Streaming Games to the Edge without Dedicated Hardware

Streaming Games to the Edge without Dedicated Hardware

Welcome to our new blog series: Edge Cloud Thought Leadership. This collection of blogs will focus on new technologies and industry leaders in the edge computing market as way to educate a broad IT audience in the future capabilities at the edge.

Company: Blacknut

Blacknut aims to fundamentally change the video game industry by creating a service where users can access a vast catalog of curated games, play without limit across all devices, and no longer have the need for dedicated hardware, all for a simple subscription.

Using the latest cloud computing technology, Blacknut digitizes video games, delivering them through the cloud directly to smart devices such as televisions, tablets, and phones or through devices like set-top boxes and dongles.

Blacknut CEO Olivier Avaro talks about their use of edge computing to enhance their game streaming offering. The video is in French; press CC for English captions

Podcast Highlights:

  • Working at a Gaming Company
  • Game Streaming and Experience
  • Elimination of H/W Limitations for Modern Games
  • Delivery of Game from Cloud vs Edge
  • More than 50% of 5G traffic expected to be Gaming
  • Edge/Cloud Infrastructure Requirements
  • Virtualized Game Platform
  • Challenges in Building the Blacknut Platform

To see more thought leadership videos from the Global Edge Forum event visit our YouTube playlist: Global Edge Forum 2019.

Previous Blogs in this Series:

·         Building Applications for the Edge with a Datacentric Model (3/12/19)

·         Multi-Cloud Management at the Edge (3/19/19)

·         Deploying Containerized Apps at the Edge (or anywhere) (3/26/19)

·         Edge Cloud – Powering Digital Transformation (4/2/19)

·         Global Reach at Massive Scale for Game Studios and esports Competitions (5/15/19)


Edge Cloud - Powering Digital Transformation

Equinix Builds ‘Network Edge’ for Global NFV

This article appeared on Light Reading on Feb 27, 2019

Equinix is launching a global NFV infrastructure as a foundational building block in its strategy to partner with enterprises and service providers in the transition of enterprise workloads to the public cloud.

Equinix has launched its network functions virtualization infrastructure (NFVi) platform in Silicon Valley, California, as well as Washington DC, Amsterdam and London, with plans to bring Singapore and Sydney, Australia, online soon. The initial use cases are virtual router and firewall virtual network functions (VNFs) from Cisco and Juniper, with SD-WAN VNFs in the pipeline from those vendors. Equinix is also working with security vendors Palo Alto Networks, Fortinet and others, Equinix says.

Equinix sees NFV as an extension of its core mission of providing network-to-network connectivity for enterprises, telcos and other communications network service providers, and cloud platforms such as Amazon Web Services and Microsoft Azure, Jim Poole, Equinix VP of business development, tells Light Reading.

“The value of Equinix lies in positional value,” Poole says. “We sit between networks and the cloud. We create a physical PoP that is a neutral control point between a corporate network and the cloud network.”

The NFVi platform launched in private beta in October, and public beta in December. A limited release will come later this year, with a plan to go into general availability in September, Poole says. Initially, the program will be available to enterprise customers, with general availability to both enterprises and operators in the third quarter.

“The goal is not to replace operators. If you're a big enough enterprise customer — the global 200 — you come directly to us, but we would be just as happy if customers come to us through an operator,” Poole says. Equinix's core enterprise customers are organizations large enough to essentially run their own carrier networks for internal use, and those are the initial users for Network Edge. Later, Equinix hopes to provide point-and-click capabilities that will enable its carrier customers to offer the VNFs as white label services to their own enterprise customers.

Beyond virtualized router, firewall and SD-WAN, Equinix exploring operator-specific VNFs, such as virtual EPC, IMS and MNO core functions. “Ninety percent of all mobile traffic today terminates in a cloud,” Poole says. Equinix sees its position between cloud and operator data centers as ideal for running those kinds of VNFs.

Initially, Equinix is running the VNFs on OpenStack. Poole declined to name the vendor. “We may not stick with them,” he says. Equinix is also considering supporting containerized VNFs with Kubernetes, and bare metal.

In addition to working with enterprises and carriers, Equinix also serves vendors who are transitioning from hardware to recurring services business model. For example, it's providing space, power and interconnect services for Ericsson's Edge Gravity and Nokia Wing Internet of Things orchestration platforms, Poole says. Ericsson and Nokia provide hardware and software.

Equinix sees its interconnection capabilities as a platform enabling it to offer new kinds of services to enterprises, cloud providers, telcos and other communication network providers. For example, Sentara Healthcare provides a customer portal accessible via Amazon Echo smart speakers that connect to both Azure and AWS, with Equinix at the center. Equinix also provides Sentara with access to Microsoft 365, SAP and Salesforce.

– Mitch Wagner, Executive Editor, Light Reading

Ericsson must not waste potential of Edge Gravity with narrow telco focus

Ericsson must not waste potential of Edge Gravity with narrow telco focus

This article appeared on ReTHINK Wireless Watch on Nov 20, 2018

The place that edge compute will play in the telco’s network, and its business model, is a topic we have been tracking with interest this year. While there is a strong logic to convergence of compute, storage and connectivity – all in locations close to the user to improve service response times – it is not clear that operators will always take the leading role in deploying and monetizing edge networks.Many industrial and IoT applications will rely on edge locations that go far beyond the sites and central offices in a telecoms grid, and as some of those sectors look for edge cloud resources on a global scale, the real power may lie with organizations that can aggregate physical assets from many owners – not just operators – and create a scalable, flexible, virtualized platform.A large operator could perform this role, as could a webscale provider like Amazon, or an industry-specific player, such as GE in the heavy manufacturing segment. Another type of company which could take on this powerful coordination and orchestration role is the equipment vendor. Nokia is combining its network and cloud products with its orchestration and virtualization software to form the basis of various edge-oriented cloud services for industries, with or without the involvement of operators (see lead item).

And Ericsson is responding with Edge Gravity, its own take on a strategy which seeks to exploit its expertise at the intersection of networks and the cloud, while keeping the plum role in the value chain for itself, not its telco customers.

Edge Gravity is an extension of the Swedish company’s Unified Delivery Network (UDN) initiative, which has its own roots in the firm’s acquisition of Microsoft MediaRoom in 2013, as the foundation of a push into the media business. Ericsson has pulled back from that sector considerably, offloading a controlling stake in the division to an equity firm in January.

However, it is clearly looking for ways to harness some of the technology assets more widely, and there is a logical progression from the original UDN, launched as an integrated content delivery network (CDN) in early 2016, and Edge Gravity. For many telcos, after all, the most immediate application of edge compute is to support CDNs that are closer to the users and can deliver lower latency, higher quality mobile video.

Edge Gravity is presented as a far broader platform though, with the potential to support many latency-sensitive or personalized applications that benefit from being closer to the device. The idea is to use UDN-type underpinnings to create a virtualized, global network of edge cloud assets, linked to data centers, and each other, via the wired or wireless networks of more than 80 operator partners. These include MNOs, wireline telcos and cablecos, said Ericsson, and in future might also add private and industrial network providers.

The initial foundations of Edge Gravity are built around a core of 22 central locations,  based in Equinix data centers and on Ericsson’s own servers and MPLS backbone. The software stack is mainly open source, using OpenStack and Kubernetes. These are connected to the edge networks supplied by over 80 partners. These will offer rack space, power, connectivity and other capabilities depending on location and requirement. The combined network has about 4Tbps of capacity so far.

Among the service providers on that list of 80+ are Vodafone group, China Unicom, Telstra in Australia, Canada’s Rogers, Telefonica group, NTT Docomo of Japan and Bharti Airtel of India.

“Rather than selling to the service providers, we are making them suppliers to us,” Yves Boudreau, CMO for the Edge Gravity unit told LightReading. “They can’t act globally as a webscale company can. This is our attempt to help our customers and the service providers come to the table collectively.”

Edge Gravity is part of Ericsson’s Technology & Emerging Business unit, which focuses on new activities housed in ‘accelerated units’ – designed to have the autonomy to behave more like start-ups. It is the first unit to use this model, and so mimic the norms of the cloud and web worlds rather than telecoms. Boudreau said it was harnessing the agility enabled by virtualization and resource flexibility, to drive global scalability.

The firm expects a wide range of companies, in sectors from content delivery to smart cities to automotive, to want to use the edge cloud platform. It might support their internal decision making (for instance, predictive maintenance in near -real time for a transport operator) or allow them to deliver new or improved services to their own customers (e.g. better video and low latency gaming) without having to build their own edge clouds.

Initial customers include Limelight Networks, which will use the network to expand the reach of its content service; Net Insight, and two start-ups, Haste and Mode. The former has built a low latency service for gaming, and the latter has developed algorithms to accelerate the decision-making process for routing low latency requests, in applications such as collaboration.

This is a similar model to that pursued by the webscale providers, and the ideas of elastic scalability, and global aggregations of cloud resources, are familiar from AWS and Azure. But the cloud giants are not necessarily the best placed to extend the model they dominate to the edge. Their power has rested on investing in vast data centers, but in many areas, their ability to access edge sites is inferior to that of telcos or other enterprise types. That is not to say AWS and others will not seek to keep pole position by turning themselves into the orchestrators of many others’ edge cloud and connectivity assets, but they are not the only game in town.

Nokia has worked extensively with AWS in the edge area, particularly on integrating its MEC capabilities with the Amazon unit’s Greengrass developer and cloud services ecosystem. That has raised the prospect of an alliance between these two companies to pool their respective strengths and offer a powerful combination in the edge cloud, almost certainly confining operators to a supporting role.

Ericsson could leverage Edge Gravity for a similar strategic relationship, or try to dominate the whole value chain itself.  It will experiment with various value chain positions, first within the starting segment of content delivery, and then moving into other, more industrial applications. In its first model, it will charge fees to the content provider for the edge CDN resources that should improve the quality of delivery, and therefore customer satisfaction. It will then share part of those fees with the connectivity provider.

Ericsson, at its recent Capital Markets Day, was keen to emphasize that efforts like Edge Gravity do not signify a return to the diversification strategy of its previous CEO, Hans Vestberg, now at Verizon. Current CEO Börje Ekholm reversed that approach, triggering the sale of the media business and a pull-back from enterprise expansion, especially activities which might involve selling directly to vertical sectors, rather than via operators.

So Edge Gravity, despite its potential to be the starting point for an AWS-style business that looks well beyond telcos, is clearly positioned in the traditional, telco-driven market of content delivery, and was positioned by Ericsson as part of a cautious move into “close adjacent markets”.

It is clinging to the change of heart it announced with Ekholm’s plan – to stop targeting enterprise customers directly, by contrast with Nokia and Huawei (though Nokia has also, in public statements if not behaviour, pulled back from appearing to challenge its own core customers). In 2017, Ericsson said: “We will build our IoT business with service providers, addressing industries based on use cases. We will continue to address clients outside of the telecom industry through our service provider customers.”

However, the risk is that, if it sticks too rigidly to that mantra, it will be confined within the same limitations as its main clients. In the edge, a platform like Edge Gravity has the potential to support broad new revenues based on a whole new position in the value chain. But just as telcos’ approach and assets leave them stuck with video delivery as their main edge compute service, rather than higher growth applications in industrial spaces, so Ericsson could clip its own wings in the same way with tactics that are too narrowly focused on telcos.

That is why, for now, Ericsson’s hopes of financial turnaround remain heavily – probably too heavily – reliant on 5G contracts. So far, it remains confident of its strategy, as do the markets – Ericsson’s shares have gained 50% in value this year. Ahead of the Capital Markets Day in New York, the firm raised its sales target for 2020, citing a 5G-driven increase in telco capex from 2019, gains in market share, as well as new revenues from the “adjacent” sectors. It maintained its target of a 10% operating margin across the entire business but raised its sales goal to SEK210-220bn ($23.4bn to $24.5bn), up from SEK190-200bn ($21.2 to $22.3bn).

Ericsson now expects its overall addressable market to grow at a compound annual growth rate (CAGR) of between 1% and 3% between 2018 and 2022.

The company said about SEK5bn ($560m) of its projected sales increase in 2020 would come from anticipated favorable currency movements; and another SEK2bn ($220m) from Red Bee Media, a broadcasting and media services company. The rest would come from higher than expected network sales. Ericsson is now guiding for revenues of SEK141-145bn ($15.7 to $16.2bn), up from an earlier forecast of SEK128-134bn ($14.3 to $4.9bn). The midpoint of that range would represent an increase of nearly 12% on 2017’s network sales.

Ericsson also said it would target sales of SEK41-43bn ($4.6 to $4.8bn) from digital services and SEK23-25bn ($2.6 to $2.8bn) from managed services for the 2020 fiscal year. In fiscal 2017, those units delivered SEK41bn ($4.6bn) and SEK24.5bn ($2.7bn) respectively.

Caroline Gabriel, ReTHINK Research

The Economics of Supply, Demand, and the Monetization of the “Edge”

The Economics of Supply, Demand, and the Monetization of the “Edge”

-5 minute read

In 1776, a Scottish author called Adam Smith published “The Wealth of Nations”, but unfortunately for the British, Smith’s book was released in the same year as the American Revolutionary War began in earnest. If it had been published ten years earlier, Parliament and King George III may have taken to heart Smith’s sentiment that “colonies should be granted representation in the British parliament proportional to their contributions to public revenues”. Had this advice been heeded, there may have been no 1776 and arguably, the America Colonies would now be ruling England. Adam Smith would not simply be a great economist and philosopher, but the founding father of the United States of North America and Great Britain.

Supply, Demand, and the Edge

As part of his thesis written some two hundred and fifty years ago, Smith hypothesized – “Price is regulated by the proportion between the quantity brought to market and the demand of those who are willing to pay.”  This simple statement is as profound today as when it was published and has provided the stimulus for numerous economic movements. Margaret Thatcher and Ronald Reagan, who arguably created perhaps the two strongest economies of the 20th century, were reputed to always carry a copy of Smiths work with them.

But what, if anything, does this have to do with edge cloud and cloud computing?The Economics of Supply, Demand, and the Monetization of the “Edge”

For cloud computing, Smith’s principle is enduring.  If we portrayed a graph with price on the y axis and quantity on the x axis and were to plot where the downward slope of demand (Cloud Market) intersects with the upward slope of supply (Compute Inventory), this point in a free market where all available cloud compute resources could be monetized.

The challenge for public cloud providers is that Smith’s principle assumes that the supply at a given price is limited and the cost of an additional unit of inventory becomes higher as the total supply expands. In most industries this is true. A good example is the auto sector. To deliver more quantity, additional workers are required and that comes at a premium as auto makers have presumably exhausted all the workers willing to function at a lower cost.

In public cloud environments, these economics of industrial production do not apply, or at least not cleanly. The cost to add capacity is somewhat negligible as compute costs become insignificant, and storage and bandwidth costs plummet.  Spinning up shared virtualized resources costs is proving to be quite cost effective compared to traditional IT. It is anticipated that 83% of enterprise workloads will run in public clouds by 2020 and thus it could be argued that public compute inventory is effectively infinite if the price exceeds the cost per unit.

This is a challenge for the public cloud companies is an increase in demand for compute has not increased the price. In fact, over the past 10 years price has generally dropped or, at a minimum, has remained constant at any level of demand. The illustration above outlines the decrease in cost per hour pricing for Linux on-demand cloud services offered by a large public cloud provider[1], while the cost of some services dropped as much as 14% in the year up to August 2018[2]. However, the increased demand observed over this period did not create a supply imbalance, and has thus resulted in a highly commoditized market place.

The Commoditization Conundrum

The Economics of Supply, Demand, and the Monetization of the “Edge”This commoditization has generally been seen as a challenge by many network operators considering edge compute as a way to monetizing their investment in their fixed and 5G access networks. Its left them struggling with a business model that, at first glance, doesn’t appear to make a lot of sense. In many cases the approach has simply been for an operator to turn over their edge to a third party while they continue to pay for power, cooling and real-estate based on the hope that they may profit from transit or metro network savings.

However, the economics of public cloud and edge cloud and compute models are very different and there is a good reason for this. The edge offers a fixed, finite inventory of compute resources. This is unique for edge compute and is the result of creating ultra-small and ultra-close access nodes that are distributed in nature with considerable geographical diversity and relatively small available power and real-estate footprints. This means that the supply of edge compute resources at a given access point remains of capacity exists. This has resulted in pricing that is steadily climbing, even as the number of edge compute applications remains relatively low compared to broad spectrum of enterprise applications – a further indication of the effect that a constrained inventory supply has on value.

Clearly every application does not require edge compute. Public cloud & traditional COLO are still a viable locations to host enterprise applications, while there will always be requirements for specific real-time functionality to reside on the “device” itself – autonomous vehicles are a great example. However, for low latency, interactively-intense applications the edge is a hard requirement and as such will command premium pricing. This scarcity of edge compute inventory limits surplus stock and establishes a constrained supply that can be unquestionably monetized, thus the average cost of edge compute will increase significantly and enable the Network Operators to extract fair value from their investment.

So as Adam Smith declared some 250 years ago, when things are scarce and in demand, people have proven to be prepared to pay more for them and there is more profit in supplying them so network operators can feel justified in investing more capital to produce them. Where there is a glut, things are commoditized and prices and profits are low, producers switch their capital and enterprise elsewhere. Unlike the public cloud, the edge inherently rations inventory and drives pricing up automatically and without the need for central direction.

The new “Machine Economy” is already stimulating the deployment and orchestration of edge dependent applications.  While media and entertainment workloads, high resolution cloud gaming and bandwidth efficient VR video delivery may seem like obvious edge use cases for an operator, there are a gamut of emerging enterprise applications that have the potential to be the “new oil”. Autonomous driving experiences, haptics and tele-operation, vRAN, SD-WAN, IoT and smart-cities all benefit from, and more importantly, help the operator to monetize edge and access infrastructure investment.

What Can Be Done?

So what is the call to action?  We believe there is scarcity and high value associated with the Operator edge network and that, if combined into a global network, will deliver exponential value and revenue. Operators also cannot wait for 100 applications to become popular before deciding whether or not to experiment. Edge is a $5Bn+ business and provides numerous opportunities to start trialing new business and deployment models, thus the time to learn is “now”. It all starts by installing some resources in 1 location. Its low cost, high impact and will help start seeding the edge.

[1] Verified by the Internet Archive
[2] Right Scale: Comparing Cloud VM Types and Prices

Ericsson Takes Startup Approach With 'EDGE GRAVITY'

Ericsson Takes Startup Approach With ‘EDGE GRAVITY’

This article originally appeared on Light Reading (November 14, 2018).

Without much fanfare, Ericsson has launched a new unit at the company, called EDGE GRAVITY, that operates a global edge cloud network that links together a core network of data centers with the last-mile networks of more than 80 partners that include cable operators, telcos and mobile service providers.

EDGE GRAVITY, unveiled in part via this Twitter post on November 9, is essentially the name of a new company within Ericsson AB (Nasdaq: ERIC) that is tied to company's relatively new Unified Delivery Network (UDN) initiative. By putting compute at the edge — and inside the network of the ISPs themselves — EDGE GRAVITY has initially focused on expanding the reach of certain content delivery networks (CDNs) and supporting an array of latency-sensitive services and apps (more on that later). (See Ericsson Wants to Add Compute to CDNs and Ericsson Poised to Disrupt CDN Market.)

EDGE GRAVITY is part of Ericsson's Technology & Emerging Business unit, which focuses on new parts of the business — referred to as “accelerated units” — that Ericsson is trying to scale up in part by giving them a large degree of autonomy and startup-like agility. EDGE GRAVITY, for example, has its own sales and marketing teams and IT system/infrastructure.

EDGE GRAVITY is the first unit there that is using this new model, according to Yves Boudreau, Ericsson EDGE GRAVITY's chief marketing officer.

EDGE GRAVITY, which has been working on the project for more than a year, started by building a core network, on its own dime and using its servers, in Equinix Inc. (Nasdaq: EQIX) data centers and on its own MPLS backbone.

That core network, with 22 locations at present, has been connected to an edge network comprised of more than 80 last-mile network providers that EDGE GRAVITY has managed to sign on so far. Under that model, the service providers provide EDGE GRAVITY with elements such as rack space, power and connectivity.

That's a turn of the table of sorts. “Rather than selling to the service providers, we are making them suppliers to us,” said Boudreau, who also leads the partnership and ecosystem strategy at the unit. EDGE GRAVITY then works on a revenue share based on the traffic it brings to the network.

A representative list of those service provider partners include Rogers Communications Inc. (Toronto: RCI), Telstra Corp. Ltd. (ASX: TLS; NZK: TLS), Bharti Airtel Ltd. (Mumbai: BHARTIARTL), Singapore Telecommunications Ltd. (SingTel) (OTC: SGTJY), Telefónica , NTT DoCoMo Inc. (NYSE: DCM), China Unicom Ltd. (NYSE: CHU), Chunghwa Telecom Co. Ltd. (NYSE: CHT), Telkom Indonesia and Mobiphone. Boudreau, who presented on EDGE GRAVITY in mid-October at the EdgeNext Summit in New York, noted then that the service provider part of the edge cloud network is about half deployed.

EDGE GRAVITY then interweaves its network with a software stack that is open source “for the most part” using OpenStack and Kubernetes, Boudreau said.

Creating a global edge
Boudreau said EDGE GRAVITY idea came about as service providers struggled to exit their own countries and extend the reach of services while watching the likes of Amazon Web Services Inc. , Microsoft Azure and Google (Nasdaq: GOOG) dent their revenues.

“They can't act globally as a web-scale company can,” he said. “This is our attempt to help our customers and the service providers come to the table collectively.”

Boudreau stressed that EDGE GRAVITY is not out to replace public cloud companies but rather to complement them and take advantage of a web-scale deployment model that allows customers to write software that will operate on EDGE GRAVITY's entire edge cloud network.

Boudreau estimates that EDGE GRAVITY's combined network has about 4 Tbit/s of capacity and that the number will rise rapidly as more partners are brought on board. That approach, he said, will enable the network to scale as data demands increase in a way that individual networks can't on their own.

“The only way we can keep up with that pace of traffic is really by doing it in partnership with operators,” he said.

Early use cases
Though deployment is still ongoing, EDGE GRAVITY has already signed on several partners such as Limelight Networks Inc. (Nasdaq: LLNW) and Net Insight AB (Stockholm: NETI-B), and startups such as Haste and Mode. (See Limelight Connects to Ericsson's Edge Cloud .)

Limelight is tapping into the network to expand its reach and tack on more capacity without handing to standup more infrastructure. The CDN angle was the “first and easiest” application for the new edge network, Boudreau said.

Mode, a startup in San Francisco, has developed software that accelerates the decision-making process for routing for services — such as collaboration software — that have stringent latency demands. The idea here is that Mode can boost the performance of its software by touching packets coming out of the operator network itself.

Haste, meanwhile, has built a low-latency service for gamers (with an eye toward other real-time apps) that it's also pitching to ISPs as a potential driver of new revenues.

Haste is now running in EDGE GRAVITY's nodes. While Haste also requires users to run the company's client on their gaming machine, the company is exploring ways to embed Haste into the network itself, company co-founder Adam Toll said at the same EdgeNext Summit event. (See Haste Sounds Out ISPs on Low-Latency Service for Gamers .)

In addition to getting current partners rolling, another priority for EDGE GRAVITY is to prepare to onboard another batch of apps and services that are optimized for this type of edge network.

Regarding cases of pushback, Boudreau notes that some companies argue that they are already doing this on their own, but on their own network. “But nobody wants to be on just one network. They want to be on all the networks that they can,” he counters.

Others, particularly smaller service providers, claim that they don't have the resources to pull this off. But that fits into a primary purpose of EDGE GRAVITY, he said, in that the company is in position to handle the heavy lifting that those providers can't handle on their own.

— Jeff Baumgartner, Senior Editor, Light Reading




Real Edge Computing