with a better link (lower latency, more bandwidth, cheaper cost). Broadly, this falls into the discussion of edge computing, putting services increasingly near(er) the actual point of data consumption and generation (user or machine).

To simplify the world, we define four types of edge available to any application today. Each edge provides services and value versus associated cost and downside.

The net new change today that makes edge computing a disruptive concept is the telecom addition of edge compute and storage services alongside the ubiquitous adoption of cloud computing and mobile devices from the developer perspective. 5G promises the next level of performance and scale to an increasingly mobile user base, human or machine. To date, telecom has only provided communication services to people outside their industry. With 5G, the promise is that compute, storage and cloud services will also be made available, where these services are consumed in the same way as public cloud services today. This has the potential to dramatically change what is possible from a service, cost, and agility perspective, transforming the landscape of digital experiences once more.

The Developing Solution Landscape

Application and service developers that have sought to explore how edge-based computing can transform offerings and power new ones have by now run into familiar challenges. The edge isn't where they need it to be geographically, it is not easy to access and it is not being built in a common way.

The edge is not a replacement of public cloud, but a bridge between device and public cloud, which allows creation and expansion of applications based on computer vision and A.I. that deliver ubiquitous XR experiences, robotics, industrial IOT, and autonomous vehicles. To implement edge and take advantage of the new applications and capabilities to boost performance, an organization has to think about the following:

Edge Infrastructure as a Service (IAAS) on-premise or in-network

Building an edge infrastructure is pretty easy in today's world as cloud is getting commoditized and edge IAAS uses the same layers and economics of public cloud. Major cloud players have IAAS offering for edge like Amazon Outpost, Microsoft Azure Stack, and Google Anthos that can be installed on enterprise premises. Traditional IT vendors like VMware and Redhat also offer edge IAAS that can run on commodity hardware from Dell, HP, and whiteboxes. The leading network operators are also implementing edge IAAS in their network so enterprise customers can avail these capabilities similar to how they consume public cloud but the infrastructure is very close to them and in network. AT&T has invested in the Linux Foundation Edge effort via the open source Akraino project that integrates Openstack and various other open source software to create an integrated solution for both enterprise on-premise deployment, as well as operator in-network deployments, so enterprises can benefit from cloud-like economics.

Connectivity Options - Public/Private LTE/5G/wifi

Depending on the campus size and use of drones or autonomous vehicles require LTE/5G based connectivity. This is where 5G brings new capabilities and private spectrum to the table where enterprises can take advantage of either on-premise or in-network edge IAAS, while taking advantage of the privacy and trust the cellular network offers. The mobile approach is quite different from that of the Internet. The connection between an application and its backend service is not an anonymous connection but a trusted connection over a secure and trusted operator network if using public LTE/5G but same benefits are available in operator-led private implementations (similar to what we get when we order home broadband from AT&T, Verizon or T-mobile). Hybrid solutions, which are part WiFi and part LTE/5G, are also becoming available.

Privacy, Security and Trust Considerations

One of the key considerations of immersive applications is the use of mission-critical video feed and privacy. The AR glasses, shop floor video processing, robotics and autonomous vehicle solutions are all using sensitive video feed to offer insights and improve productivity. Having an edge solution, which leverages the application deployment and management model similar to public cloud but where the application is totally firewalled from the internet solves the privacy concerns and allows the enterprise to take advantage of the new generation of productivity applications, This is where using a private 5G for connectivity also increases the security and trust profile of the application. The cellular control plane is quite different than WiFi-based connectivity because the users and devices use SIM card and IMEI number as identity and there is a strict admission control before the user can access the services, unlike WiFi which is easily spoofed. There is a need for software capabilities that work equally well with WiFi and cellular connectivity, but take advantage of cellular control plane and identity and location verification when present, to ensure a higher degree of security and trust.

Edge Platform as a Service (PAAS) for Application Deployment/Management

There are a plethora of open source and commercial solutions available to deploy and manage your applications. Vendors are creating open source device SDKs to integrate with a wide variety of edge applications and new generation of devices for enterprise deployment and Edge PAAS layer to offload while taking into account latency, privacy and trust that is required for this class of applications that see and process the mission-critical enterprise applications. The edge PAAS layer makes compute seamlessly available to end users and applications so they have the same experience as if running in a public cloud but the device and clients actually utilize the virtualized/containerized backed close to users.

Summary

Applications that have been designed only in the context of the existing public cloud will incrementally benefit from the new capabilities of the edge-cloud. The even greater value envisioned today will ultimately be discovered by the application and service builders that actually develop edge-native applications.

About the Author

Sunay Tripathi is the CTO and executive VP of engineering and product at MobiledgeX. He has an extensive 25+ years of software background, and products created by him are currently used in mission-critical environments such as banks, telecom networks and data centers worldwide. He holds more than 100+ patents in areas of containers, virtual switching and software-defined networking.

Prior to MobiledgeX, Sunay founded Pluribus Networks where, as founder/CTO, built a distributed network operating system, Netvisor, to run on switches and bring software-defined networking (SDN) to the datacenter world.

His early career was at Sun Microsystems where he was the chief architect for kernel/network virtualization. He created container networking and virtual switching that became the mainstay of today's microservices and cloud technologies.