Why cloud makes sense

step one to a cloud engineering job

howdy avatar

i am going to teach you how to get a cloud engineering job. The first step for that is to understand what advantages cloud provides vs the previous paradigm of ‘on-premise’ datacenters.

COST STRUCTURE

For on-prem investments into equipment, or PP&E as our accountant friends like to say, is an investment into servers, wiring, high powered cooling, and etc. Specifically the servers and other electronics are sunk costs, capital expenditures (CapEx), and fixed costs. These kind of costs make our accounting friends sad because they are fixed cost investments into depreciating assets. On top of that the firm then has to hire and retain labor, networking specialists, reliability specialists, cooling and water specialist. The list goes on. Summary: fixed cost capital expenditure and labor.

Cloud on the other hand is set up as pay as you go, or better, pay as you scale. Why is this better? This is a variable cost based on USAGE. It is already more optimal than buying several servers and over or under estimating on demand. The utilization can be more granular, you are not paying for excess or idle compute power. This may seem obvious, but it was revolutionary at the time. Having many customers in a single data center then creates economies of scale for the provider, as they then want to minimize idle resources and can offer competitive pricing. Summary: variable cost that is pay as you scale, better utilization of resources for provider and customer.

SCALING

On-prem means one needs to accurately estimate the compute power needed, purchase the required servers, and then purchase extra servers to be idle to handle spikes in traffic. Why is that an issue? The firm then has unused compute power not being used to make profit. What if the firm needed to scale rapidly? Buy more servers. Shipping + installation + network configuration + application setup = the opportunity to scale probably passed you by. Startups could effectively miss their viral window to take off because they did not have enough servers. Big companies could crash during holiday season. Summary: Over-Purchase compute to be ready to scale, beyond that order more compute.

Modern cloud providers have enabled customers to scale with simple button clicks or even automated processes. Compare that to on-prem paradigm. Customers can scale up or down automatically (future substack post?), or with simple clicks. The ease of scaling empowers customers to be able to granularly purchase enough compute and quickly and affordably scale at a moments notice. Revolutionary compared to ordering a server and paying expensive labor to install and configure the rig. Cloud computing played a pivotal role in the ability for startups to scale. Startups can focus on their core competency and let the cloud provider focus on the infrastructure and interfaces. Overall, remember the scaling is virtualized Summary: Scaling is automated or simply done with a couple clicks.

DISASTER RECOVERY

Aside: Distaster recovery is having infrastructure seperated from the main stack in order to have business continuity in the case that a major issue occurs on your main stack. Eg: having compute in California when the main stack is in Virginia.

For the on-premise paradigm one would have to open or lease a data center in a region isolated from your main compute resources. Expensive, annoying, and more overhead. If you cannot see why this is worse than the cloud section, read again. Summary: hard in the on-premise paradigm

Cloud service providers have the advantage of creating data centers all over the world. As a customer you can take advantage of this by being able to failover to almost anywhere in the world in a moments notice. Depending how the resources are set up, failover could be near real time to anywhere in the world. This section is pretty simple. Summary: Cloud wins.

SECURITY

On-premise much work goes into physical security as well as the regular cyber security. Controls for access, operational security, etc. In terms of cyber, many firms fail just due to the amount of labor and diligence that goes into creating a secure infrastructure. Everything must be built or an expensive external tool. Its almost impossible to hit on every part of security that needs to be addressed. But when the firm has to build tools, audit, and verify compliance to NIST SOX, etc, it adds up. Caveat: I am aware of some instances where government entities or firms that deal with certain data prefer or are required to use on-prem due security. With the most recent launching of AWS Top Secret, it seems like this number might shrink soon. Summary: expensive, overhead, and diligence.

Cloud gives several advantages here. The engineers at the cloud service providers are some of the top minds when it comes to cryptography and other security subjects. They have built tools to be hardened and robust. And built them in such a way that firms should be able to maintain a strong security compliance. All the tools you would need are already build by the provider, and they’re built well. The customer just needs to purchase and configure them according to the documentation. The chance for you to mess something up that bad is low. Summary: Well built tools, pay as you go, configure instead of build tools.

Wrapping things up

hopefully at this point you are drawing some conclusions. I am sort of rambling on, but hopefully you can see that leveraging a cloud provider has many advantages for most firms. I would be somewhat skeptical if someone said they were making their own data center for their startup. Does not make a lot of sense.

Disclaimer: None of this is to be deemed legal or financial advice of any kind. These are opinions written by a celt avatar