Feedback from recent demos at Cloud Native PDX and ACM/IEEE Symposium on Edge Computing (SEC)

[scottmbaker]

I’m excited to relay some of the feedback we had from recent engagements with the community about the Edge Manageability Framework.

Cloud Native PDX (November 2025)

This was a meet-up in Portland, Oregon attended by professionals who work in the cloud native software space. Most attendees were very familiar with Kubernetes and similar tools, though some were unaware of the differences between orchestrating Kubernetes Clusters at the Edge versus at the Cloud. Much of the talk consisted of describing the differences between Cloud and Edge and the diverse types of hardware that are prevalent at the Edge. The Edge offers many unique challenges, such as disconnected operation and troubleshooting issues with distributed hardware. Intel vPro generated much interest, and the audience could envision the scenarios where remote out-of-band hardware management would be necessary, such as dealing with a bricked hardware device without involving a costly physical truck roll.

AMC/IEEE Symposium on Edge Computing (December 2025)

This was a conference in Washington, DC. We conducted demo sessions on both the Edge Manageability Framework and Edge Microvisor Toolkit, though I will primarily discuss the EMF presentation in this post. The audience at this conference was well versed in the unique challenges that the Edge presents to orchestration, and immediately appreciated the contributions that EMF makes. There was specific interest in the Cluster Connect Gateway and Application Service Link features. The audience had experienced challenges being able to interact with hardware that was behind NATs and firewalls, and these features facilitate connectivity to edges via the centralized orchestrator. One audience member discussed how his team had implemented a similar feature, but not to the level of production readiness that EMF strives for.

Intel vPro again generated interest, and some members were already familiar with AMT. We had productive discussions contrasting in-band management approaches with out-of-band management.

As part of the EMF demo, we also included a couple of deployed AI Suite and AI Library applications – Pallet Defect Detection and ChatQnA. The audience particularly liked seeing the AI applications as practical demonstrations of edge orchestration and appreciated that they were deployed on physical edge nodes at a remote location, the same type of scenario that would be seen in the field.

Green energy usage at the Edge was a topic of several papers at SEC, and one audience member felt that an orchestration solution like EMF with its open APIs could be used to facilitate automatic migration of workloads to Edges as energy availability changed.

Another hot topic of the conference was federated learning and various schemes for partitioning AI workloads between handheld devices, nearby edge sites, and/or the cloud. Presenters advocated doing for the usual reasons – latency, privacy, or economics. This is another example where workloads may be highly dynamic, with the need to schedule edge sites as demand changes.

Looking forward, I feel that spending some time promoting EMF’s API and providing some simple examples of integrating external automation controllers or schedulers could build on the momentum of these areas of interest.

[biapalmeiro]

@scottmbaker you have been such a great partner and community champion. Love all the energy, effort, time and passion you put into that. Thanks a lot!