Honeycomb, running on AppAgile, is a multi-tenant and customizable platform-as-a-service (PaaS), a result of joint collaboration between Roambee and T-Systems to operate a scalable, smart logistics platform – enabled through Red Hat OpenShift. This integration is the first known use of OpenShift and AppAgile as an IoT platform.
The goal is to provide a trusted, on-demand, end-to-end, and a Smart Logistics service. This means, we compute to put a dot on the map that shows the shipment location as well as additional shipment health conditions with a high level of accuracy.
We do that while intelligently and proactively notifying you about your shipment in real time. And, we accomplish that while providing devices, connectivity, cloud hosting – an entire inclusive service.
Also, our architecture needs to scale easily as we collect increasingly more messages from the sensor devices (Bees) and process it to derive business information.
After running on AWS for a year, we are migrating to T-Systems’ OpenShift-enabled AppAgile platform. This blog post shares our learnings with developer evangelists who are evaluating various cloud options for their enterprise scale IoT applications and offers insight into how Roambee, T-Systems and OpenShift are aiming to deliver value to customers.
Roambee offers an on-demand shipment monitoring service built on Honeycomb, our IoT API Platform. Honeycomb is architected to collect sensor data and translate that into in-transit visibility information, predict disruptions by using machine learning and recognize patterns to improve our customer’s supply chain.
When we think of an IoT platform, we think in terms of clean, composable, and robust APIs, alongside solving different business use-cases.
Honeycomb was originally developed on AWS but in due course of time, in spite of battle tested DevOps convenience of AWS, it became clear that our customers wanted many different flavors of cloud deployment and hence we started considering a cloud agnostic approach for Honeycomb.
Honeycomb-3 will be cloud agnostic leveraging T-Systems’ AppAgile PaaS, which supports DevOps, microservices, cloud and container technology. Today AppAgile enables customers of every size and in nearly every vertical to build disruptive software with enterprise grade. The core container management of AppAgile is based on Red Hat Openshift.
Tenets of IOT:
Any IOT platform naturally comprises of device events, processing pipeline, rules engine and analytics to extract intelligence and actions from the data, and a robust set of APIs that can be accessed and leveraged to build smart systems using the information. Let’s take a look at what makes up the Roambee solution.
Honeycomb – MPP (Message Processing Pipelines)
As our customers grew from handful of Bees to 100’s, there was an immediate need to scale the data collection and processing pipeline. Since message collection and processing is the backbone of any IoT application we created Beetle.
Beetle includes the ingestion, parsing, processing and enrichment semantics with an ability to handle large volumes of data and hence scales out maintaining the reliability and availability and other business driven metrics, powered by a node.js microservices stack.
Beetle supports different device communication protocols such as TCP, UDP, HTTP, MQTT, and their variations so that developers require minimal learning to support various IoT use cases. The platform makes use of big data compute semantics to bring streaming data and compute together in an elastic, scalable, and resilient manner producing reliable and fast results, with real-time notification model.
Honeycomb – Analytics
We created Instinct – our analytics pipeline that supports different aggregation and statistical frameworks. Instinct uses clustering mechanisms to build interpretive and predictive models using big data.
Honeycomb – APIs
Bubble is our robust, stateless and scalable API layer that empowers developers and business owners to accelerate and simplify IoT adoption. The APIs provide device interoperability, IoT data management, event handling, rules development and application integration in a toolkit. It provides developers with a built-in cloud deployment machinery designed to simplify, automate, and eliminate the steps associated with setting up the foundation for your application. You can get your IoT application deployed much more quickly, and iterate, adapt, and extend it more rapidly over time. With composable, and complete APIs, our clean interfaces and abstractions can handle any IoT application needs – from device registry and message delivery to analytics and everything in between.
Why OpenShift for Roambee?
- The self-service and easy to provision middleware components with capabilities to induct containerized software, as well as source-control repositories, makes it practical to use OpenShift as a strong container application platform offering for our IoT message processing pipeline – Beetle.
- In the rapidly growing IoT-Market, we have within AppAgile & OpenShift the possibility to evaluate and test new technologies more quickly than we could on our own.
- The model enables us to have full end-to-end management of our software lifecycle — including simple and flexible deployment options and CI/CD integration.
- A few frameworks we required,such as Storm, were not out of the box, but OpenShift offers us the ability to easily augment that with our own solutions.
- We wanted to be able to eventually run on any infrastructure provider.
We plan to evolve Honeycomb in the coming months to support various use cases. If you want to help accelerate or prioritize the pace of innovation, contact Vinay Mishra at firstname.lastname@example.org.
You can Learn More about how our companies are working together to define the IoT space, here: http://www.honeycomb.to/, or contact us at email@example.com .
Vinay Mishra is the Sr. Director, Engineering at Roambee responsible for product engineering, scalable cloud architectures, and data analytics. Prior to Roambee, Vinay worked at Yahoo, leading teams on signal driven targeting and user behavior analytics platform using Map/Reduce, Hbase and Storm technologies. He also has experience in designing solutions for large scale computing with a good understanding of cluster and parallel architectures as well as high-scale or distributed RDBMS and knowledge of NoSQL platforms. Before Yahoo, he led business critical engineering products and projects at eBay and IBM. He can be reached at firstname.lastname@example.org