Cloud Performance: Some Cloud Platforms Simply Don't Measure Up

If you aren’t familiar with the intricacies of cloud architecture, it’s easy to think of cloud platforms as interchangeable. After all, cloud storage and cloud compute are available from many different vendors. From the outside there doesn’t appear to be much to differentiate them. To steal from a great American writer: a cloud is a cloud is a cloud.

The reality is more complex: cloud platforms differ on a number of fronts, from support to availability and from scalability to performance.

Performance in particular can have a significant impact on the cloud experience. Depending on the trade-offs a cloud vendor deems acceptable, cloud platform performance can vary from extremely low latency to slow as mud. That doesn’t mean organizations shouldn’t choose the cloud for infrastructure hosting, just that they should be circumspect about which cloud platform they choose.

Choosing the wrong cloud vendor can have negative implications for a business’s projects, particularly when they depend on the lowest possible latencies. If a business is using a cloud platform to store backups, latency might not be much of a problem. But for user-facing applications, financial services, analytics and machine learning applications, and a host of other applications, performance really does make a difference. It makes a difference not just to the experience of an application’s users, but to the bottom line of the cloud customer.

A poorly performing cloud platform can eat IT budgets at an alarming rate. Clouds are great as scaling, but that doesn’t necessarily make scaling the ideal solution to every problem. If your web application is performing poorly, you can certainly throw more servers at the problem — they’re trivially easy to create. But every server costs money and the better the basic performance characteristics of the cloud platform, the less often customers have to scale to solve their performance problems.

Performance is a key issue for applications that rely on low-latency IO, including most user-facing web applications and eCommerce stores. A platform that has chosen to accept IO performance trade-offs simply can’t get data to where it needs to be fast enough. Whether that matters depends on the specifics of the application, but if IO performance matters to your application, your choice of cloud platform will have an impact on both end-user experience and cost.

The effects of poorly architected cloud platforms can be particularly vicious for startups encouraged to scale quickly and prioritize growth before profit. As we’ve already discussed, cloud platforms have no problem scaling, but many startups find their runway being rapidly consumed by infrastructure costs their investors balk at paying for.

Now, it’s possible to pay the biggest cloud vendors for better performance, for “fast lanes” that are capable of moving data as quickly as needed, but that massively inflates the cost and doesn’t solve the problem. The better option is to choose a cloud platform optimized for performance from the ground up.

If you want to talk to one of our infrastructure experts about a cloud platform optimized for performance, availability, and scalability, don’t hesitate to get in touch.