Platform as a Service (PaaS) adoption has been slow to gain broad acceptance in enterprises. We’ve seen some large enterprises embrace PaaS within in a specific product line or vertical within a company, but it is rare that we see PaaS as an architecture standard adopted across an entire company.
The promise of a company embracing a single PaaS solution has frequently fallen short as well. Enterprises are often large and disconnected and many have subcultures within IT. One silo of UI/UX developers might embrace a mobile backend PaaS, sometimes called MBaaS. Another silo within the company may build on the public cloud and embrace the PaaS solutions and services provided by AWS, Microsoft, or Google. Still another silo may focus on big data solutions and embrace a database as a service (DBaaS) platform like Hortonworks. Yet another silo may buy into the a cloud agnostic PaaS that sits on top of an infrastructure, such as Pivotal’s Cloud Foundry.
All these different PaaS solutions are adding IoT functionality to their products and services. Some of the most popular IoT features focus on tasks such as ingesting IoT data, providing hardware developer kits to integrate with a variety of sensors and chips, persistent state management, and IoT gateways. But there is one big piece all these solutions are missing.
How do I get access to the sensor data?
All the above mentioned platforms are great solutions once you have access to the data streams or feeds. What they don’t address is the layer outside the cloud, which is called the fog. The following image shows the complexities lying in the fog that need to be solved before these cloud platforms can begin to process even one byte of data.
In the fog, there are a number of different connected “things” generating real-time information. These things may be a sensor, a smart phone, a surveillance camera, a drone, a connected car, or a countless array of connected devices. All these things must communicate through some known protocol, so the communication hardware that picks up the signals knows how to process the data and where to send it.
The number of permutations a developer would need to code for to interface with a wide variety of things, communication protocols and communication hardware, quickly becomes unmanageable. Often the developer does not have the technical expertise to work within this fog layer.
The solution lies in a new type of PaaS called fog platforms.Without a fog platform that abstracts all the complexities within the fog, building IoT applications would be very challenging and time consuming. In addition, developers would have to stay on top of technology changes within the fog layer and would constantly be maintaining code to deal with device and networking issues, instead of working on solving business problems.
Fog platforms will therefore be a key component for building IoT applications. In fact, we are already seeing cloud PaaS solutions starting to integrate with fog platforms. So the dream of leveraging a single platform to build IoT applications is just that, a dream. The reality is we will need to work with multiple platforms--in the cloud and in the fog--that integrate with each other.
Industry specific platforms
We’ve seen the emergence of industry specific clouds focused on verticals, such as government and healthcare. The same thing is happening the fog. There are platforms for connected cars, connected homes, healthcare, manufacturing, and many other verticals. The future architecture might look like a cloud platform that integrates with an all-purpose IoT platform in the fog, which integrates with a variety of industry specific IoT platforms.
I’ve been asked if I thought an industry specific platform was a good idea or a bad idea. I think it is both. It is bad from the standpoint that it adds much more complexity to the ecosystem. But it is good, because an industry specific platform can be stripped down to focus only on what’s needed for that industry. For example, a manufacturing plant does not have to be HIPAA compliant and will not have to integrate with a large number of sensors and machines that are common in the healthcare industry. In addition, the builders of these vertical platforms have domain experience and will add more granular features than those you find in an all-purpose platform.
This space is so new that it will most likely get even more complex until we gain a lot more experience building IoT solutions. It reminds me of the early days of cloud computing and how many features needed to be added before enterprises could take the public cloud seriously. After several years of getting user feedback and establishing better standards, public clouds are now widely embraced within the corporate world. IoT will need to go through that same evolution.
Summary
IoT will drive up the demand for PaaS, but expect to live in a multi PaaS world. Architectures are getting very complex and very distributed. The old days of building apps from the front end to the back end with just code will be ancient history. We will have to rely on platforms because our business partners are counting on us to deliver new features to the market at the speed of business. And the speed of business is getting exponentially faster every day.
