White paper Expert Community Edge computing
Thought Leadership Executive Summary Tables of contents
This whitepaper provides an overview of 04 Edge computing: definition, motivation and drivers edge computing technology, use cases and Motivation existing market developments. It elaborates on What is edge computing? analysis of the existing market status building a classification which entails consumer, software, Use cases hardware and born at edge categories. In 07 Edge market status addition, it relates to on-going initiatives and opportunities for further growth being explored The consumer edge by Atos, including the introduction of the Atos The software edge Edge Server. The hardware edge Born at edge
10 Challenges for adoption of edge
10 Atos developments and opportunities
12 Conclusion and call to action
02 Expert Community: edge computing Expert Community: edge computing 03 Edge computing: definition, motivation and drivers
Motivation What is edge computing?
The unstoppable development of Internet upstream business systems – or even in communication, make it impossible to achieve Edge computing locates computing and • Cloud services: they develop management a number of important specific characteristics of Things is expected to exponentially grow other platforms that can make use of the the necessary response times required to storage resources at the edge of the functionalities for both edge and IoT devices to edge computing9. Among them: the number of connected devices worldwide. data out of their original context (e.g. a car trigger and execute the actions in closed network, with the intention of “getting data and they perform long-term data storage and • Wide physical distribution: in contrast According to the latest estimates, these insurer using in-vehicle data to support a loop environments. and computation at the right place and right analytics. Additionally, they provide the point to centralized cloud, the services and could reach more than 20 billion by 20201. pay-as-you drive offering). time in a more decentralized manner”4 and for integration with other enterprise systems. applications in edge computing are In this context, the challenge is not only the As part of addressing this issue, there is the to elude non-essential data transmissions In some advanced scenarios, such us decentralized and distributed by nature. management of these things on and across The number of device connections, the volume pressing need to move information close over the network. Gartner defines edge autonomous vehicles, the limits among • Need for near real-time interactions: edge the various platforms, but also, how to cope of data, the latency across different locations to the source, transferring analytical model computing as “solutions that facilitate data these three layers blur, the car being a computing aims to enable IoT actuation with the volume of data and content that and networks, together with the asynchronous execution near to the sources of data, processing at or near the source of data rich environment of IoT devices and edge scenarios by bringing computation close to all these connected things will produce. nature of many connections between data so to provide compute capability inside generation”. At time of writing there still capacity all integrated in a single device. data sources. In doing so, there is a strong Researchers estimate that in the last two flow and analytical cloud services, make an an environment where connectivity and remains in the market significant confusion requirement for immediate response times. The recognized advantages of edge years we have created 90% of all available existing IoT approach unsustainable in the long response times can be controlled. with regards to what is Edge computing • Autonomy: edge computing installations computing are: data on the internet, and this is expected to term. Besides, today’s heterogeneous networks and specifically which are the features and are not subject to data center boundaries To avoid sending all data to be grow at 40% pace in the coming years due are not yet ready for managing the expected To address the above points a new class capabilities it offers. Diverse approaches • Privacy: and well-known fault tolerance practices. stored and processed on cloud servers. to the explosion of the number of available incredible growth in the number of end-points of edge connectivity and computing has exist, coined under diverse terms: Edge Therefore, they are prone to circumstances • Connectivity: To avoid and reduce costs connected devices2,3. and the volume of data to be transferred, for emerged. It deals with network and local device computing5 is widely used; but also Fog that can affect their availability and associated with streaming/sending all raw which smart ways to distribute the data are not management, pre-processing of the message Computing6, as named by Cisco; and connectivity. data to cloud services. Following the pattern we have today in the yet available. streams (e.g. data can be pre-processed, Multi-Access (Mobile) Edge Computing7 , • Heterogeneity. Heterogeneity is a key To improve existing IoT set-ups, generated sensor data aggregated or filtered and distributed) to focusing in RAN (Radio Access Network) • Latency and reliability: characteristic of Edge computing which is responsiveness and reliability by would be transmitted over the wide area Apart from the previous considerations, ensure that the subscribers of a specific data technological aspects. It is important to note present both in the variety of devices which maximizing processing at the edge and network in order to be centralized, processed production critical environments need yet to stream only receive the data that is relevant for that this whitepaper we take the approach form and connect to Edge computing thus minimizing dependence on cloud and analyzed - creating additional volume address requirements for immediate response them, and that even the execution of analytical to Edge Computing as defined by the Linux instances, but also in the diversity of network connection. of (enriched) data. This data, in addition to times. Nowadays, in many environments, models and resulting actions is produced Foundation Open Edge Computing Glossary8 and security protocols they are exposed to. analytical models, are supposed to trigger the stability of the connection to centralized swiftly and in a distributed model. which defines the Edge Cloud as “Cloud-like Edge and cloud manage processing, • Interoperability and federation. Typically, actions either on the thing itself or in the systems, as well as the latency incurred in the capabilities located at the infrastructure edge, networking and storage resources. In this an Edge installation works together with including from the user perspective access regard, they share essential mechanisms a federated Cloud offering. For this, both to elastically-allocated compute, data storage and attributes (for instance virtualization). Edge and Cloud environments inter- and network resources. Often operated as a Nevertheless, edge computing specifically operate for operations such as Edge device seamless extension of a centralized public or focuses on addressing latency issues management or long-term data storage. private cloud, constructed from micro data detected in large IoT scenarios and this brings centers deployed at the infrastructure edge.”
A common topology for an edge computing Figure 1: Edge computing layered stack installation is composed of three-layers. From bottom to top these are: - Management Cloud • IoT devices: IoT devices are connected to - Long term data services store & analytics an edge device. IoT devices communicate Wide Interoper via diverse communication protocols Physical Autonomy ability and with the edge environment acting as data distribution federation sources. Edge - Near real-time data • Edge nodes: they enable data processing processing nodes - Temporal data close to sources of data through near real- storage time data analytics and model execution. It offers divers communication and near Heterog messaging protocols for data acquisition real-time eneity - Messaging interaction from near-by IoT devices and acts as - Data temporal data storage. - Actuation IoT devices
1. Gartner, Leading the IoT, https://www.gartner.com/imagesrv/books/iot/iotEbook_digital.pdf 4. Bonomi, F., Milito, R., Natarajan, P., & Zhu, J. (2014). Fog computing: A platform for internet of things and analytics. In Big data and internet of things: A roadmap for smart 2. Big Data, for better or worse: 90% of world’s data generated over last two years, http://www.sciencedaily.com/releases/2013/05/130522085217.htm environments (pp. 169-186). Springer, Cham. 3. IDC, The digital Universe of Opportunities: Rich Data and the Increasing http://www.emc.com/leadership/digital-universe/2014iview/executive-summary.htm 5. FL Edge, https://www.lfedge.org/ 6. https://www.openfogconsortium.org/ 7. Multi-Access Edge Computing, https://www.etsi.org/technologies/multi-access-edge-computing 8. https://github.com/lf-edge/glossary 9. YM. Yannuzzi, R. Milito, R. Serral-Gracià, D. Montero and M. Nemirovsky, «Key ingredients in an IoT recipe: Fog Computing, Cloud computing, and more Fog Computing,» 2014 IEEE 19th International Workshop on Computer Aided Modeling and Design of Communication Links and Networks (CAMAD), Athens, 2014, pp. 325-329.doi: 10.1109/ CAMAD.2014.7033259 04 Expert Community: edge computing Expert Community: edge computing 05 Edge Market Status
Use cases Edge computing enabling technologies are a major opportunity for existing IT stakeholders and new emergent ones. Gartner estimates that, “Currently, around Edge use cases are mainly related to those • Asset tracking and monitoring (including These use cases are sometimes mixed where real-time (or almost real-time) reaction predictive maintenance): this has been to create business solutions that fill all 10% of enterprise-generated data is created and processed outside a traditional is needed and the latency to move to a focus area so far, aligned with Industry requirements that customers may have in centralized data center or cloud. By 2022, Gartner predicts this figure will reach 50%.” central cloud services does not ensure the 4.0 initiatives. The fact that machines today’s complex scenarios. As an example, necessary velocity on reactions. So, we may equipped with sensors can be monitored we may think of healthcare and, in a more In terms of market size, analysts measure the market growth as, “6.72 USD billion by identify as key use cases the following: in real-time. Together with the amount of concrete manner, a hospital. We may have data available to use with ML algorithms to the asset tracking scenario to check the 2022 at a Compound Annual Growth Rate of 35.4%”. • Autonomous driving: Tesla is one learn from patterns and behaviors, allow health and status of all equipment but also example of using computing in the for real-time monitoring to prevent failures receive sensor information from patients edge but also centralized ML practices and even, being able to forecast them and have some decisions being taken on The edge market is still in its infancy, • Born at edge: while all the previous to improve the overall behavior of therefore reducing the cost of down time, the edge to improve customer care. All however its complexity is starting to be categories refer to existing vendors Figure 1: Edge computing edge devices. We see examples in all waiting for spare parts and repairs. this increases the resilience of the overall recognized. The edge computing market who are - in different ways - extending layered stack transportation subsectors (cars, trucks, structure having all actions and resources • Employee, customer and citizen will allow diverse market actors to have their or adjusting their existing products ships, drones, etc). being used at the proper time and location say in providing edge computing products or services to cover the edge market monitoring: the use of smartphones, depending on priority, latency and availability. • Process monitoring and operations wearables and tagging devices with and services. These range from system spectrum, here we refer to specialized Consumer analytics: the monitoring of business their related beacons (Bluetooth, NFC) integrators, major cloud providers, hardware edge computing vendors whose offerings edge process and operations through the and an increasing number of image and software vendors, telecom operators only focus on this market. extensive use of sensors and information recognition systems are putting the and more. • Hardware edge: in this category we systems are making inroads to increase monitoring of persons (and reacting to present actual hardware device vendors Software In this scenario, different players and edge efficiency of operations. For instance, that) center stage of new edge computing strategies come into play. With the objective who are including, as part of their we may think of refrigeration checks applications. Examples include using to provide an overview of existing market products, the development or packaging Edge for food security and safety, where person/Face Recognition Systems, Human developments we have structured our analysis of software tools that enable edge Market conditions are verified and warnings Behavior analysis (through image and of the edge market in the following categories: workload management. Atos is classified issued (or actions taken) if values fall below location analysis), customer experience under this category. • Consumer edge: represents vendors in specific targets. Systems can be end user monitoring (and influencing though Born at the field of consumer electronics who are For some of these categories we will edge devices (personal fridges) or industrial targeted advertising and notifications). likely to address the opportunities that be presenting examples of vendors to ones. Actions may range from a request Keeping in mind privacy concerns, there edge computing brings by embedding (or show how the edge opportunity is being for additional stock when empty or for are also a wealth of uses around safety Hardware allowing to interact) edge intelligence in addressed. Note that this analysis will support assistance if values point to errors. and compliance that will be a major push edge their devices. present information available at time of in this area. writing (July 2018) and does not aim to be • Software edge: these are vendors, mainly an exhaustive market guide instead we major cloud providers, who offer software aim to provide exemplary cases for the platforms to be used in conjunction with established categorization. their IoT cloud services and installed in edge devices provided by the user.
10. Gartner Blog, “What edge Computing Means for Infrastructure and Operations Leaders”, https://www.gartner.com/smarterwithgartner/what-edge-computing-means- for-infrastructure-and-operations-leaders/ 11. Markets and Markets, “Edge Computing Market worth 6.72 Billion USD by 2022”, https://www.marketsandmarkets.com/PressReleases/edge-computing.asp 12. Gartner, “Edge Computing Challenges Go-to-Market Strategies in IoT“, https://www.gartner.com/doc/3877186/edge-computing-challenges-gotomarket-strategies
06 Expert Community: edge computing Expert Community: edge computing 07 The consumer edge The software edge The hardware edge Born at edge
Edge can scale from personal devices Major cloud vendors such as Amazon Web significant effort. Whether using Azure’s While vendors in this category have The Raspberry Pi could be said to have The development of the edge computing (i.e Fitbit) to high performance compute in Services and Azure are taking the edge IoT edge and IoT Hub or AWS Greengrass, traditionally focused on devices and started (or perhaps re-started) a computer market is also bringing the emergence of wind turbines and complex manufacturing computing opportunity to extend their cloud you can now allow edge distributed logic to equipment, increasingly they are trying to revolution, bringing a new generation of specialized providers that offer targeted (i.e Siemens). It can include B2B and B2C and services and to provide an additional entry be written in the same format as what can embrace both hardware and software stacks, users to ‘home-brew’ computer. The low products and solutions. even C2C scenarios. Edge has the potential to point for their consumption. run centrally, easily allowing a centralized so as to build overall edge solutions as part cost, low power and commodity nature of add value to almost all devices, but success model to move to an edge deployment at of their value proposition. Examples of this the Pi has made a ideal starting point for For instance, FogHorn products specifically is about more than internet connectivity and Azure IoT edge15 is available as Open Source a later stage. Serverless code, whether it approach are Cisco and Dell. producing commodity IoT edge devices. tackle the Industrial IoT market and centers capturing data. The point is that the edge platform16. It can execute both in Windows be Azure Functions or AWS Lambda, allows Both Azure and AWS include guides via its offering in its Complex Event Processing device allows capture and analysis of data and Linux powered devices. IoT edge local decisions to be taken rapidly and even Cisco’s was a precursor in identifying the value Github on how to use these commodity technology which is described as the “world’s with the goal of providing insight. modules run as Docker containers. Its runtime insight and analytics of data, with low latency, for computing at the edge, coined under the devices and integrate them in to your IoT most compact, advanced and feature rich”22. provides monitoring and workload execution even in scenarios where internet access Fog Computing term19. Cisco sees the need for environment. Obviously, such a commodity This technology is provided in two different Even now embedding intelligence into devices functionalities at the edge. It allows data pre- might be down or intermittent, while at the a handling the volume, variety, and velocity of device is not for every circumstance but platform packagings, in standard and micro is generally missing. Only a few market processing on-premises before sending it to same time allowing aggregation of data back IoT data in a new computing model. Recently, has helped encourage a rapid growth editions. Both editions include remote edge developments are starting to tackle these Azure Cloud services. Azure services, which can to the central cloud for global scale reporting. of experimentation at the edge. Beyond flee devices management including device and opportunities. General Electric is embedding run on edge devices include Azure Machine Cisco has launched Cisco Industrial Compute this initial experimentation with Edge platform monitoring, configurability, as well as, edge computing into industrial internet Learning, Stream Analytics Azure Functions, AI The ability to train a ML model rapidly in the Gateway IC300020. This is equipped computing, progressively we observe in IoT automated deployment of custom, platform and and providing an edge to cloud distributed services and the Azure IoT Hub. cloud and then roll that model out locally to with its Kinetic edge and Fog Processing applications not only the need of having analytics software. Differences among these intelligence through its Predix platform13. thousands of distributed locations empowers Module enabling a software stack to rapid data processing close to the sources, are in their deployment memory requirements, Another recent example is provided by AWS Greengrass17,18 software stack is the edge. Models that even a few years ago distribute computing. In addition, the Cisco but the crucial requirement to provide less than 256MB devices for micro edition and Amazon, that is reported to be building specific available for both ARM and x86 devices with would seem impossible to run on the edge IOx application framework offers security secure, intelligent monitoring and even, 2G for standard edition. More in detail, Micro AI chips for its Alexa platform14. minimum required capacity. It also takes into suddenly become feasible on small and low capabilities as well as consistent management handling of the physical environment so to edition provides a C++ SDK for application account communication and management power hardware, both in generic processors, and hosting across network infrastructure enable next generation, transformative AI development, data pre-processing, analytics, A very significant field of development for the for micro-controllers (Amazon FreeRTOS). but also potentially in ASIC based hardware. products, including Cisco routers, switches, and and IoT applications. machine learning and visualization capabilities. future are autonomous vehicles for which Atos AWS Greengrass is a Serverless computing compute modules. Features available in standard edition include is developing crucial technology in the context platforms-based AWS Lambda which also Taking it a step further, Azure Stack brings Therefore, next generation Edge computing micro-edition’s features with extended support of the European Processor Initiative11. provides: data messaging and edge to cloud another potential to run considerably higher Dell edge Gateway Series21 is purposefully environments need to cope with the emerging for Industrial protocols, data access protocols, sync via interoperability with AWS IoT Cloud power applications, bringing the power of built for developing industrial automation. necessity of offering AI inference and cloud interoperability and local data persistence. services. Similarly to Azure, it has recently Azure Cloud in to your own data center with These Gateway Series offer specific IoT sophisticated data analytics at the Edge, which Cloud technologies or providers offering further developed ML inference capabilities hundreds of cores and terabytes of memory hardware equipped with Intel base Atom encompass steam video, image and audio compatibility with this solution are not specified. in Greengrass ML. These ML models, which for local delivery of Azure native services, processor enabling integration with diverse near real time processing. FogHorn presents as a main novelty of its have been developed and trained in the whether it be for latency, geographic availability protocols (ModBus, BACnet and ZigBee) products edgeML technology, which aims to cloud, can be deployed and executed locally or data sensitivity purposes. The key is the as well as, network access means (Wi-Fi, enable execution of machine learning models in in the edge device. same models from public Azure have the WWAN and Ethernet). edge constrained devices. potential not only to be moved to smaller In either case, the major cloud players bring lower power hardware, but also to distributed Increasingly commodity cheap hardware such Another example is the French company Tell the opportunity to have a unified ecosystem, scenarios offering huge levels of local as Raspberry Pis and off the shelf x86 servers Me Plus. This company has been chosen as but potentially locked in to that vendor. processing power. are a clear platform of choice to provide edge part of Microsoft’s “Accelerator Program”, where Running the software as containers on Gateway features, playing a significant role Microsoft looks to equip participants with tools top of Linux allows easy deployment, but due to affordability and huge scale, which is to help them develop their product and grow the code written may well end up locking achievable depending on the requirements. their organization. Similarly to FogHorn, Tell you in to that vendor longer term without Me Plus software platform, predictive Objects, targets the Industrial IoT sector.
Other novel general-purpose edge platform vendors are Crosser and Rigado.
Besides, we also start to encounter niche offerings which are addressing specific needs of edge solutions. Examples of these are: IoTium which focus on edge security aspects; and Wirepas that addresses the needs of IoT networking at scale.
13. GE, Predix Platform, https://www.ge.com/digital/predix-platform-foundation-digital-industrial-applications 19. Amazon Web Services Greengrass FAQs, https://aws.amazon.com/greengrass/faqs/ 14. The Verge, “Amazon is reportedly following Apple and Google by designing custom AI chips for Alexa”, https://www.theverge.com/2018/2/12/17004734/amazon-custom- 20. CISCO, “Fog Computing and the Internet of Things: Extend the Cloud to Where the Things Are”, https://www.cisco.com/c/dam/en_us/solutions/trends/iot/docs/ alexa-echo-ai-chips-smart-speaker computing-overview.pdf 15. European Commission, “European Processor Initiative: consortium to develop Europe’s microprocessors for future supercomputers”, https://ec.europa.eu/digital-single- 21. CISCO, “Helping Enterprises Scale IoT Deployments with Secure Computing at the edge”, https://blogs.cisco.com/digital/helping-enterprises-scale-iot-deployments-with- market/en/news/european-processor-initiative-consortium-develop-europes-microprocessors-future-supercomputers 16. secure-computing-at-the-edge Azure IoT edge, https://azure.microsoft.com/en-us/services/iot-edge/ 22. 17. Azure IoT GitHub. https://github.com/Azure/iot-edge/ Dell edge Gateways for IoT, https://www.dell.com/en-us/work/shop/gateways-embedded-computing/sf/edge-gateway 18. Amazon Web Services Greengrass, https://aws.amazon.com/greengrass/ 23. FogHorn products, https://www.foghorn.io/products
08 Expert Community: edge computing Expert Community: edge computing 09 Challenges for adoption of edge Conclusion and call to action
While the edge market is rapidly developing and benefits of edge computing relate Atos is in a privileged position in order to build a compelling edge computing offering to improved applications performance derived from reduced latency in data which takes advantage of the uniqueness and diversity of Atos capabilities. processing, it comes with a series of challenges related to its adoption:
The first relates to the novelty of the concept computing completes cloud computing, is not yet a clear understanding of required Closely related to the definition of its IoT term, these providers won’t be able to adapt At the same time, the novel market actors and lack of awareness of the benefits. While offering an alternative solution for which standardization efforts and vendor adoption in strategy, edge computing offering can be a these to the specific needs of their software are focusing on the industrial internet arena the edge technology is still under development existing services in the cloud are not sufficient, the market. key development for the coming years for platforms. In addition, it is important to note, which nowadays concentrates the majority and not fully consolidated in the market, users the interrelations and interoperability among Atos by agglutinating and exploiting a series these providers are offering their customers of edge commercial efforts. The list of niche have difficulty in getting concrete figures both approaches still appear uncertain to It is worth mentioning that edge technology of capabilities that makes it exceptional in a not negligible vendor lock-in (even at products available today is significantly and information on performance gains and potential consumers of this technology. also brings a number of important challenges the market: ability to deliver its own edge programming model). limited talking solely security and networking. applicability to its concrete usage scenarios. not only to users to adopt technology but to compute servers cutting-edge IoT security Despite the fact that some initiatives such the commercial providers of these offerings assets; capacity to build a software platform Hardware providers which are strong and From this analysis it becomes clear that In relation to this, there still remains confusion as OpenFog, ETSI and edge Computing such as service management in geographically with robust analytics and machine learning capable in developing and deploying devices there is not yet a provider able to offer in some instances about the complementary consortium start to emerge in order to offer distributed environments outside the data inference features and interoperability with at scale, are trying to build software platforms an edge computing end to end solution or substitutivity between edge and cloud. both certification and standardization schemes center limits. own and major cloud providers services (as from scratch, and are not yet embracing comparable to Atos’ potential in the field. Whereas common perception is that edge for this new technological environment there Atos cognitive analytics evolution); as well possibilities of building partnerships with Atos is able to build on top of horizontal as, detailed sector specific knowledge and cloud vendors for this purpose. Therefore, to assets (hardware platforms, software consultancy skills to bring this novel offering some extent, they are lagging behind at the security, analytics and interoperable cloud to the market. level of the software capabilities they offer software platforms) specific customizations to customers having to build applications. to vertical sectors and IoT platforms which Atos developments and opportunities It can be observed from current market Although the use of commodity cheap could, complemented with domain specific developments detailed in Section 2, that hardware is increasingly becoming popular, knowledge and consultancy services, major cloud providers are focusing on it has not yet had a reference Open Source facilitate adoption paths for customers of this Atos, via it’s cognitive analytics offerings, is already a significant player in the IoT provision of software platforms, relying on a edge computing project that delivers future crucial technology. network of hardware providers on which to a software stack which helps to build a marketplace. The Atos cognitive analytics tagline is «Turning Data into Business make their offerings compatible. In the long complete solution beyond pilot execution. Outcome», which implies that we are supporting the full loop from the Edge device to collecting data from the device to trigger actions either in the device itself or in the associated business systems.
Edge computing plays an important role in itself is analyzed on the edge device to provide announced as «European Processor any of these setups : it provides the means to fast response. The opportunity to build and Initiative with Deep Learning acceleration» Authors connect the devices to manage the (in some integrate towards such scenario requires (EPI) for future availability. Head of Next Generation Cloud Lab at Atos Research and Innovation, Atos Distinguished Expert in Cloud Domain and member of cases bi-directional) data flow and associated Atos capabilities across the board – IT/OT Ana Juan Ferrer At the level of software stack, Atos Research the Atos Scientific Community actions, as well as, the volumes of data that connectivity, Integration of shop-floor and and Innovation is participating in a set of are produced. top-floor data, analytics on both the central Adam Dolman Head of Public Cloud for Atos Global IDM ESO, a member of the Atos Scientific Community, and an Atos Expert in the Cloud Domain related H2020 initiatives taking key aspects systems as well as on the edge and creation Akshay Bagewadi Cloud Expert & System Consultant Lead at Atos More strategically, Atos will launch its first of future edge computing developments: of micro-service or apps on the edge itself to Edge server, BullSequana Edge. AGILE project24, which builds an IoT gateway Alex Caballero CTO IDM Iberia, Atos Senior Expert in Cloud Domain, and member of Atos Scientific Community provide decision support or to trigger actions. that offers data and workload portability with Atos is one of the few companies that can Claude Derue HPC & Big Data Marketing, BDS Global From a connectivity and data management key public cloud offerings; DITAS project25, provide edge computing solutions at this level. Darren Ratcliffe Distinguished Expert & Cloud Domain lead, Head of Technical Strategy within Business & Platform solutions side, Atos own edge solutions («cognitive that develops and implements the concept The prototype description highlights the ability analytics Connectivity Platform» (CCP), «Cloud of Virtual Data Containers and specifies data Gary Burt Atos Distinguished Expert, Cloud Product Manager and Hybrid Cloud Business Strategist to provide AI computing power at the edge Industrial Supervision» (CIS)) help us build movement approaches across edge and with rich security features. Moving up the stack Gauvain Girault CTO Big Data, BDS Global platforms to connect devices (like sensors cloud environments; and mF2C project26, - using edge devices that provide connectivity Gert Prieber BullSequana Edge Product Manager, BDS Global on the shop floor, vehicles, power valves) which focuses on providing across edge and and device for their compute capability opens and help distributing compute and analytical cloud workload orchestration techniques. Klaus Ottradovetz Atos Distinguished Expert Cloud & IoT platforms, CTO office on cloud and IoT platforms and technologies the door to move analytical execution close tasks. These are software solutions built on to the devices. Analytical models are being Prashant Chamarty Expert Sales at IoT and Cloud Centre of Excellence, Atos Senior Expert in Cloud Domain top of hardware, assuming end-points are developed (often with the help of Machine available. CIS is an Atos cognitive analytics IoT Ripal Kaila Sr. Expert Application domain, Sr. Solution Architect Cloud, SAP HANA Learning and Artificial Intelligence) on the component that actually helps to control the centralized platform and the (sub-) models Roy White Global MS Digital Cloud, Infrastructure & Applications flow in such distributed scenarios. are distributed to be executed on the edge Sameep Bhate Solution Architect device so to either reduce the volume of data With regards to long term development transmitted or to trigger automated actions. of edge computing, Atos is working with In complex business scenarios – e.g. remote its Bull subsidiary on edge devices that factories relying on Manufacturing Enterprise provide massive compute and analytical Systems (MES) data – edge devices will start capabilities (incl. Machine Learning caching external data from the MES system, and Artificial Intelligence) - this is being use real-time data from the shop floor which
24. Agile Project, http://agile-iot.eu/ 25. DITAS project, https://www.ditas-project.eu/ 26. mF2C Project, http://mF2c-project.eu
10 Expert Community: edge computing Expert Community: edge computing 11 CT_190425_MA_ExpertCommunity-EdgeComputing_WP About Atos Brochure information owned by Atos, to beusedby therecipient only. This trademarks oftheAtos group. April2019. ©2019Atos. Confidential Atos, theAtos logo, Atos Syntel, Unify, andWorldline are registered For more information: [email protected] Let’s adiscussiontogether start atos.net/career atos.net Find outmore aboutus Paris index. stock CAC40 the Atos Worldline. and Syntel, is listed on Unify Atos, Atos brands the under operates and Games &Paralympic Olympic for the Partner Technology Worldwide Information is the Group The sectors. all across business clients its of transformation digital the supports Atos knowledge, industry and technologies cutting-edge its industry. With payment the in leader European the Worldline, through Factory, services as well as transactional Transformation Digital its through solutions Workplace Digital and Applications Business Data, Big Cloud, Hybrid Orchestrated end Group provides the end-to- Computing, High-Performance and Cybersecurity in Cloud, one number European billion. revenue annual and over of €12 countries 73 employeestransformation in with 120,000 in digital Atos leader is aglobal approval from Atos. circulated and/or distributed nor quoted without prior written may notbereproduced, ofit, or any part document, copied,