Edge computing has emerged as an important architecture for applications that need to deliver high bandwidth, low latency, and autonomy to geographically distributed users. Examples of edge computing applications include: Internet of Things (IoT), Augmented Reality, Autonomous Cars, and Massively Multiplayer Online Games (MMOG).

Edge computing is going mainstream, and one of the leading software stacks supporting this architecture is the Ormuco Stack. Ormuco has been a pioneer in edge computing, cutting their teeth on MMOG, providing gaming companies with a low-cost, scalable, computing stack, leveraging open source software and running on commodity hardware.

New call-to-action

The Ormuco Stack is used to push applications, data and services away from centralised hosting centers to the endpoints of the network, closest to the users. In addition to minimizing long distance network traffic, an edge computing stack enables local autonomy by supporting analytics and intelligence at the the logical extremes of a network. In a multiplayer game, for example, if you shoot your opponent, you need to know if their character died instantly. The analytic computations for determining the damage inflicted must happen close to your computer.

Since Ormuco started, the need for edge computing has grown far beyond the niche of online video games. Today, the world around us is increasingly populated with intelligent sensors from home security systems, to real-time video sensors, to patient monitors. This rapidly growing Internet of Things requires edge computing to become ubiquitous. Ormuco has evolved their technology to keep pace and now provides a robust and flexible software architecture ready to meet this growing need.

Edge Computing Example: Preventative Maintenance for Manufacturers

To understand the value that Ormuco provides in the edge computing ecosystem, it helps to start with an example application such as preventative maintenance in factories.

Modern manufacturing plants, such as auto assembly lines, contain a lot of sophisticated equipment: stamping presses, robotic welding machines, torque wrenches, CNC machine tools, conveyor belts, die casting equipment, etc. When a piece of equipment breaks, it can cause an expensive work stoppage, or even a serious injury. As a result, manufacturing companies are increasing installing sensors on all this equipment to collect data that can be used to predict failures. That data is fed to machine-learning algorithms to distill out parametric values that drive probabilistic models that anticipate failures. When a potential failure is anticipated by the model, a maintenance inspection is scheduled. In this way, the plant management team anticipates failures before they happen, preventing expensive breakage, and improving safety.

New call-to-action

This is a classic IoT application. There are lots of sensors collecting data. This data needs to be processed by machine learning in real time. Moreover, the distilled analytics data produced by the machine learning algorithms needs to be interpreted to know when to apply preventative maintenance. Some situations require real time responses. Say a person walks in front of a robotic welding arm. The arm needs to stop immediately. It cannot wait for a probabilistic model to identify danger. Some situations require careful analysis. Suppose that 17 measurement variables need to be analyzed to determine when a die needs to be replace in a casting machine. In this case, centralized intelligence needs to run a model that determines when the variables are collectively outside the desired tolerances and maintenance should be performed. This kind of processing is important but does not require a real time response.

Now consider what the computing architecture needs to look like to support this kind of application.

null

At the factory, sensors of all kinds are sending video and other measurement data to an edge cloudlet for processing. The cloudlet applications, then communicate with centralized applications for the predictive modeling and maintenance schedule. What’s new here is the cloudlet that provides low latency services on the edge of the network, close to the factory.

Cloudlets and Edge Computing

A cloudlet is a small-scale cloud datacenter hosting resource-intensive applications in close proximity to the end user. In this case, the end user is a factory which could be anywhere in the world. A cloudlet can be set up nearby, in a managed colocation provider, to provide the bandwidth and latency needed by the applications. Connectivity from the factory LAN to the Cloudlet may be wireless or wired. Minimizing latency is a key concern, to provide for rapid response, as in the case of the robotic arm that needs to be stopped quickly to prevent accidents.

The purpose of the cloudlet is to bring cloud services closer to the edge of the network. The cloudlet provides the key middle-tier of an edge computing architecture. In this case, the IoT sensors are the endpoints of the architecture and a regional cloud hosting facility is at the top.

null

In the example we are considering, the factory may be one of many, geographically distributed manufacturing centers owned by a single company. Each factory is linked to a local cloudlet hosting the highly responsive applications, while the cloudlets are integrated at a central cloud location hosting predictive modeling and maintenance applications that are used across the company. The cloud may also host high-latency, processing intensive, applications like deep learning. In this case, the deep learning application could be finding new patterns in the data to improve the predictive model. 

What is the Ormuco Stack?

The Ormuco Stack is a software framework the enables you to set up and manage your own cloud services. A distributed manufacturing company, like the one in our example, could use Ormuco to implement and manage the cloudlets near each of their factories. Or a managed service provider could lease the cloudlet infrastructure to the manufacturer and administer the edge computing infrastructure for them using Ormuco.

In either scenario, Ormuco enables the edge computing provider to provision and manage the computing resources in each cloudlet as needed to support the local factory. The diagram below shows some of the components that can be configured in the Ormuco Stack. This illustration is by no means comprehensive, but provides an idea of the services provided by Ormuco.

null

Access

The cloudlets supporting each factory can be managed from the Integrated Control Panel. This control panel includes a User Portal for point-and-click administration and also provides REST APIs that give you programmatic control for developing orchestration applications.

Support and Intelligent Operations

Immediately beneath the access layer, the Ormuco Stack provides an integrated operating environment built on the most recent and stable OpenStack release. Ormuco delivers their platform using continuous integration and continuous delivery (CI/CD) to meet your SLAs.

The Ormuco Stack is highly instrumented and designed to be self-healing. Artificial intelligence and machine learning algorithms are incorporated to detect and correct problems. As a result, the administrator of an edge computing infrastructure built on Ormuco does not need to worry about supporting the nuts and bolts of the environment, and can turn their attention to the business applications.

New call-to-action

IaaS and PaaS

The components of the cloudlet infrastructure can be managed through the user portal. Anyone familiar with managing an AWS, or other public cloud, environment will appreciate the ease and convenience with which infrastructure (IaaS) can be deployed using Ormuco Stack. Here you can configure your computing resources at various levels including: virtual machines (VMs), pods (Kubernetes), and containers (Docker). You can even incorporate bare metal servers with the user portal.

Software defined storage can also be configured, including block, object, and file storage systems. Networking and administrative components like security groups, load balancers, DNS vRouters, etc. are configurable. Platform services (PaaS) are included as well, including provisioning for a variety of database engines including MySQL, PostgreSQL, Hadoop, MongoDB, and more.

Then there are enterprise services, like backup and disaster recovery options (defined and managed using Freezer) and autoscaling. There is even a billing engine to track resource consumption, assign costs, and generate invoices - particularly useful for managed service providers.

Hypervisor

Under the hood, most customers are using Ormuco Stack with Kernel-based virtualization (KVM). But, other hypervisors are supported, such as Hyper-V, and if you are migrating from an existing virtual machine environment (including VMware), Ormuco will work with you to import your VMs into their environment, potentially saving on licensing costs.

Hardware

Ormuco is a software company, so, unlike public cloud providers, they do not provide the hardware that runs your cloudlets. However, they do partner with colocation centers to install the Ormuco Stack on commercial off-the-shelf (COTS) or commodity hardware. They also have an enterprise-grade partnership with Hewlett-Packard Enterprise (HPE) to delivery service provider-ready solutions installed on HPE gear.

Simplifying the Complex

Ormuco’s goal is to simplify the complex and make edge computing solutions manageable for IT organizations. To that end, in addition to delivering a full stack for edge computing, the company has partnered with IIS to support adoption of their software in North America.

IIS - Ormuco’s North American System Integrator Partner

International Integrated Solutions (IIS) is Ormuco’s exclusive system integrator partner in North America. IIS has deep data center expertise and is a distinguished HPE partner, winning HPE Global Partner of the Year in 2016 and Arrow’s North American Reseller Partner of the Year in 2017.

As your Ormuco systems integrator, IIS can help with:

  • Sizing - determining the hardware specifications needed to support your edge workloads running Ormuco.
  • Architecture - designing the hardware solution for the datacenters hosting your cloudlets.
  • Colocation Support - ensuring that your colocation partner installation is successful.
  • Service Desk - while Ormuco supports the core stack, IIS can provide help desk support above the VM level and act as a single point of contact for your IT staff.

Edge Computing Has Arrived

The rapid adoption of IoT, wearable devices, and artificial intelligence is driving the need for computing solutions that push computing power out of centralized datacenters and closer to the edge of the network. The Ormuco Stack provides a path for IT organizations at companies, as well as managed service providers, to implement and manage edge deployments that provide decentralized computing services that can be centrally managed.

Jeff Smith

Written by Jeff Smith