A group of vendors and analysts released a new report tackling the impact edge computing is having on developers, telecom operators, internet service providers (ISPs), and end-users. It’s also an effort to simplify the semantics surrounding the ecosystem.
The report — State of the Edge 2018 — is sponsored by Packet, Vapor IO, Rafay Systems, Ericsson UDN, and ARM. It’s written by Edge Research Group Analyst Jim Davis and Structure Research Analyst Philbert Shih. Th focus is on architectures for edge deployments and what the technology looks like today, and gives guidance on the future of edge computing.
The report notes four principles that define the edge: it’s a location; it’s the edge of the last mile network; there is an infrastructure edge and a device edge; and compute will exist on both sides in coordination with the centralized cloud. As the edge ecosystem evolves and enterprises begin to take advantage of that evolution, it will impact the way software is used; how computing resources are purchased; and will shift expectations of service levels, physical security, hardware lifecycles, and refresh cycles.
“The CNCF community sees edge computing as a natural extension of cloud native practices, both on the infrastructure side as well as on the device side, “ wrote Chris Aniszczyk, CTO and COO of the Cloud Native Computing Foundation (CNCF), in the report. “Edge computing will drive a shift in how all applications are designed and managed.”
The report provides five broad implications and predictions for the future of edge computing. These include that edge will not be the end of the cloud and instead it will create a more open and collaborative model between the edge and the centralized cloud; the edge, specifically edge data centers, will bring a greater need for automation; edge computing will require a different supply and service chain; applications will need to become more network aware as locality evolves; and cloud resource purchasing will change with the adoption of a multi-tier marketplace.
These results will lead to the emergence of a new distributed computing architecture due to the adoption of specialized hardware paired with speciality software that is quickly replaced. The resulting supply chain will need to effortlessly deploy and manage a diverse portfolio of resources at a number of locations. This complex logistical and physical dilemma will need to be addressed with new hardware and software.
One of the compelling features of infrastructure edge computing is that, similar to centralized cloud, it can provide compute, storage, and network resources to application workloads. This will require greater access and control to every network layer, requiring applications to become more network aware, and will require service providers to evolve to meet this need.
“In a centralized model, a user might want to answer the question: did my packet arrive?,” the report states. “At the edge, the question is likely to evolve. How did it arrive? Which networks is the compute I’m using attached to? What is the past performance of those networks? … What is the latency to various points on various networks?”
The report also has an Open Glossary of Edge Computing that is stewarded by The Linux Foundation. It’s meant to encourage collaborative definitions in this new realm. There is also a market map of the Edge Computing Landscape to encourage the edge compute ecosystem to be inclusive.