IoT edge applications will need a data pipeline that stretches all the way from devices to the data center and treats devices as ‘first-class nodes’, say Nutanix executives.
The cloud essentially exists as the underpinning foundation for the IoT – and the connection channels that extend from the back end outwards to its ‘edge’ are an increasing focus for the enterprise tech industry and its customers.
The world’s top three cloud providers (Google, AWS and Microsoft), for example, have somewhere close to 3.5 million servers in their datacenters. This massive surface area is a natural hunting ground for Nutanix, a company that previously describe itself as a hyperconvergence specialist, but now prefers to present itself as a provider of a cloud operating system (OS).
Either way, Nutanix makes software that allows customers to use, view and manage, migrate, monitor and optimize their use of cloud computing ‘instances’. This refers to those chunks of cloud resource that customers sign up for to provide them with data storage, core processing and analytics.
Read more: Nutanix maps out its edgy IoT cloud vision
Back-end base, front-end view
Although focused on the back-end infrastructure aspects of cloud services, Nutanix also has its eye on the surface layers of computing, including the workloads kicked off by the IoT and the role of so-called edge computing.
Nutanix executives explain that the company focuses on IoT use cases that marry real-time edge intelligence with core cloud computing. Customers can use Nutanix as an ‘intelligent edge’ for IoT applications based on Google Cloud Platform (GCP), by deploying TensorFlow (an open source library for machine intelligence) for edge processing, while training machine learning models and running analytics on the processed metadata in GCP.
But, they add, edge applications will need a data pipeline that stretches all the way from devices to the data center, as processing is now shared across a wider transept. What this means for software application developers is that functions (in terms of actual compute execution tasks) may now be more ephemeral (typically existing as smaller scale, discrete microservices) and more flexible in nature.
Satyam Vaghani, vice president of technology at Nutanix says that his firm’s Prism management interface has been engineered to appreciate the existence of the IoT itself. Because of this, it is able to treat ‘edge’ locations as ‘first-class nodes’, so that they can play an appropriately important role in total networks.
Vaghani clarifies his firm’s understanding of the current trend to build explicit distributed systems, saying that in this distributed architecture world, we know that edge applications are typically very vertical-specific and engineered to serve specific use cases. On paper at least, this could make them very difficult to replicate and deploy for additional, but different use cases.
In terms of product updates, Nutanix now says it will enhance support for applications using large unstructured datasets. This could typically be areas such as big data analytics, data warehouse applications and large-scale IoT deployments. To optimize delivery of these applications, Nutanix will be extending its data management capabilities to include object-based storage, which application developers can use as a native service.
Complex stuff? Yes it is, that’s Nutanix… but the company is seeking to provide a simpler means of managing clouds tasked with IoT workloads through its Prism management interface. Nobody said architecting for para-virtualized microservices running on hyperconverged server nodes for edge IoT deployments was going to be easy now, did they?
Read more: Are we edging closer to IoT Edge Computing?