The Shift to Public Cloud

There are many approaches to building infrastructure and designing systems. There is no silver bullet for how things should be architected. Depending on the use case, how the data can be accessed, and the technical requirements, it may make more sense to build a stateless application on serverless systems or to run a database on bare metal. As we work through understanding and contrasting the differences in these architectures, it’s important to also provide examples of when certain approaches make more sense. Companies have been transitioning from on-premises, private cloud infrastructure to cloud-native, public cloud infrastructure, but this doesn’t mean that everything should be stateless and serverless. Depending on the architecture of current systems, what performance thresholds are required relative to cost constraints, and other factors, ultimately, customers will make the decision based on individual workload needs and the business value...

The value of adopting public cloud

At first glance, running a private cloud on-premises is a great way to build technology infrastructure. The company would have ownership and control over the land, capacity, and power requirements to ensure that they can plan for and meet the needs of their future customers. You can forecast future capacity requirements, work with vendors to develop a procurement plan, and work with customers to commit to using the future capacity.

One of the common themes that comes up with customers who have taken this approach is the idea of control. They can control where infrastructure is deployed, what application is running on that infrastructure, who can access the data center, and what hardware is used to deploy that infrastructure. However, with control and choice comes complexity and cost. Although they can choose routers, switches, and servers, does that nuance truly matter to the business? At the end of the day, the business just needs the application...

The impact of cloud adoption on finance teams

Beyond reducing the amount of planning and capital required to build and operate technology companies, the shift from large capital expenditures to a consumption-based expenditure model also represents a big shift in how companies and finance departments think about technology costs.

Building and running private cloud environments is very costly. It requires massive amounts of capital and significant time to activate that capital. Therefore, over the past couple of decades, finance departments have understood that when it comes to procuring technology, you should expect large deal sizes and for the capital to be spent upfront. This meant that checks needed to be sent to all the different vendors to buy the different components of the data center before the data center would come online. There would be months if not years of lead time between when a company was able to purchase something and when they were able to use it.

Given this...

Private, public, or hybrid?

Now that we’ve explored the benefits and challenges related to private and public cloud environments, let’s discuss when each of those environments makes sense or when you may use both! The term hybrid cloud refers to an organization that leverages both private and public cloud environments. It is quite common to find organizations that are running hybrid cloud environments when they’ve been around for a while and have already made significant investments in building out their data centers. Rather than completely discounting historical investments, organizations may choose to slowly sunset their existing data centers based on existing contractual and hardware life cycle constraints. For example, if they refreshed all of the hardware in one of their data centers a year ago and the hardware still has 4 years before it becomes obsolete, that company may choose to migrate infrastructure from other data centers to the public cloud while retaining...

An introduction to networks

In this section, we’ll dig into networking basics such as terminology and architecture.

An IP address is a unique identifier that is assigned to any machine that is connecting to a network. This identifier helps ensure that traffic is routed correctly. For example, when you send a request to google.com, the web page needs to know who made that request to provide the answer to the correct machine. You can think of an IP address as a home address for your laptop, mobile device, or server. It designates where the request is coming from and where the response needs to go.

Every machine connected to the internet has an IP address, including web servers hosting websites. What we type into a browser and understand to be the domain of the website (www.google.com) is not used by machines to understand the destination. Rather, these human-legible names for sites get converted into IP addresses for the machine to understand where the request is coming...

Public cloud with Google Cloud

Google has built a global network of data centers that is connected by proprietary cables and networks. This means that although Google may work with or partner with ISPs in certain cases, all traffic traveling over the Google network is traveling outside of the public internet. This is unique to Google as it is the only hyperscale cloud provider to have built out this global network. The reason for this infrastructure is that to support the expectations of users for Google Search to be lightning fast, Google engineers decided to build their own global, private network. This prevents them from being at the mercy of latency, bandwidth, or security issues related to working with regional ISPs and ensures that they will have the networking capacity required to support future growth – even if local ISPs don’t meet Google’s technical requirements. The following figure highlights Google’s global network across the existing network...

Summary

The shift from private to public cloud was catalyzed by the needs of employees and customers. Whether it’s teams looking to be more productive or customers asking for better availability or new features, the public cloud is a great way to build infrastructure that’s highly secure, available, reliable, and performant. While the same technical benefits may be achieved or nearly replicated in private cloud environments, organizations would be hard-pressed to deliver the same quality of service at the same cost relative to public cloud.

Building data centers is no small feat and entire teams would be required for the planning and management of the physical components of the infrastructure. By specializing their engineering teams and focusing on building performant code and new features, companies can out-innovate their competition.

Understanding the shift to public cloud as well as networks as a whole provides us with the foundational knowledge upon which to...