The Internet of Things (IoT) and the industrial version, the Industrial Internet of Things (IIoT) are revolutionizing everyday life and are writing the future of Industry 4.0. With the use of sensors, they can store and cross-reference data flows in the cloud, which are then analyzed by Big Data and Artificial Intelligence technologies. This is how we helped Kerlink to automate the deployment of its IoT platform while avoiding vendor lock-in.
Data can be about the geolocation of devices, machine temperature in a factory or even calculating the flow of people. The possibilities are endless, as we’ve already seen in the LTE-M challenge organized by Orange and the French railway company, SNCF. The opportunities afforded by IoT are so promising that analysts like Gartner have predicted they will be a major source of innovation in the next decade.
This proliferation requires the ability to quickly deploy these solutions while meeting the customer’s demands, such as proximity, control, etc.
Good infrastructure is IoT ready
If they are to reach their potential, IoT technologies require flawless infrastructure and platforms from data harvesting through to its analysis. Kerlink, the network and application layer provider, relies on the LoRa protocol and has chosen to provide its software offering on Amazon’s cloud (Amazon Web Service or AWS). We supported this company with their global infrastructure management, and I shall build on this experience to identify guidelines that may be useful to other companies wanting to launch similar projects.
Host the solution or use the cloud?
Originally, Kerlink’s solutions were based on their customers' architectures, on-premise, or, for the multi-client SaaS version, hosted by a local cloud provider. As the business was growing its client-base globally and responding to specific regulatory requirements for cloud hosting, it needed to store data as near as possible to its customers. Kerlink selected AWS for its global coverage. The need to speed up their commercial deployments, as well as flexibility and automation requirements, equally led to the use of the cloud. Today, Kerlink offers its customers a dynamic infrastructure as an IaaS offering, available anywhere in the world. However, the company plans to remain cloud-agnostic in order to offer a multi-cloud service.
Beyond the choice of AWS – expert support
The Breton-based company, Kerlink, needed to improve its knowledge of the AWS cloud infrastructure and its associated tools, so it looked for a business to support it.
AWS offers a large range of managed services, network configurations, virtual machine templates and types of storage. Faced with this extensive range, Kerlink quickly realized that it would need support in choosing these services from experts in the field. This selection was carried out using reversible criteria to avoid too much focus on AWS services, such as resilience and cost.
Development of the existing IoT platform
The Orange experts tried to minimize the impact on the software while making the steps as automated as possible, from the deployment of IaaS resources to the deployment of the Kerlink software as well as all the operating activities, e.g., updates, upscaling and return to operations after a failure. This was done to avoid Kerlink needing to maintain two versions of the solution, i.e., the on-premise and the AWS versions, and to allow operational teams to intervene in a solution deployed in the same way, whatever the platform. The strategy adopted had an agile approach, which has now been replicated many times.
Supporting Kerlink’s IoT platform deployment on the cloud in four steps:
1 - Requirements analysis: Building on the customer's issues and requirements and helping them acquire skills to use AWS services (e.g., platform and tools)
2 - Tool selection to carry out the project: Selected from the AWS services (e.g., Amazon VPC, Object Storage S3, Security Groups, EIP) and chosen add-ons (in this case Terraform and Packer)
3 - Implementation and service delivery: During this phase and during testing, exchanges with the customer are constant. The aim is to build the product together and enable the customer to quickly take full ownership
4 - Testing
As for the technical aspects of the project, the choice of Terraform and Packer made it possible to automate and considerably reduce the deployment time. These choices were both technological and strategic and fitted with Kerlink’s goal to avoid vendor lock-in and becoming too dependent on a cloud provider’s solution, which leads to issues for their own customers when changes occur in the architecture or with the service provider. The required solution was therefore the adoption of agnostic technologies to run the cloud platform, ensuring its complete, flexible reversibility as necessary.
Expertise is indispensable
With its aim of implementing its IoT platform in the cloud to meet its customers’ expectations in relation to speed and ease-of-implementation, Kerlink was faced with the difficulty of understanding and managing the AWS cloud. Its choice of partner – in this case Orange Business Services – enabled it to better understand the catalog and the functionality of the Amazon cloud platform, to manage the cloud implementation in an agnostic way with independent tools, and to carry out the necessary developments to adapt to the cloud solution with an agile, iterative approach. Working in this way has allowed the company to remain focused on its own business while its teams quickly got to grips with the technology and improved their skills – all to the benefit of its customers.
For more information, read the case study: Kerlink facilitates its global implementation by using a multi-cloud business application.
As a member of the Professional Services team, I support our clients with cloud transformations, using methods and technologies linked to DevOps. I am passionate about the Open Source world, cloud computing and automation (Infrastructure as Code and scripting).