Before I get down to discussing bare metal servers, virtualization, virtual machines, containers, and related technologies, it’s important to have a fundamental understanding of the different layers involved in the cloud computing infrastructure stack.

We’ll begin this lesson by having a look at the cloud infrastructure diagram. Then, we’ll discuss each layer of the stack step by step.

Illustration 1.41 - Cloud Computing Infrastructure Stack

Physical infrastructure

Starting from the bottom, the base layer of the cloud computing stack consists of the physical infrastructure. This includes the compute servers mounted in a server rack, physical storage consisting of the hard drives and Flash SSDs, the RAM attached to the compute servers, and the data center network that connects all the physical infrastructure.

Platform software

Above the physical infrastructure layer lies the platform software layer composed of device drivers, firmware, operating system, and so on. Platform software manages the physical hardware and provides an abstraction layer for the distributed systems software to interact with the hardware.

Cluster infrastructure

The cluster infrastructure layer runs on top of the platform software layer. This consists of distributed systems that manage the nodes and are responsible for cluster management.

They facilitate operations such as message passing between the nodes running in a cluster, distributed node coordination, cluster synchronization, and reaching a consensus at the cluster level. Some examples of distributed-system software are Apache Zookeeper and Google File System; in addition, some of the distributed storage systems include Google Bigtable, Hadoop, and Google Spanner.

Just like an OS manages a physical server, cluster-level software manages groups of server nodes. We’ve already discussed this in the clustering chapter.

The cluster infrastructure layer provides an abstraction to the complexity of distributed systems via APIs. The application programmer does not really have to worry about the internals of how data is dynamically partitioned across several nodes in the cluster, how complex memory operations are executed, how the system ensures fault tolerance, and so on.

Cluster-level software also takes care of cluster resource management, like distributing tasks to the less occupied server nodes to keep a balance amongst the different nodes in the cluster.

Kubernetes is one good example of a cluster resource management software that tracks resource consumption across the container cluster. More on Kubernetes and containers in the upcoming lessons.

Application software

The application software layer contains the applications that we build using the APIs of the cluster-level software. These applications are our business applications like social networks, fintech apps, food delivery systems, online multiplayer games, and so on.

Distributed systems enable us to run and scale our applications across thousands of nodes running in several different clusters. Running a service in a distributed environment ensures scalability and high availability.

System monitoring

To ensure good system health and continual uptime, we need to continually monitor all the layers of the computing stack. As you see in the diagram, the system monitoring layer is a vertical layer that applies to all the layers of the cloud computing stack.

Monitoring the system, including keeping track of the server uptime, performance, latency, throughput and resource consumption helps in weeding out bugs, getting rid of system bottlenecks, optimizing the resource consumption, setting up alerts, making the code efficient and so on.

All these things are vital in building highly-performant software. All the stats are ideally monitored with the help of a web-based dashboard.

Grafana, an open-source monitoring tool, is one good example of this. Google Cloud Stackdriver monitoring is another example of a cloud-based monitoring tool. More on monitoring in detail in the upcoming lessons.

In the next lesson, we’ll unpack multitenancy with regards to application hosting.