Real-Time Analytics at the Edge
Quickly distribute your solution at scale.
Join industry leaders and a special guest for a distributed cloud virtual event on 9 March 2021 at 11 AM ET. Register now.
The number of devices used in edge networking situations continues to grow exponentially year over year. By the end of 2022, according to Gartner, 75% of all data generated will be collected, analyzed and acted on at the edge. By implication, to do that work at such volume and scale, streaming analytics solutions need to become commonplace in enterprise IT strategies. And, by definition, those solutions must operate with very low latency.
Returning results of data analysis in more than milliseconds may be too slow for user experiences that depend on the outcome for making quick choices in the moment. A retail offer, for example, could be irrelevant if received after a pedestrian has walked further down a shopping street; or, more seriously, a medical device that is about to fail may be in use before predictive analytics sends a maintenance alert.
Besides needing to operate with very low latency, many business applications may need to process data at the edge as part of complying with local finance regulations.
Market pressure across the economy has led companies of all sizes to adopt cloud technology as a competitive necessity. To achieve business goals, large companies that operate across regions are especially challenged to define a cloud-based strategy that also enables a scalable solution at the edge that meets latency and data residency requirements.
One approach has been to adopt multiple clouds that, together, afford geographical reach and agility in satisfying business needs. But with that comes complexity in managing operations related to the different clouds used. In fact, performing operations can come to outpace delivering new value to customers in application experiences.
Additionally, many companies with established estate IT on-premises have been reluctant to move workloads onto a cloud platform for lack for internal skills and because the security risks are seen as too high. For them, creating hybrid solutions makes sense — modernizing part of an application in the cloud and connecting back to hardened systems on-premises. However, such hybrid cloud architectures often do not meet low-latency and data-local requirements of edge use cases.
Distributing cloud services where needed
Distributed cloud is an emerging solution to the constraints of hybrid cloud and the complexity of multicloud strategies. With a distributed cloud architecture, users have the flexibility to consume cloud services on-premises, in public clouds and at the edge.
IBM Cloud Satellite is a distributed cloud solution that includes the benefit of being offered as-a-service. Satellite customers flexibly consume cloud services without the burden of maintaining them; instead, IBM does that. Satellite includes access and identity management across all environments — or “locations,” as these sites are called. Satellite Link provides secure tunnels through which teams run their own networking policies. A single view in IBM Cloud makes visible all services and applications running across locations. Teams use that same view to manage provisioning and deployments.
Scaling a streaming analytics solution at the edge
Let’s look at a scenario. WorkSafe offers a workplace safety solution that operates at job sites. Through artificial intelligence software, video cameras monitor visitors for hard hats; in their absence, the system triggers an alert that impedes entry. Meeting the system’s low-latency requirements depends on processing video data as soon as its generated.
With the onset of the COVID-19 pandemic, customers asked WorkSafe to adapt their solution to focus on enforcing compliance with health guidelines. Part of the change involved teaching the system to consider different movement rules during monitoring. Cameras would be able to determine the distance between people on site and trigger an alert should the minimum distancing rule be violated. And, with the addition of thermal-sensitive cameras, the artificial intelligence software was retrained to detect employees’ temperatures instantaneously.
Having developed software for the new use case, WorkSafe uses IBM Cloud Satellite to deploy in customer sites in the same way they did with the initial workspace safety solution. Red Hat OpenShift clusters are already set up to receive the containerized software in each site. Customers likely would choose to expand their OpenShift clusters to run the new software in addition to the software that supports the existing use case. Assuming the video cameras send data to servers running both versions of the analytics software, the system simultaneously can follow rules related both safety use cases. Satellite’s single, centralized console provides a comprehensive view of all application activity.
Bring public cloud to the edge
Companies that provide solutions at the expanding edge of digital business need to leverage the benefits of public clouds without being limited to the data center footprints of providers. More specifically, IT strategy needs to satisfy these requirements:
- Meet data latency and residency requirements.
- Consistently deploy, operate and manage software at the edge.
- Transform existing applications in days, not months.
- Scale delivery of solutions despite the general scarcity of internal skills for doing cloud native development and operations.
IBM Cloud Satellite meets these requirements by enabling teams to launch, use, monitor and manage consistent cloud services anywhere.