The concept of edge computing has been in use for quite a while now in IoT networks. It makes immense sense for compute to be shifted to the edge in these kind of networks since we can save on costly bandwidth by avoiding unnecessary traffic over the WAN. Sensors that sent data up only when a metric changed Or just when thresholds were breached helped save on precious WAN bandwidth. More importantly this also dramatically reduced the amount of data that had to be processed.
More recently this trend of computing at the edge of the network is being increasingly embraced in the enterprise with several not-so-obvious yet key benefits.
Let’s look at some of these.
Point of deployment
Network monitoring is best done near the point where its performance is perceived. If you’d like to measure how the network is treating its users, its best done at the edge. Hence apps like Cisco ThousandEyes are ideally suited to run in devices like the Catalyst 9300 and 9400s that form the access layer of the network.
Apps that guard the network are best deployed where attacks strike first. An app that simulates a honeypot or one that looks for suspicious rogues is best when deployed where such attacks strike first.
Apps that monitor the OT(Operational Technology) network have to be deployed at the point where the IT and OT networks intersect. The Cisco Cybervision app hence belongs here.
As we see from the above examples, Edge Apps have to be deployed in those parts of the network where they can perform their function the best.
Point of compute
The fact that network bandwidth is abundant in the enterprise may lead us to believe that it does not pay to process information at the edge. Computing at the edge of the network has another not-so-obvious benefit. In reducing the amount of traffic we send up to the controller or data center, we save on precious compute resources at those points in the network where traffic is aggregated like a wireless controller or the management station in the data center. This frees up compute for the other more important functions that these network equipment deliver like scaling to more sessions or ensuring better performance.
Enterprise IoT convergence
Customers had to deploy overlay networks in order to achieve IoT use cases like ESL (Enterprise Shelf Labels) tags or for managing devices over ZigBee or other IoT protocols. With an extensible USB dongle in the Cisco Catalyst range of access points, the same infrastructure deployed for providing WiFi networking services can be used for IoT use cases too. This results in lesser management and operational overheads.
Cisco DNAC Spaces embraces app hosting on edge equipment to support legacy IoT devices that run the not-so-latest protocols.
What’s changing now ?
The coming together of three different complementary technologies from Cisco stables is truly changing the game here.
Apps
Compute at the edge is only useful when we have killer apps that offer business critical functionality. Cisco ThousandEyes which already loved and used by enterprises world over is now available to be run on the Catalyst range of switches. So is Cisco Cybervision which helps with industrial visibility and security and Cisco Edge Intelligence which helps get IoT data to the right application at the right time.
App hosting capability
Cisco has been working on the app hosting capabilities on the supported range of Catalyst switches and access points to enable several more capabilities that the apps can exploit. Be it greater available bandwidth for operations, the ability to utilize the in-built SSD on the device in-lieu of external storage or even getting access to more privileges as a Docker container running on the switches, apps on switches can now do so much more.
Cisco Catalyst Access Points support a third party dongle that apps can access and use. This ability is what apps like SoluM and ImagoTag use. Plans are afoot to grant the apps access to many more capabilities on the access points.
Orchestration of apps via Cisco DNA Center
Unlike networking configuration, apps on device could pose a very different challenge to the network administrator. Apps have a life of their own – they need to be configured at start, be monitored while in action, be restarted when in need and be upgraded to fix vulnerabilities or avail new features. Network admins can ill afford to use another tool to this purpose since it would mean that the network inventory needs to be kept in sync on both tools.
Managing apps in enterprise networks require an integrated console that can help manage the network and the apps running across it.
Cisco DNA Center fits this ask to the T with the app hosting capability. Apps can be managed across hundreds or even thousands of devices (switches or access points) with this capability.
The Cisco Thousand Eyes can be installed across on several vantage Catalyst devices with this one workflow.