Articles

Who accounts for the economic and carbon cost of SaaS data centre power? And can it be brought under control?

Few consumers think about the electricity that powers their applications. But for the enterprise sector, whose reliance on SaaS is growing, this is not something that can be ignored.

That leaves many questions. How is the energy use and cost being priced into the services such as SaaS? Are SaaS providers paying too much for power?

SaaS providers are focused on cost per transaction; meaning this becomes a question for the data centre operator. Should SaaS providers demand energy cost transparency and that data centre power provision be more flexible and adaptable?

SaaS Power

The figures below reveal there is little doubt that SaaS is taking over the global enterprise software market. So, what are the power cost implications for SaaS players and their data centre partners?

At first glance SaaS, like its enterprise application perpetual licence forebears, may appear many layers above and removed from the physical equipment and energy needed to make it work.

The promise of SaaS (and cloud as a whole) is that it fulfills the desire of IT organisations to abstract their own service provision away from the infrastructure layer.

But if the enterprise is no longer managing the infrastructure then this falls to the SaaS provider to source - through building or buying – the data centres and power to deliver these applications.

The energy consumed by SaaS instances doesn’t usually surface as a consideration when selecting which SaaS platform to use. (At least not in the public domain). But as the world focuses on reducing energy consumption and GHGs this is likely to change.

How SaaS applications are powered has huge implications across the entire supply chain of utility companies, data centre providers and ultimately end users.

SaaS Eating the Enterprise World

The revenues in enterprise applications are expected to be around $450 billion in 2020 and the market is expected to reach nearly $500 billion in 2021.

https://www.statista.com/statistics/203428/total-enterprise-software-revenue-forecast/

If we unpack this, an increasing share of the enterprise application market is going to be SaaS based.

According to FinancesOnline much of the enterprise applications world is already SaaS based.

“As a key component of the cloud computing market, the global SaaS market size is estimated to end up at around $158.2 billion in 2020. With a CAGR of 11.7% in the next four-year period, the projected SaaS market size will be about $307.3 billion by the end of 2026. In 2020, it is estimated that the overall worldwide SaaS penetration rate will be at 36%.” In the collaboration application segment SaaS penetration is getting the lion’s share at 81% of the market, it says.

The biggest SaaS applications are collaboration, human capital management, CRM, ERP, BI, SCM and Content.

Any list of the world’s top SaaS companies usually places Salesforce at the top, even before it announced its intention to buy Slack for $28 billion. Microsoft is usually number two followed by firms such as Adobe, Box and Amazon Web Services (AWS) SaaS. From the traditional enterprise software world firms such as Oracle, Docusign and IBM software are not far behind along with firms like Servicenow and Workday.

What every SaaS provider has in common is knowing how to sweat their assets.

Where SaaS Resides

The big SaaS companies can be split between those that own and operate data centres, Microsoft, AWS and others that buy data centre services. SaaS companies generally run a hybrid mix of data centres from the commercial colo space and cloud providers.

The delivery model is based on redundancy. Salesforce says: “Customer success drives our data center strategy and delivering the highest standard in availability, performance, and security is our top priority. To that end, we build and serve each Salesforce instance from two geographically diverse data centers to have availability our customers have come to expect from us. At any given time, your Salesforce instance is actively served from one location with transactions replicated in near real-time to a completely redundant, secondary location. We regularly site switch between the locations for maintenance, compliance, and disaster recovery purposes. As we continue to expand and improve our global infrastructure presence, we recommend customers build their applications free of specific data center requirements to support a seamless Salesforce experience.”

Salesforce-managed data centers [operate]…in: Chicago, Illinois, United States (USA); Dallas, Texas, United States (USA); Frankfurt, Germany (DE); Kobe, Japan (JPN); London, United Kingdom (UK), London North, London West; Paris, France (FRA); Phoenix, Arizona, United States (USA); Tokyo, Japan (JPN); Washington DC, United States (USA), Washington DC North, Washington DC South.

It adds: “In addition, we have instances served from AWS Cloud infrastructure in the United States, Canada, and Australia. These instances are located in two or more separate Availability Zones within each respective country.”

So, the model is clear. SaaS provision operates on redundancy from paired data centres. This is true for every enterprise SaaS player.

Why Power Costs Matter

All very interesting but what has this to do with the price of butter?

Price per transaction is a priority for SaaS companies. The SaaS business model demands providers look to drive out costs and get the greatest ROI from their assets wherever and however possible.

But those SaaS companies which rely upon colocation data centres don’t have control over the power element of their cost base, because the way data centre power topologies have been designed dictates how power is provisioned. They are fixed and wasteful. As SaaS continues to scale, power costs as a percentage of cost of goods can only rise. If SaaS providers are paying too much for power or having to pay for power they don’t use then that becomes an additional cost to be passed on to the end user.

The Answer lies in “Power as a Service” 

Here are some questions:

1. Will enterprise customers be forced by regulators to account for the energy and carbon cost of their SaaS applications?
2. As customers of commercial data centre operators, will SaaS providers demand access to flexible and adaptable power?
3. How can data centre operators respond?
4. Can inflexible power provision in data centres be addressed without ripping and replacing the existing infrastructure?

To address these questions, indeed to answer many other questions which are starting to surface about cloud provision of all kinds, ‘Power as a service’ (PaaS) needs to become an available offering from service providers. Fortunately, technology already exists which can enable the sorts of fixed data centre power systems which are ubiquitous to fulfil the adaptable and redundant power requirements of SaaS applications.    

Adaptable Redundant Power (ARP) from i₃ Solutions Group, is one such solution designed to address many of the flexibility requirements to make PaaS a reality for SaaS environments. By making energy use more economical and flexible, ARP technology brings agility to data centres by accessing trapped power and reconfiguring the power system topology to provide granularity to match SaaS service levels. The obvious benefits are power resilience better aligned with SLAs, more efficient utilization and the reduction of needless emissions.

In the end all costs of production and delivery are ultimately charges the customer must bear. Unlike consumers, enterprises cannot ignore the cost of electricity that powers their applications. At the same time, SaaS market players will likely face more regulatory and investor scrutiny of their power use and carbon footprint. Neither they nor their data centre partners can afford to risk letting the cost of power spin out of control. Smart commercial data centre operators who are hosting, or wish to host, large SaaS customers can add value to their offering by providing PaaS based on technologies such as ARP.