IEEE 11th International Conference - COMSNETS 2019 - Last MilesTalk - Jan 2019. This talk is for Beginner or intermediate levels only. Kubernetes and related edge platforms are discussed.
3. 3
Edge computing is a method of optimizing cloud computing systems by
performing data processing at the edge of the
network, near the source of the data.
4. 4
Key Requirements
1. Maximize the computing at Edge
- Real time Scheduling
- Distributed Application Execution
- Resource Optimization and Compute Efficiency
- Efficient Orchestration, Monitoring
- Resource utilization across the cluster
- Efficient Runtime Support (Container, LWC, Serverless)
- Low Latency
2. Offline Scenarios and Communication
- Edge Node/Cluster Offline Working
- Vendor Agnostic Cloud Interface
- East – West Communication
- Reverse Proxy, Address Resolution, Routing
- Workload-Workload, Device to Workload Commn
3. Security & Privacy
- Workload to Workload Secure Communication
- Device Identity and Authorization
- Node level identity
- Private Data Isolation
4. Scalability – Platform and Clusters
- Edge Cloud - Clusters
- Microservice based core platform
- Platform extensions and plugins
5. Device Life Cycle & Management
- Device, Node, Application Provisioning
- Repositories and Registry
(Device, Mapper, Node, Workload)
- Discovery (Device, Node, Application/Service)
6. Data & Data Analytics
- Data Storage, Sharing, Distributed
- Distributed and customizable Data Analytics
- AI/ML, Big Data, Streaming Data processing…
7. Remote Management & Visualization
- Consolidated and Efficient Dashboard
(nodes, devices, workloads, resources so on)
- Dashboard at North and South
- Upgrade, Rollback, Reset, Enable/Disable
8. Efficient Energy Management
- Energy aware workload scheduler
- Energy optimized Devices/Nodes
- Energy Monitoring
5. 5
Edge Node 1
Edge Node n
D1 Dn
D1 Dn
Edge Node
Edge Node 1
D1 Dn
D1 Dn
Central Cloud
Edge Node n
D1 Dn
Deployment Simplified…
6. 6
Some Reference to State of the Edge Computing…
Edge computing: The state of the next IT transformation:
The cloud is going to edge in the coming decade
https://www.zdnet.com/article/edge-computing-the-state-of-the-next-it-transformation/
Hype Cycle for Cloud Computing, 2018:
Edge is sitting at top; According to Gartner, for example, 80 percent of enterprises will have shut down their
traditional data centre by 2025
https://www.gartner.com/doc/3884671/hype-cycle-cloud-computing-
State of the edge 2018:
Infrastructure Edge at Operator side & Device Edge at last miles; Acceleratoirs like Smart NICs and GPUs are
central role to play; great opportunity lying ahead.
https://www.stateoftheedge.com/ [Detailed Report Available along with Edge Landscape]
Edge Market 451 Research:
The global Fog market opportunity has the potential to reach more than $18bn worldwide by 2022; Initially,
hardware will have the largest percentage of overall Fog revenue (51.6%), followed by Fog applications (19.9%)
and then services (15.7%). Over time, we forecast that standardization plus the incorporation of Fog functions ‘as
a feature’ into existing hardware driving this hardware component down to 42% of the total market, with the
primary ‘upside’ benefactors being Fog applications and services.
https://www.openfogconsortium.org/wp-content/uploads/451-Research-report-on-5-year-Market-Sizing-of-Fog-
Oct-2017.pdf
8. 8
Open Source edge platforms
A Linux foundation common open platform for IoT edge computing (MEC) – April 2017. Third party central control, System management capability
through API, Device service SDK, start in less than 10 sec, user interfaces, binary data support, automated performance testing, exporting application
services to support multiple end points. Supported by AMD, Dell, Samsung.
https://www.edgexfoundry.org/
Linux foundation project, AT&T & Intel initiated – Feb 2018. carrier-scale edge computing applications running in virtual machines and
containers. The Akraino Edge Stack community anticipates releasing project code in the Spring of 2019. Open source software stack to improve the
state of edge cloud infrastructure
https://www.akraino.org/
Open stack foundation project. Intel and Wind River have jointly open sourced this software. Run the minimal services required at the edge, yet
provide robust support for bare metal, container technologies and virtual machines.
https://wiki.openstack.org/wiki/StarlingX
Reference platform design for edge cloud in OPNFV – main focus. Edge clouds share a similar but simplified NFV+SDN architecture as the core DC but
has more implementation constraints. Service Provider's Management of Edge Cloud
https://wiki.opnfv.org/display/PROJ/Edge+cloud
Looking Edge & ONAP components in terms of data collection, processing, policy management, resource management, control loop models, security.
https://wiki.onap.org/display/DW/Edge+Automation+through+ONAP
There are several projects under Eclipse Foundation supports IoT and Edge platforms. Eclipse foundation cloud platform stack, Eclipse foundation IoT
working group & Eclipse foundation with CNCF for IoT Edge are some of them.
9. 9
https://www.edgexfoundry.org/
• Dell has seeded this project with FUSE
source code base under Apache 2.0.
• Analog Devices, Dell and Samsung are
the platinum members (as on Dec, 2018).
• More than a dozen microservices and
over 125,000 lines of code.
• ~70 Organizations (as on Dec, 2018).
11. 11
AKRAINO Edge Stack
A Linux Foundation open source Project:
AT&T & Intel initiated – Feb 2018. carrier-scale edge computing applications running in virtual machines and containers.
The Akraino Edge Stack community anticipates releasing project code in the Spring of 2019.
Akraino Edge Stack will offer users new levels of flexibility to scale edge cloud services quickly, to maximize the applications
or subscribers supported on each server, and to help ensure the reliability of systems that must be up at all times.
https://www.akraino.org/
https://wiki.akraino.org/display/AK/Akraino+Edge+Stack
13. 13
Openstack StarlingX
Open stack foundation project. Intel and Wind River have jointly open sourced this software. Run the minimal services
required at the edge, yet provide robust support for bare metal, container technologies and virtual machines.
StarlingX is both a development project and an integration project that combines new services with many other open source
projects into an overall edge cloud software stack.
https://wiki.openstack.org/wiki/StarlingX
https://www.openstack.org/edge-computing/
15. 15
ONAP - Edge
To efficiently and effectively deploy 5G network supporting ultra low latency and high bandwidth mobile network, we need to
deploy variety of applications and workload at the edge and close to the mobile end user devices (UE or
IoT). https://wiki.onap.org/pages/viewpage.action?pageId=28381325
16. 16
The Open Source stack
The IoT Cloud Platform represents the software
infrastructure and services required to enable an IoT
solution.
An IoT Cloud Platform typically operates on a cloud
infrastructure (e.g. OpenShift, AWS, Microsoft Azure,
Cloud Foundry) or inside an enterprise data center
and is expected to scale both horizontally, to support
the large number of devices connected, as well as
vertically to address the variety of IoT solutions.
The IoT Cloud Platform will facilitate the
interoperability of the IoT solution with existing
enterprise applications and other IoT solutions.
Some of the projects are:
Eclipse Kapua – manage IoT gateways
Eclipse OM2M – For telcos for M2M spec
Eclipse Hono – API for devices & extensible framework
Eclipse Mosquito – MQTT broker
Eclipse Leshan – OMA LWM2M
Eclipse hawkBit – Software update device & gateways
Eclipse BIRT – Dashboard & reporting
https://iot.eclipse.org/cloud/
Eclipse Edge : Eclipse IoT Cloud Platform Stack
17. 17
One project that Bosch is leading is Eclipse Hono, which does large-scale MQTT data ingress from as potentially many as millions
of end devices, converting it into AMQP and then routing it to the application that cares about the data from that particular device.
Hono runs atop of Kubernetes.
Eclipse Hono™ provides remote service interfaces for connecting large numbers of IoT devices to a back end and interacting with
them in a uniform way regardless of the device communication protocol.
Hono specifically supports scalable and secure ingestion of large volumes of sensor data by means of its Telemetry and Event
APIs. Hono's Command & Control API allows for sending commands (request messages) to devices and receive a reply to such a
command from a device in an asynchronous way.
https://www.eclipse.org/hono/
Eclipse Edge : Eclipse Hono
18. 18
Eclipse Kura™ is an extensible open source IoT Edge Framework based on Java/OSGi. Kura offers API access to the hardware
interfaces of IoT Gateways (serial ports, GPS, watchdog, GPIOs, I2C, etc.). It features ready-to-use field protocols (including
Modbus, OPC-UA, S7), an application container, and a web-based visual data flow programming to acquire data from the field,
process it at the edge, and publish it to leading IoT Cloud Platforms through MQTT connectivity.
an extensible open source IoT Edge Framework based on Java/OSGi. Develop new Components and Application, Drag-and-
Drop new modules from the Eclipse IoT Marketplace.
https://www.eclipse.org/kura/
Eclipse Edge : Eclipse Kura
19. 19
• The Eclipse IoT Working Group is a collaboration between
organizations and individuals who share the goal of creating
an open IoT. The collaboration focuses on the development,
promotion and adoption of open source IoT technology. Our
members provide a wide range of projects and services built
on top of Eclipse IoT technology.
https://iot.eclipse.org/working-group/
https://wiki.eclipse.org/IoT
https://github.com/kubernetes/community/tree/master/wg-iot-edge
Eclipse Edge : Eclipse Foundation IoT working Group
20. 20
Eclipse Edge : Eclipse Foundation with CNCF for IoT Edge
• Formed a new Eclipse working group focused on improving
Kubernetes IoT and edge deployments, Supported by - Red
Hat, Bosch, Eurotech, InfluxData, Siemens, Vapor IO, and
VMware. Is a collaboration among the 40-member Eclipse IoT
Working Group and the Kubernetes community
• Plan to provide reference architectures for various IoT/Edge
environments. https://iot.eclipse.org/working-group/
• The complexity of orchestrating IoT systems is a problem
domain for which Kubernetes is a perfect fit, Eclipse
Foundation's Executive Director, Milinkovich added
• https://thenewstack.io/cncf-eclipse-explore-kubernetes-driven-
internet-of-things-edge-computing/
21. 21
Open Source Edge platforms - more…
1. Mirantis Cloud Platform Edge (MCP Edge) with kubernetes - https://www.sdxcentral.com/articles/news/mirantis-rides-kubernetes-to-charge-open-
source-edge-ecosystem/2018/10/
2. Edge compute platform - Deutsche Telekom and Aricent have partnered for the creation of an Open Source, low latency available to operators, to
enable them to develop and launch 5G mobile applications and services faster. - http://www.eenewseurope.com/news/edge-compute-platform-open-
source
3. Open-RAN Mobile Edge Computing Platform - Saguna Open-RAN improves quality of experience (QoE) by providing web services with high-bandwidth,
low-latency communication to mobile users and connected ‘things’. With fully virtualized software architecture, Saguna’s MEC platform enhances
network agility and promotes the transition to network function virtualization (NFV) using cost-effective commercial-off-the-shelf (COTS) solutions -
https://www.sdxcentral.com/products/saguna-open-ran-mobile-edge-computing-platform/
4. Open Edge Computing Initiative is a collective effort by multiple companies, driving the business opportunities and technologies surrounding edge
computing. The members are Microsoft, Intel, Carnegie Mellon University. http://openedgecomputing.org/
5. Radisys Open Mobility Solutions - Radisys provides an end-to-end solution for M-CORD that combines the principles of virtualized RAN, virtualized EPC,
SDN and NFV in one platform for MEC, IoT and 5G deployments. http://www.radisys.com/mobilityengine/solutions
6. OpenVolcano: An Open-Source Software Platform for Fog Computing - exploit Network Functions Virtualization (NFV) and Software Defined
Networking (SDN) to support personal cloud services. https://www.semanticscholar.org/paper/OpenVolcano%3A-An-Open-Source-Software-Platform-
for-Bruschi-Lago/32c6e2224d1b61de46fdc8875dfb52dbed77b867
7. MFX-1 IoT EDGE COMPUTING GATEWAY - COMPLIANT WITH LINUX FOUNDATION EDGEX FOUNDRY OPEN SOURCE FRAMEWORK -
https://www.mainflux.com/edge-computing.html
8. FogLAMP - open source platform for the Internet of Things and an essential component in Fog Computing. It uses a modular microservices architecture
including sensor data collection, storage, processing and forwarding to Historians, Enterprise systems and Cloud-based services.
http://dianomic.com/platform/foglamp/
9. Open Carrier Interface - An Open Source Edge Computing Framework - https://dl.acm.org/citation.cfm?id=3229579
10. Project Flogo - A Lightweight Runtime for Edge Computing - https://thenewstack.io/project-flogo-a-lightweight-runtime-for-edge-computing/
11. Forest Giant - Open source projects made/used by the Forest Giant team. Distributed discovery and streaming key value store.
https://github.com/forestgiant & https://platform.forestgiant.com/
12. Intel, Alibaba launch a joint edge computing platform - Intel now produces a range of different AI accelerators, including CPUs, field programmable
gate arrays (FPGAs) and Movidius vision processing units (VPUs). Its AI technologies also include OpenVINO, a toolkit designed to bring computer vision
and deep-learning inference to vision applications at the edge. https://www.zdnet.com/article/intel-alibaba-launch-a-joint-edge-computing-platform/
13. Baidu Cloud launches its open-source edge computing platform - platform will include features like data collection, message distribution and AI
inference, as well as tools for syncing with the cloud. https://techcrunch.com/2019/01/09/baidu-cloud-launches-its-open-source-edge-computing-
platform/
14. And more………
23. 23
kubeedge – Open Source Edge Platform Initiated by Huawei.
• www.github.com/kubeedge , Nov, 2018,
• Kubernetes IoT Edge WG
V o lum e C o nfig m a p Po d Pro b er Event ...
M etaM anag er
Ed ged
C o nta ine rs
D eviceTw in
D a ta
Sto re EventBus
Ed geH ub
Po d /V o lum e/...
D evic es
M Q TT Bro ker
kubectl
A PP
SD K
Note : Here the edge-controller is in cloud. Currently most of the cloud providers want the controller and coupling with cloud services.
This is more of business than technological reason
24. 24 24
Edged: Edged is an agent running on edge node for managing
user's application.
EdgeHub: EdgeHub is a web socket client, which is responsible
for interacting with Huawei Cloud IEF service, including sync cloud
side resources update, report edged side host and device status
changes.
EventBus: EventBus is a MQTT client to interact with MQTT
server(mosquitto), offer subscribe and publish capability to other
components.
DeviceTwin: DeviceTwin is responsible for storing device status and
syncing device status to the cloud. It also provides query interfaces for
applications.
MetaManager: MetaManager is the message processor and
between edged and edgehub. It's also responsible for storing/retrieving
metadata to/from a lightweight database(SQLite).
kubeedge – More Details…
Edge Computing
With business logic running at Edge, volumes of data can be
secured & processed locally. It reduces the bandwidth request
between Edge and Cloud; increases the response speak; and
protects customers' data privacy.
Simplify development
Developers can write regular http or mqtt based applications;
containerize and run anywhere at Edge or Cloud.
Kubernetes-native support
With KubeEdge, users can orchestrate apps, manage devices
and monitor app/device status against Edge nodes like a normal
K8s cluster in the Cloud
Abundant applications
You can easily get and deploy complicated machine learning,
image recognition, event processing and other high level
applications to your Edge side.
Advantages Key Components
25. 25
Kubeedge Next**…
• Data & Data Analytics
• Multi Cluster
• Security (spiffe etc)
• Function as a service
• More protocols and device types
• Edge Market Place and Ecosystem
• Pluggable North Bound Connection – Support Vendor Agnostic Central Cloud (though it is challenging to make
central cloud edge service agnostic)
• Service Mesh, CNCF Projects Integration
• Improve monitoring, health, remote management, and CI/CD
**Add more community driven features to build the open ecosystem. Collaborate with CNCF and other partners for Edge Computing Alliance
https://github.com/kubeedge/kubeedge
Slack channel: kubeedge.slack.com
India Contact : sanil.kumar@Huawei.com / Krishna.m.kumar@Huawei.com
• To build an ecosystem for edge computing
• To build a generic light weight open source edge platform with
competitive features to support your use cases
26. 26
Key Challenges
Managing the Offline Scenarios
Fragmented Device Specifications and Protocols
Heterogeneous Edge Node / Compute
Distribution and Best utilization of Edge Cloud Resources
Security & Privacy
Real time performance
Huge set of user scenarios for same use case, Huge number of devices/nodes
connectivity & its management
28. 28
Roof Computing 1931.1
• The development of the IEEE Standard for an Architectural Framework for Real-time Onsite Operations Facilitation
(ROOF) for the Internet of Things.
• This standard defines an architectural framework, protocols and Application Programming Interfaces (APIs) for providing
Real-time Onsite Operations Facilitation (ROOF). ROOF computing and networking for the data and the devices include
next-hop connectivity for the devices, real-time context building and decision triggers, efficient backhaul connectivity to
the cloud, and security & privacy.
• The standard covers interoperability, collaboration and autonomous operation of an Internet of Things (IoT) system with
computing required for context building, security, access control, data storage, data aggregation and ability to choose
different cloud and application service providers [5].
• Furthermore, this standard defines how an end user is able to securely provision, commission and decommissions the
devices.
• It leverages existing applicable standards and is complimentary to architectural frameworks defined in broader IoT
environments.
29. 29
https://www.openfogconsortium.org/
Multi-access Edge Computing (MEC) offers application developers and
content providers cloud-computing capabilities and an IT service
environment at the edge of the network. This environment is
characterized by ultra-low latency and high bandwidth as well as real-
time access to radio network information that can be leveraged by
applications.
.
https://www.etsi.org/technologies-clusters/technologies/multi-access-edge-computing
Some Edge Platform Alliances…
specifies signalling requirements and architecture of
intelligent edge computing to provide intelligence to the edge
network for efficient data processing within the network.
https://www.itu.int/ITU-T/workprog/wp_item.aspx?isn=14275
31. 31
Edge Computing is identified as one of the key trends
The vast potential of Edge Compute to support all smart and
intelligent solution for end users.
Possible integration of all cutting edge technologies viz., AI/ML, Big
Data, Cloud, IoT, Blockchain…
Research with Industry, Community and Academic Collaborations..
Edge Computing is at at its Phase 1…!
32. 32Thank You! Kumar Brothers! Krishna Kumar & Sanil Kumar
Disclaimer: Images & many data are taken from Internet and only used for information sharing. We do not claim any other rights/correctness to it. Also no commercial usage of these slides allowed.