Pursuing Certification(s)
After graduating from Flatiron in May with my certificate in Data Science, I began my search for employment. It quickly became clear that the best way to get my foot in the door, rather than simply putting down my skills on a resume, was to pursue certifications in my skills to pass along to recruiters. Fortunately for my, while scrolling through LinkedIn, freeCodeCamp posted a link to 115 free Coursera courses, each offering their own certification at the end of the course. Going through the list, I elected to enroll in many of the available courses, ranging from information architecture to computer vision.
I’ve started my journey by pursuing certification in Industrial IoT on Google Cloud Platform. As I’ve learned, Google provides access to many different cloud computing sources that can be used to analyze big data from potentially thousands of sensors for any company. The Internet of Things, however, is not simply made of sensors and devices. The Internet of Things is made up of these devices, as well as their cloud counterparts that help process the data and produce insights for future growth and innovation. With the right sensors, actionable data even offline can be captured to anticipate events proactively. And Google Cloud Platform provides the resources to capture and process this data.
In the main, IoT is configured through devices, the gateways in which the devices access the internet and send/receive data, and then the cloud, where data is stored, processed and analyzed. Devices connected to the cloud use their environment, oftentimes, to gather data or perform actions. Oftentimes devices are embedded into other pieces of equipment, machinery, perhaps. Information from the physical world is received, and then transmitted digitally.
Gateways are put in place between devices and the internet, to make sure that data is sent securely to the cloud. Examples of gateways include cellphones, computers, microprocessors, etc. These devices sometimes perform the calculations that would otherwise occur on the cloud. This is called edge computing, and it is often instituted to alleviate some of the strain put on cloud processing when there are large amounts of data. Any useful cloud computing must be scalable. That is, it must be able to add, remove, or modify devices that support the transfer of data. The protocols and pipelines that support the transfer of data between devices and the cloud must be able to adapt to be agile, and produce actionable insights.
The Internet of Things is made upon the client/server model, where devices are authenticated, authorized, and connected to various nodes across a network. Without this architecture, data bottlenecks occur; errors cascade, and lead to poor insights. Devices fall into the client/server model with edge computing, to help carry some of the burden of the computational process. Data analysis, machine learning, and data gathering can, of course, be done on the cloud, but with edge computing some of these tasks can be preprocessed, and then verified on the cloud.
With these IoT capabilities, Information Technology and Operational Technology have begun to overlap. Where Information Technology has historically been centered around data, Operational Technology was used to create insights from monitored events. Now that devices are embedded into machinery, IoT has merged the two fields.
Along the pipeline, data must first be ingested, then processed, and then stored. This, of course, must be done quickly to be valuable. With edge computing, only cleaned metadata is sent to the cloud, accelerating the cloud’s ability to compute, analyze, and create ML insights. Other than edge computing, asynchronous communication can also alleviate some of the strain of processing so much data.
Google has created a special platform just for edge computing, called Cloud IoT Edge; this is a platform that provides resources for real time analytics and machine learning. But it is not as simple as this. Google IoT Edge is made up of Cloud IoT Core, Cloud Functions, Pub/Sub, and Dataflow — all of which I will explore in my next post.