Choosing a cloud provider with caution as SMB
These days, everyone is talking about the big three cloud providers, AWS, GCP, and Azure. These cloud providers offer a variety of services and products around the traditional compute layer one used to describe as a virtual machine or virtual servers. The undoubtedly advantage is, that all these services are supported natively, whether it is managed databases, notification gateways or high infrastructure environments like lambdas.
These advantages come with a higher median price than the actual computing is usually tagged for. George H. sampled and analyzed the price differences in the article “Is a billion dollar worth of server lying on the ground” .
The large cloud players vs traditional ISP providers
If the business model implies an unpredictable variable in the user load, the flexibility of these cloud providers is invaluable. Particularly for social media branded startups, which user base overnight tenfold and heavy compute is involved. Yet, these environments do not apply to all use cases. Most of the organizations have some typical up and downs in their traffic, but the average and median load are somewhat predictable given a typical margin of 20-40%. If that falls true, why bother to spend 160% or 250% more on your infrastructure to guarantee flexibility in potential scalability that doesn’t apply to your business?
Most small and medium-sized businesses (SMBs) trust rightward infrastructure providers like netcup and others. In the following, we want to briefly outline that most modern technology templates can be easily and natively implemented on their infrastructure as well. One of these modern frameworks are canary releases. Long running and stable, just for a fraction of the typical cost base.
Be aware of Vendor and Employee lock-in
One of most IT decision factors is the risk of lock-in. While most executives are aware of vendor lock-in strategies, a more subtle way is the employee lock-in. These days, hiring a DevOps specialist implies the skill-set of vendor specific knowledge. Like the AWS Cloud Practitioner certification or all similar learning paths for GCP, azure and AWS. Skills are much more narrow specialized for particular vendors, and not generic anymore on how deployment or DevOps tooling can be applied vendor-independently. Imagine, all of the IT staff is trained for a specific vendor, how hard will it be to migrate away or even assess alternative infrastructure options.
It became truly unique finding proper IT staff that understands fundamental concepts of virtualization techniques, handling IT infrastructure apart from UI-centric deployment of out-of-the-box solutions. This, in contrary, makes the triage of IT problems much harder.
Traditional ISP providers offer same flexibility as cloud providers
The art of releasing new feature versions to just a subset of users is known as canary release, a technique in modern software engineering that balances early user feedback and low risk profiles. In this way, the new feature can be testing before rolling them out on a large scale, yet with real users in a controlled environment. Randomly sampled users can enjoy new features, observe their experiences, and make adjustments upon convenience. Yet, in case of increasing failure or problems, the infrastructure automatically and immediately rolls back the designated releases to the stable version without impacting the broad audience. This combines a variety of tools, including A/B testing, early releasing, and user centered feedback.
Implementing such modern techniques doesn’t require sophisticated cloud providers or agencies. Still, it is as simple as traditional continuous integration and deployment (CICD) pipelines coupled with a loadbalancer and a few working nodes. Knowledgeable DevOps engineer knows how to leverage CICD SaaS services like TravisCI, CodeShip or Gitlab runners to offer continuous deployment. Wrapping these in docker and deploying them to Kubernetes is no magic. And no fancy cloud provider service is required that charges on top. These services allow easy governance policies for large enterprises, but are these necessary for SMBs?
The proof of the pudding is in the eating.
There is nothing better than trying it out yourself. For various of years, we stick around with the traditional ISP provider netcup. Over the past decade, we tried out many different vendors but netcup is the best provider in our eyes and offer absolutely fair prices.
From small servers to heavy compute machines, all are available. To kickstart you, a collection of a few discount codes are available for trial, and if appreciated, there is an additional 10% discount on all products if you like our article (just reach out directly).
 George H., Is a billion dollar worth of server lying on the ground, https://cerebralab.com/Is_a_billion-dollar_worth_of_server_lying_on_the_ground, Nov 2020