The release of the Raspberry Pi 4 with a quad core processor and 8GB of memory opens up new possibilities for enterprise level applications on a small form factor. At $75, multiple boards can be purchased and incorporated into an appliance form factor. By clustering the boards you can achieve enhanced performance and improved availability.
One use for such an appliance is what I call a Federated Data Caching Appliance. This drop-in appliance allows you link information from various data sources together, build views into the data based upon a schema you define, cache the information for quick retrieval and surface the views in a variety of different protocols. I have based this on technology from Radiant Logic, but other technologies can be substituted.
Imagine taking data from your HR, CRM and inventory systems and joining the information into a common view. What insights could you gain from that information? How could your applications leverage that data? How about building a view that linked a salesperson, his or her manager, his or her vacation schedule, what the salesperson has sold and the inventory available of those items. With that information, alerts could easily be generated for a manager when a client is running low on a product and inventory is available and the salesperson covering that client is out on vacation. It's a complicated scenario, but any data that can be pulled together and correlated can then be made available for consumption by applications. By separating the view from the physical representation, you have complete control over how the data is represented and made available through multiple protocols.
Radiant Logic's solutions provide the following capabilities:
With its sophisticated methods allowing you to quickly link to underlying data sources, define a schema for the data, join it, and deliver the data through multiple protocols, the technology provides a good engine for the Federated Data Caching Appliance. Additionally, the solution supports clustering for high availability and scalability. The solution requires three servers, but more can be used.
This appliance could be designed to house three (or more) Raspberry Pi's into a single highly-available device at a low cost point. By adding a second appliance you gain external high-availability as well. With the web-based administration and dashboards available in FID, a UI for managing the appliance could be quickly created:
The appliance could be designed something as follows:
The appliance has three Raspberry Pi's for the FID cluster which are powered over ethernet. The box also has redundant power. Two of the units would be deployed for high-availability.
Granted, there are some challenges to this approach. A build of FID that runs on ARM Java would have to be made available. Additionally, the default microSD-based storage would have to be replaced with something more scalable. However, this is an interesting experiment.
Have fun!