A Match Made in the Cloud: CloudSigma’s Ampere Adventure

In 2008, I sat down with my friend Patrick Baillie at a bar in Zurich. We had studied Economics at Cambridge together, and had moved in different directions: I had become an entrepreneur, and he was working in IT at Credit Suisse. Some people might discuss politics or sports after a few beers and Jack and Cokes. We talked about virtualization and the potential of cloud-based services.

It’s amazing to see an idea leave your head and begin to take shape. We outlined our guiding principles on a sheet of paper and founded CloudSigma in 2010.

When we launched CloudSigma, the cloud was still in its infancy. VMware and some smaller players were making a name for themselves, and AWS had just launched their cloud offering. But I knew a big thing when I saw it. We realized there was potential in a customer-defined, customer-centric, European-based cloud service. Instead of creating a proprietary cloud environment and asking customers to modify their systems to work on it, we made it as neutral and flexible as possible.

Today, CloudSigma is a global federation of service providers and partners with 25 cloud locations in London, Tokyo, Mumbai, and Washington, DC. Our customers have evolved, too, growing more sophisticated in their understanding of cloud technology. Every change we make is for our customers’ benefit, and a recent change put us ahead of the pack—opening new avenues of business for us and new possibilities for those we serve.

A Flexible Cloud Experience

Customers of Amazon, Azure, and other public cloud providers must adapt their systems to fit with those proprietary environments. That’s fine if they’re happy, but there’s a lot of friction if they want to change providers or move their applications on-prem because they must re-engineer and redesign their computing resources.

Instead of creating a cloud platform with a lot of proprietary technology, CloudSigma adopted a better model from the start. We gave our customers the building blocks to design what their cloud would look like and easily lift and shift their environments to our platform. To do this, we built a flexible cloud that operates like a utility. Our customers don’t buy server or storage space, instead paying for aggregate CPU, RAM, and storage use. Our billing system is transparent and fair, offering freedom and flexibility that puts them in control.

When we launched, we were strictly a cloud infrastructure provider, but over time, we evolved a partnership model and started offering our customers cloud-as-a-service. This fully managed end-to-end solution can be white labeled by service providers who can resell high-quality, competitively priced cloud services to their clients out of the box. Within three months, we can architect and deploy a fixed-cost cloud solution and regional network, while also allowing these providers access to our global network no matter where they are. Doing this allows us to compete with Amazon, Google, Microsoft, and other major players who offer global connectivity.

Our role in these partnerships requires us to investigate, research, and build new solutions for our customers. We are constantly looking for cutting-edge technologies that can serve as foundations for new products and services, and we work with our vendors to incorporate their latest innovations into our designs. When HPE presented us with their next-generation servers built around sustainable cloud-native Ampere Altra 80-core and Altra Max 128-core processors, we jumped at the chance to test this new technology with typical workloads.

However, we don’t redesign our systems for the sake of it and we don’t add hardware and applications just because they’re shiny and new. Everything we do is to meet our customers’ increasing demands, and let’s be honest, customers rarely demand less. They’re always asking for new services, and we must find ways to meet their changing expectations. It was our job to determine whether this technology would do that. 

Becoming Alpha Testers for Ampere Technology

As our customers have grown with cloud technology, they’ve developed increasingly specific uses that we wanted to address:

• They have started asking for solutions-as-a-service, preferring that we manage, monitor, and secure their clouds instead of assigning these duties to their internal IT teams. They increasingly want us to handle every aspect of their cloud implementation, not just infrastructure.
• We are continually looking at the carbon footprint and environmental impact of our products and services because our European customers will soon be required to report on their energy usage. We are building a green ledger that will measure the carbon impact equivalency of our customers’ cloud systems with full transparency into the power consumption of their computing resources and processes.

With these ideas in mind, we scheduled a meeting of minds with the folks at HPE and Ampere. They flew people from Houston to our testing site in Bulgaria, installed pre-release servers that use Ampere’s ARM-based processors, and spent the day showing us how they worked. These were early versions of what became known as HPE ProLiant RL300 Gen 11 servers.

We became alpha testers, paying close attention to whether Ampere’s cloud-first chips addressed our customers’ need for increased performance and lower power consumption. We tested some typical workloads to compare the performance of current mid-range industry standard server processors and Ampere Altra CPUs. The difference was staggering. The price per core of Ampere chips is 70% lower than the industry standard processors, and you get a minimum of 64 cores with Ampere Altra CPUs versus 20 with the industry standard processors.  

ARM-based CPUs also use less power, which is a crucial concern for European customers who need to comply with incoming green legislation and where electricity costs have risen by about 200% in the last two years. The Ampere Altra chips make sense on the CapEx, OpEx, performance, and environmental fronts, providing everything we need to meet our customers’ increasingly sophisticated cloud computing requirements at a lower cost. 

A Smooth Transition to Next-Gen Servers 

We deployed HPE ProLiant RL300 Gen 11 servers, with Ampere Altra processors, internally to create object storage and caching services and new cloud products we can sell as managed solutions. We also offer our customers the option to migrate their services to Ampere-based infrastructure to lower their delivery costs.

While storage and network costs remain the same, Ampere Altra chips substantially reduce electricity and computing expenses, especially for customers using fixed-cost services. 

CloudSigma is using HPE ProLiant RL300 Gen 11 servers as the foundation of our future products and services. We have already transitioned some of our bigger customers to the platform and expect to move hundreds of smaller accounts to this technology as they adopt these new services. The migration is relatively straightforward on the infrastructure side, but some customers have to recompile their Intel-based apps for ARM-based processors.

That said, hundreds of open-source projects are already running on Ampere Altra CPUs, so they can easily migrate to the new architecture. I expect more of our customers to adopt ARM-based computing because so many of their apps have already been validated on this new architecture and can move over as-is. ARM-based processors represent a seismic shift for cloud services providers because they have little impact on our customers’ behavior but significantly lower their costs and boost their performance. 

A Seismic Shift for Cloud Services Providers

HPE ProLiant RL300 Gen 11 servers and Ampere Altra chips scale more linearly and predictably at a lower cost per core, but our customers have to rethink their approach to CPU technology. ARM-based chips are based on mobile processors, which were previously less capable than their desktop and server-grade counterparts. As tablets and mobile phones became consumers’ primary computing devices, these processors became more powerful. And because they’re already optimized for low energy consumption, they’re perfect for energy-efficient data center and cloud operations. Considering that people are shooting and editing feature-length films on tablets and smartphones, which are processor-intensive activities, it makes sense to scale mobile technology to enterprise applications.

Although we are still transitioning our cloud products to HPE ProLiant RL300 Gen 11 servers and Ampere Altra CPUs, we are already eyeing new applications for this exciting new architecture. We are especially interested in using ARM-based processing for AI inferencing to run live data through a trained model. This application previously required pricy, energy-intensive GPUs, but now runs on cost-effective, low-energy chips like the Ampere Altra and Apple M families of processors.

My co-founder recently found the sketch of CloudSigma’s original guiding principles. It’s now framed and hangs in our office as a reminder of where we started. We never could’ve imagined how technology like this would open doors for more customers, more services, and more business worldwide.