Recently, Platformer reported that Twitter has stopped paying its bills to both Google Cloud and AWS. While I’m not advocating for not paying your bills, this highlights a larger issue with cloud services. Companies like AWS, Azure, and GCP have pricing structures that make businesses pay for everything – every bit, every byte, every CPU cycle, and every transaction. The more successful and well-adopted a platform gets, the faster the costs rise.
Our team of seasoned platform engineers have managed critical, large-scale services for millions of users over the past two decades. Needless to say, we’ve been watching Twitter closely, admittedly with a morbid fascination in recent months.
Take, for example, the recent failure of the Twitter Spaces interview hosted by Musk, where not even tens of thousands of users could join the livestream. 20 minutes in, Musk gave up and decided to record the stream for his audience to watch later. Twitter could have become a “global town square,” serving as a real-time source for news. But not even a billionaire is immune to disruptive infrastructure and network issues.
Here at NetActuate, while we may not be billionaires, we do know what it takes to avoid performance issues like these. Every day, we have the privilege of collaborating with innovative customers to deliver their products and services to end users around the world. Our network is currently the third largest in the world in terms of peers, and is the go-to choice for platforms requiring seamless connectivity to their data centers, clouds, or edge services.
Every week, we help new clients with six-figure public cloud bills significantly reduce them with customized, alternative deployment models.
Hundreds of Ways to Drive Up Your Costs
Just how does a six-figure public cloud bill happen? The primary culprit is these providers’ vast and intricate service stack. A quick visit to the AWS console shows hundreds of services, each supported by dedicated product and engineering teams.
It doesn’t take long to realize how large an organization it takes to design, build, and then operationally deliver hundreds of highly available services that can be easily called by an API or SDK. By charging for each and every service – and making it so easy to (intentionally or unintentionally) use them – they are able to cover their costs and much, much more.
Go-to-Market Pressure Can Drive Cost Inefficiency
Modern application development and architecture, such as the adoption of Kubernetes and serverless computing, simplified and added abstraction layers to processes that once demanded extensive backend engineering. However, this convenience comes at a cost.
These distributed design patterns help products get to market faster, as they are easier and faster to build. But these patterns are often extremely cost inefficient.
One of our recent customers had 14 different transformations and APIs for every call an end user made to their app.
That’s 14 ways for their cloud provider to charge them for CPU cycles, bandwidth transfer, and more – for every. single. request. Once we got to the bottom of their existing exorbitant cloud bills, we learned that their full stack developers were heavily leveraging prebuilt tools and services to get them to market.
We partnered with them to re-architect and design their applications with a more traditional, monolithic approach. They reduced their application complexity to just three paths for each call, and brought all operations in-house. This restructuring let them leverage new features like custom hardware with GPU offload. The result? A win-win: significant performance improvements AND cost savings.
Twitter’s current financial challenges shed light on the broader issues surrounding cloud services and escalating costs. As a tech company, NetActuate strives to make sure every customer’s deployment is fully optimized to be both high-performing and cost-efficient. That’s why we like to say we help businesses scale without fear – as your platform adoption grows, you shouldn’t have to worry about exorbitant infrastructure costs impacting your bottom line.