For the second example, consider a manufacturing company deploying quality control sensors along its assembly line.
Instead of trying to inspect a completed item—a set of high-end headphones—at the end of the manufacturing process, they will use robotic measuring systems to catch out-of-specification parts and assemblies at several points in the manufacturing process.
Some of the quality control inspectors who currently run a complete suite of tests over each completed set of headphones will move into positions where they receive alerts about a failed test on a tablet. Once notified, these “roving inspectors” will examine the failed part, determine if a pattern of failures is emerging (with the help of a data analytics system), and decide whether to move production off the affected workstation. These inspectors will also be trained to perform some repairs locally or open tickets for a trained technician to come in and repair the impacted equipment.
Corporate leadership is very protective of information about the production line’s efficiency, failure rate, and failure modes.
The chief information security officer (CISO), an expert in industrial espionage, is concerned that access to this information would give competitors a good overview of the company’s proprietary manufacturing process.
Figure 17-5 illustrates an application design that would work for these requirements.
Figure 17-5 Private Cloud Deployment Example
Figure 17-5 shows another multi-stage application:
• Factory floor equipment has wired network access to a set of on-premises data collection servers.
• These virtual servers do some initial processing, sending the data to an off-site colocation facility for storage.
• Analytics servers in the colocation facility look for patterns, failures, etc., and send raw notification streams to the on- premises notification servers. These servers are off-premises but still part of the private cloud.
• The on-premises notification servers filter and schedule notifications to quality control inspectors.
Even though some of the servers in this example are physically located at a colocation facility, they are owned and managed by the company. Therefore, this example follows a private cloud deployment model.
Private cloud deployments supply cloud-like services using hardware and software owned and managed by the organization.
Customer Analytics
Figure 17-6 shows one final example of deploying cloud services.
Figure 17-6 Public Cloud Analytics Example
In this case, a company has decided to build an analytics dashboard showing sales information. This dashboard should be accessible to anyone with the correct security credentials (username and password) through the global Internet. The company’s CISO has reviewed the shared data and determined the risk of breaching the data available through the dashboard is manageable.
The application shown in Figure 17-6 contains the following:
• A data lake collects all company’s available information into a single unstructured database. This data lake may be implemented in a public or private cloud; the data source is unimportant.
• Information is drawn from the data lake into an analytics engine running on public cloud resources.
• Analysis results are transferred to a front-end server, and the data is deleted.
• The front-end server presents the data through a web interface.
• Users can reach the web interface through the global Internet.
Because large quantities of data are transferred into the cloud service, and small amounts of data are accessible through a web-based dashboard, this solution works with how cloud providers structure their service. Deleting the transferred data once the analysis is completed reduces the amount of data an attacker might access if they breach the cloud service.
This scenario is an example of a public cloud-native application.
A public cloud-native application is designed to run entirely in the cloud, where data enters the cloud but does not leave and is accessed through some web interface.