Edge Computing for your Organisation

Background

We will explain how Edge Computing can be leveraged to benefit your organisation. Edge Computing is a technology created and used at Nexedi since 2008. It has been combined with Amarisoft vRAN stack to form Rapid.Space.

Nowadays, Rapid.Space provides edge computing solutions for telecom industrial automation, AI, big data, aerospace, government, Industry 4.0, etc.

Agenda

  • Edge Computing Introduction
  • Edge Computing Success Cases
  • Edge Computing by Rapid.Space
  • Rapid.Space Edge Acceleration Programme

We will go through four steps, how, why, where does it come from and then in the field success cases. We will move next to how Rapid.Space provides Edge Computing service in practice and then our Edge Acceleration Program that can make your organisation adopt Edge Computing and reap the benefits within as little as a few weeks to a few months of adoption.

Edge Computing Introduction

Early forms of Edge computing has actually existed since the 1990s with one key application: Content Delivery Networks (CDN). CDN evolved in 2004 with proprietary implementation of popular application runtime (ex. J2EE) part of the CDN. But it is only in 2008 that appeared a general purpose form of edge computing that could replace cloud.

The creation of Edge (2008)

Jean-Paul Smets at Nexedi is one of the creators egde computing with Christophe Cérin at Paris University in 2008. Christophe was a specialist of grid computing and Jean-Paul was a specialist of management of ERPs. Both thought if they could manage automatically a distributed system with an ERP. Just like an ERP can be used to automate sales, e-commerce or factories, an ERP could be used to automate the deployment of corporate IT services on a distributed grid computing system.

Tthat's how SlapOS was born, an open source edge operation management system, where the ERP takes the orders, provisions the computers and generates the invoice for arbitrary IT services such databases, virtual machines, IDEs, web applications, etc. A grid computing node does all of the technicalities on the computers that are spread all over the world to convert orders into reality. 

Essentially edge was created as a combination of ERP and Grid Computing. We were very interested at that time in edge because we were afraid that clouds with centralised and few data centres were posing high risk to our societies. For example if Facebook or WhatsApp has an outage there is an extensive impact on many people. We considered that one of these data centres could easily be attacked with a graphite bomb, potentially melting hardware and leading to an outage which could even be worse than the World Trade Centre attacks in 2001 in terms of negative economic impact.

What is Edge Computing

  Cloud Edge
Topology a few DC with thousands of computers  thousands of DC with a few computers
Automation Fully automated operation management of arbitrary IT service

Edge Computing is very similar to cloud in that it completely automates all the aspects of an IT Service. Rather than paying a system administrator to install, update, monitor, optimize every service, this is done by software completely automatic. The more customers one has, the more we can automate and accelerate how a database works, how a CDN works, how a virtual machine uses RAM. That's automation and there is no difference in that aspect between edge and cloud. The big difference is the way a cloud system is designed is to operate a few data centres with thousands of computers each, whereas edge computing is designed to automate and operate thousands of data centres with a few computers each. This is fundamentally the difference.

How does Egde look like?

  DC Campus IoT
Product     
Distance < 100 km < 1 km < 1 m
IoT < 1 ms < 10 µs < 10 ns

How does it look? There is not a single kind of edge computing. We try to see edge computing as 3 different types: edge Data Centre, Campus edge and edge IoT.

An edge Data Centre is like having your cloud data centre less than 100 km away from a factory or office. With Campus Edge, and edge server is placed in an office, factory, student campus with less than 1KM away. And with edge IoT, the edge computing workload is running directly into a connected thing.

And edge Data Centre is around 16 to 32 servers in a rack, circa 50KM away and will use the same hardware as Facebook or twitter for their own cloud, called Open Compute. Looks like a mini cloud close to where you live.

Campus type Edge looks like a small server placed in a very short-size rack in the factory or the office, somewhere in a storage room. Low power, low-profile, small size.

Edge IoT will be using very small embedded systems that can feed in the drone or a machine tool or 5G base station and that will look more like embedded boards.

Edge Data Centre is suitable for use cases with require at most 1 millisecond latency such C-RAN. Campus Edge is suitable for at most ten microseconds latency use cases such industrial automation. Edge IoT is suitable for at 10 nanoseconds latency use cases such as radio unit, industrial coupler or autopilot.

What are the advantages of Edge Computing

  1. resiliency (splinternet, geopolitics, energy cost & availability, CO2 emissions)
  2. latency
  3. global availability
  4. legal immunity and compliance (incl. security)
  5. economic efficiency (no diminishing returns)

Edge Computing has five basic advantages: resiliency, latency, availability, legal immunity, economic efficiency.

Resiliency that's when you are running campus edge, you can still run your office or company even if the internet has been cut. Nowadays the internet can be cut more frequently that we used to imagine. In Algeria in North Africa, each time they have exams for university they cut country-wide access to the internet so students cannot use it for answers to exams. This also cuts the internet in factories, this is why if you use cloud then you are cut, while if you use campus edge then you are not cut. Geopolitics also create barriers with congestions or deep packet inspection (DPI) which prevent you accessing cloud. For example developers in Iran cannot use GitHub in the USA. Many enterprise web sites hosted in US are blocked in Chine. Many enterprise web sites hosted in China are blocked outside China. With an edge architecture, such as DC Edge or Campus edge you can bypass these geopolitical challenges.

Energy also has a significant cost now, for example hosting edge in your home in France you get much better price than in your office due to energy subsidies that are only applicable at home, or hosting infrastructure in certain countries, such as Spain and Germany, you get a better price in one country because the rules for energy calculation are different. Availability is also important because sometimes energy becomes less available, in some countries there are energy shortages. Some countries use coal to produce electricity while others use hydroelectric/renewable power. With edge you can make your own decisions to navigate energy costs, geopolitical challenges, environment, everything is your decision. 

The second advantage is latency. If you have to drive machineries in a factory, you can't have your server on the other side of the Atlantic or Pacific because that would mean 0.2 seconds latency which is huge and means that if for example a worker's hand goes into a machine and the sensor would only realise after 0.2 seconds which could be life threatening. The latency is also important for 5G applications and robotics drones.

Availability is when you need cloud and there is no hyperscaler service such as in Namibia, they need Edge in order to have all the services of cloud which can be used locally. 

On immunity and compliance, as you can decide where the edge node is located in the factory or in the country then you can decide which law applies. For example, the Cloud Act in the United States has extra-territorial reach, just like the Chinese Cyber Security Law. Whenever your data is hosted in a data centre outside your country there is a risk that your data is being monitored without your knowledge.

Economic efficiency. Economy's of scales are a good thing. Both cloud and edge give the ability to scale and implement automation, however when huge data centres are made, after a certain size diminishing returns ultimately lead to inefficiency and make you less competitive. With Edge the size can be adjusted to be small enough to mitigate the problems of diminishing returns and thus maintain efficiency and competitiveness. 

Edge Computing Success Cases

Edge Computing has been deployed in real world much before it gained popularity. Early success cases consist of deployments based on cloud automation technologies in various context where public cloud is not an option.

Resiliency: Stellantis in Algeria

ERP deployed on factory with SlapOS (Campus Edge) to circumvent nationwide Internet outage during university exams

We had to deliver an MRP, system to manage a car factory, for Stellantis Group (merger of Fiat Chrysler, Opel, Peugeot, Citroën) in North Africa. This is campus type Edge. The problem is that in Algeria for a few days each year the internet is completely cut meaning no use for any individuals or companies. So we placed an edge node, a small computer called EdgePOD, running SlapOS, the system we created in 2008 and that automates the complete deployment of the factory ERP, the campus CDN and the low latency networking in the factory.

Which meant that Stellantis did not have to hire highly skilled people locally and the system was very cost effective to maintain and a few thousand cars could be built with the system.

Latency: Viettel VRAN

C-RAN type vRAN based on SlapOS and Amarisoft (DC Edge) + autonomous 5G base stations (IoT Edge)

Rapid.Space has a memorandum of understanding (MOU) to deploy C-RAN solutions of vRAN, a DC type edge with a few servers in it computing the radio signals and connected to radio units. We use SlapOS, combined with Amarisoft, the leading company in the world for everything related to Virtual Radio Access Networks (vRAN). In addition we provide Autonomous 5G Base Station, an all in one system that could be placed in factories for private networks.

We see two cases of Edge: data centre edge and IoT edge. In both cases latency is essential. In a centralised system the distance between the BBU (Base Band Unit) and the RU (Radio Unit) should be less than 1 millisecond or 100 km, which is the time Amarisoft stack is capable processing digital radio signal in advance. In practice less than 50 km. The same is important in factory systems, you need the radio and the computer processing the signal to be very close. So edge is mandatory to get the kind of latency that is necessary to deploy a 5G network.

Latency: Olimex OPC-UA

IEC 61131-3 PLC (Campus Edge) controlling an Olimex coupler (IoT Edge) over OPC-UA TSN

In another case, we worked with Olimex in Bulgaria to create a machine tool based on the OPC-UA standard. Olimex is one of the leaders in the world for open source hardware for industry. We automated a production line for sorting cherries, if the cherries had a lot of sugary or not. This is fully automated with a camera looking at the colour and depending on the colour it would go to line A or B. With this we need decisions computed in a couple microseconds and the only way to do that is to have a campus edge system running a programmable logical controller (PLC), to take the decision A or B, respectively sweet or not sweet cherry. The programmable logical controller is connected to a coupler, which is a sensor that gathers the information on the colour or an actuator which executes the decision to push the cherry in a specific direction. This is IoT edge. It requires latency between the PLC and the couplers in 10s to 100s of microseconds, and for the coupler itself in 100s of nanoseconds.

Legal Compliance: Woelfel Big Data

Collection of vibration data into Big Data Hub in compliance with German Law on critical infrastructures

Woelfel is collecting a lot of information from wind turbines in Germany to try to detect and predict failures of the turbine, listen to noise in bank data centres to ensure that nobody is going to break in illegally or to control the noise level of the works in the streets of Munich. For some of these applications it is absolutely forbidden to have devices connected to the internet or to have servers placed anywhere else, except in the company that is doing the service. Due to legal compliance there was no other way than to have on-premise deployment of the edge nodes, talking about campus edge type, to collect the data and process it. If we would have used cloud in the traditional way in a 3rd party location, it would have been likely illegal because the wind turbine etc. is considered in Germany as critical energy infrastructure.

Legal Immunity: Airbus TSXX ERP segment

Cloud-native hosting of the ERP segment of a dual destination satellite without foreign legal interference

We also have in Germany, deployment of management segment for dual destination satellites, civilian and military and these have to be hosted inside on-premise yet we want to use automation techniques like cloud but it has to be in the offices of Airbus behind very strict firewall. This is the typical campus edge setup. We also need the technologies to be disconnected from foreign governments, i.e. Chine, Japan, USA, because of sensitive information. So being able to navigate legal compliance challenges, makes edge computing very useful. 

Availability: Stellantis in Namibia

Cloud-native hosting in a country without hyperscaler zones

Furthermore, in Stellantis cars factory in Namibia, the challenge was that there were not many local cloud providers so it is much easier to put in a campus type server to provide cloud locally than wait for some hyperscaler to create a local zone.

Edge Computing by Rapid.Space

Rapid.Space has designed a consistent offer to accelerate the adoption of edge in your organisation and benefit from the unique competitive advantages of SlapOS edge operation technology.

Edge by the creators of Edge (2008)

  • Proven for 15 years at Stellantis, Airbus, Kyorin, Toyota, MBDA, Woelfel, SANEF, ODOT, etc.
  • Buil-in PaaS ✅
  • Built-in SDN ✅
  • Built-in CDN ✅
  • Built-in IoT gateway ✅
  • Built-in PLC ✅
  • Built-in vRAN ✅
  • Built-in PTT ✅
  • Hard real-time ✅
  • Cloudless computing ✅

As the creators of general purpose Edge Computing in 2008, Rapid.Space has 15 years of experience. Rapid.Space's edge technology has already been deployed at Stellantis, FiatChrysler, Airbus, Kyorin (Japanese drug company), Toyota, MBDA (missles), Woelfel (engineering for German industry), SANEF (highway company), Oregon department of transportation, etc. All these companies or government bodies currently use Rapid.Space edge technology. That's why our feature set is already quite wide. It has a built-in Platform as a Service (PaaS) to automate all developer work, it has a software defined network (SDN) that reduces latency globally, content delivery network (CDN) that can accelerate website similar to Cloudflare or Akamai, but in our case open source. We have IoT gateway in order to buffer data from factories or from sensors before they are sent to big data lake. We have virtual PLC for industry for 4.0 to automatically control machines. Also we have support for vRAN that is currently used by telecom industry for 4G/5G and embedded push to talk server. Hard-real time to reach hard limits in terms of deterministic operating system, sub 10 microseconds.

One of our latest innovations is called cloudless computing which further simplifies the adoption of edge and drastically cut costs by making 90% of the server infrastructure irrelevant.

All of this is available in Rapid.Space.

Resilient by Design

  • Automated disaster recovery on multiple sites or clouds
  • Self-optimising IPv6 overlay network to reduce latency and congestion
  • ITAR-free hardware and software for legal immunity
  • Zero knowledge credential management protects privacy by design
  • Fully Open (open source, open hardware, open process) ensures global availability

How we are different from other vendors is that all our products are resilient by design. We provide automated disaster recovery by copying and rebuilding systems on multiple sites in the world, completely automatically. The sites can be your sites, our sites, depending on you. It has self optimising IPV6 overlay that can circumvent congestion and/or sometimes blockages by governments. For example, if you have a factory in Europe and China then they can communicate and exchange data with no challenges. We provide ITAR free hardware and software for legal immunity which means the hardware we use some is from USA components but we also have "made in China" CPUs which are free from any form of export control or restrictions which actually protects from the risk of being under the scope of certain laws whether in Europe or USA. Of course we also supply made in Europe hardware to suit European customers. Overall providing legal immunity.

The credential management has a zero knowledge feature where we can construct things so that we don't know any of your passwords or whatever is required to access your system. Only you know the credential ensuring that even if in Rapid.Space we have unforeseen bad actors, we won't be able to see into your edge system. And we won't be able to provide your credentials for foreign governments.

We are fully open, you can copy our hardware and software and fully copy our open process. This way, whatever happens you can fully replicate our edge service at any time, guaranteeing that no matter what happens with geopolitics, operation of your services can go on. True resiliency from the outside.

Interoperable by design

  • Standard based (OPC-UA, IEC 61131-3, PLCOpen, TSN, RFC 8966)
  • CPU independent (Intel, AMD, NXP, ST, Hygon, etc.)
  • OS independent (any Linux OS or POSIX OS) for edge native deployment
  • Support of Windows OS through virtualisation
  • Deployment on public clouds (AWS, Azure, GCP, Alicloud, Tencent, etc.) for multi-edge-cloud
  • Deployment on private clouds (VMWare, OpenStack, Vates)
  • Deployment on off-the-shelf hardware (OCP, Dell, HP, Inspur, Rapsberry Pi, Olimex, etc.)

We are interoperable by design, much more than others. For example, we use OPC-UA industrial standard, IEC-61131-3, PLCOpen, TSN. Our routing is based on RFC 8966 also known as Babel. We are standard based.

We are CPU independent, whichever you prefer is OK: Intel, AMD, ARM, NXP, ST, Hygon, etc.

We are OS independent, which means any Linux distribution, FreeBSD or any POSIX operating system can run native edge cloud workloads without requiring virtualisation. This is quite unique to our technology.

For Windows, we use virtualisation.

We are already deployed on AWS, Azure, GCP, Alicloud, Tencent cloud. All of these already run some Rapid.Space workloads with SlapOS, our underlying edge stack. This means we are a already a multi-edge cloud provider.

We can deploy our edge on VMWare, Openstack or VATES xcp-ng. We are fully interoperable and deployed on 3rd party private clouds.

And you can purchase hardware such as Open Computer (OCP) servers, Dell, HP, Inspur, Raspberry Pi, Olimex and simply run Rapid.Space on it. Nothing can block Rapid.Space running on any off-the-shelf hardware environment.

Fair Pricing

Pure edge service
  Setup Recurring Extra
DC 0€ 40€ / month / node 0.5€ / instance / month after 80
Campus 0€ 20€ / month / node 1€ / instance / month after 20
IoT 0€ 10€ / month / node 2€ / instance / month after 5

For pricing, if you have to run up to 5 instance that is 10€ per mont, up to 20 instance that is 20€ per month, 80 instance is 40€ per month, and beyond 0.5€ per month per instance. This pricing is flexible enough to support the different types of edge computing: DC, Campus and IoT. 

Rapid.Space Edge Acceleration Programme

Now that you know that Edge works, Edge is efficient, has many advantages and is interoperable with essentially everything, how can your organisation adopt Edge?

We created different programmes to accelerate edge adoption in your organisation.

EdgePOD

  • Target: companies with technical expertise in Linux programming or system administration
  • Outcome: pilot plan to adopt edge computing and port of the first application
  • Includes webinar "how can edge computing benefit your organisation"
  • Online learning track for developers and system administrators
  • Dedicated support for one developer or system administrator
  • Prerequisites : knowledge of Linux programming or advanced Linux system administration
  • Price :  450 € + 40€ / month (24 month commitment)

The EdgePOD programme is for companies already with some engineers that can do programming on Linux or system administration. It comes with a plug-and-play, 8-core EdgePOD server.

With EdgePOD programme what you get is the pilot plan to adopt edge computing in your organisation and port the first application to edge, which includes a webinar, how can edge computing benefit your organisation. We will study what is in your organisation in terms of IT landscape, what can be beneficial to your organisation through edge computing and then we will try and support your developers with an online learning track that's easy for developers and system administrators. We will provide an electronic work group and we will support one of your developers or system admins daily until you get success. For this program there is a prerequisite of good knowledge of Linux programming or advanced knowledge of Linux system administration skills. 

The cost is 450€ for the full programme and 40€ per month with 24-month commitment, which is roughly under 2,000€ to move from no edge skills to a pilot plan with your first application ported and something up and running. 

EdgePOD Executive

  • Target: companies with managers of digital transformation
  • Outcome: pilot plan to adopt edge computing and port of the first application
  • Includes webinar: "how can edge computing benefit your organisation"
  • Includes 80 hours of support by experienced developer to port one of yours applications to the edge
  • Prerequisites : access to source code or binaries of one application to port to the edge
  • Price :  4450 € + 40€ / month (24 month commitment)

Second plan is the EdgePOD Executive plan which is similar to EdgePOD but for managers who understand they need digital transformation but do not have necessarily have the skills in terms of development or system administration. The outcome is the same in both programmes, with a pilot plan to adopt edge computing in a beneficial outcome for your organisation and port of the first application. The webinar will also be used to describe the landscape in terms of IT and applications, but we will bring our experienced engineers and give you 80 hours of developer support who will help port one application to the edge. For this to happen, they will need good access to source code of binaries for at least one application which can be ported to the edge. The cost for this program is 4450€ set up + 40€ per month with 24-month commitment.

We estimate that within as little as 3 weeks, your organisation will be running edge computing. In more complex environments this can also take some months.

Open Radio Station (ORS)

  • Target: companies with technical expertise in networking
  • Outcome: pilot plan to adopt 5G vRAN and deployment of the first 5G private network
  • Includes webinar "how can edge computing benefit your organisation"
  • Online learning track for network administrators
  • Dedicated support for one network administrator
  • Prerequisites : knowledge of TCP/IP networking
  • Price :  850 € + 80€ / month (24 month commitment)

For some applications, edge has to be combined with 5G, we offer the ORS programme targeted at companies that have some expertise in TCP/IP networking. It comes with a plug-and-play ORS 4G/5G base station and edge server. It does not require any 4G/5G expertise.

You will get a pilot plan as an outcome for the ORS acceleration programme where the plan will explain how to adopt 5G vRAN and deploy 5G private networks with edge in your organisation with a webinar focused on 5G benefits for your organisation. We will support your sraff with an online learning tracking for network administrators and dedicated support for 3 months until they can take the system live. Prerequisties is knowledge TCP/IP networking knowledge. Price is 850€ + 80€ per month with 24-month commitment.

POD

  • Target: companies with big data requirements or hundreds of applications to host
  • Outcome: pilot plan to adopt DC edge computing and port of the first application
  • Includes webinar "how can edge computing benefit your organisation"
  • Online learning track for developers and system administrators
  • Dedicated support for one developer or system administrator
  • Prerequisites : knowledge of Linux programming or advanced Linux system administration
  • Price :  15,000 € + 40€ / month / sled

The POD programme is suites for governments and large companies that need to host hundreds of applications. It is also suitable for smaller organisation with big data requirements, especially organisations that are facing "obscene" invoices from traditional cloud providers. It comes with 10 barebone Open Compute servers and all the parts required for mounting in a 19-inch rack cabinet.  

With POD programme what you get is the pilot plan to adopt DC edge computing in your organisation and port the first application to edge, which includes a webinar, how can edge computing benefit your organisation. We will study what is in your organisation in terms of IT landscape, what can be beneficial to your organisation through DC edge computing and then we will try and support your developers with an online learning track that's easy for developers and system administrators. We will provide an electronic work group and we will support one of your developers or system admins daily until you get success. For this program there is a prerequisite of good knowledge of Linux programming or advanced knowledge of Linux system administration skills. 

The cost is 15,000€ for the full programme and 40€ per month per server sled, which is, when adding the price of CPU, RAM and storage, roughly under 25,000€ to move from no edge skills to a pilot plan with your first application ported and and an Edge DC up and running with a few servers. 

EdgePACER

  • Target: academia, open source projects, NGOs, home developers
  • Outcome: deploy a personal edge server on Raspberry Pi or Olimex Lime2
  • Online learning track for developers and system administrators
  • Prerequisites : knowledge of Linux programming or advanced Linux system administration
  • Price : 20 € one-time
  • Limitations : up to 5 instances

Our last programme is the EdgePACER program, aimed for academia, open source projects, NGOs and individual developers who are interested in learning what edge computing is, a more education option.

Similar to the EdgePOD programme but you will be running a personal edge server on a Raspberry Pi or on an Olimex LIME2 micro-server. You will have access to the online track for developers. You need Linux programming skills or advanced Linux system admin skills. Price is 20€ one time but you can only host up to 5 edge instances, only for Raspberry Pi or Olimex LIME2.

Edge Acceleration Comparison

Rapid.Space Edge Acceleration
  EdgePOD
EdgePOD Executive
ORS
POD
EdgePacer
Price 450€ + 40€/month 4,450€ + 40€/month 850€ + 80€ / month 15,000€ + 40€/month/sled 20€
Commitment 24 months 24 months 24 months None None
Online learning track
Webinar  
Dedicated support  
80h developer        
Max. instances no limit no limit no limit no limit 5
Edge type Campus Campus Campus DC IoT

 

In summary.

EdgePOD 450€, for organisations with Linux developers or system administrators.

EdgePOD Executive 4,450€ for organisations with digital transformation executives.

ORS, 850€, if you need private 5G.

POD, 15,000€, if you need DC Edge.

EdgePACER, 20€, for education and IoT.

Depending on the programme you will have dedicated support, up to 80 hours of developer expertise or technology tailored webinar.

Thank You

  • Rapid.Space
  • 147 Rue du Ballon
  • 59110 La Madeleine
  • France
  • +33629024425
  • contact (@) rapid.space

For more information, please contact Jean-Paul, CEO of Rapid.Space (+33 629 02 44 25).