Ed Butler, CEO and founder at Amito, explores what organisations need to know when evaluating cloud provider efficiency.
As cloud hosting providers, we understand better than anyone why most organisations want to avoid migrating to a new provider.
Moving servers between data centres, migrating to a cloud infrastructure or a SaaS platform, the process is potentially risky – as a cost and distraction. Naturally, businesses will put off a second migration for at least three years.
No CTO wants to discover, a year into contract, that a performance gap or lack of sustainability means they need to move providers again. It’s why it’s so important to evaluate the efficiency of your prospective cloud provider or data centre operator before taking that all-important leap.
Here’s the thing: people and power are the two biggest costs for data centre and cloud deployment and will represent a major proportion of what you’re paying for.
For the service price to be both competitive and sustainable, how can you ensure that these costs are consistently and transparently controlled? Check these five key areas next time you’re looking into your cloud provider’s efficiencies.
Understand the impact of power efficiency
Colocation customers will usually focus on the power costs of a data centre as this is an obvious part of their monthly bill. But power efficiency matters as much when it comes to costs of IaaS, SaaS and PaaS environments too. Fluctuating costs are inevitably passed directly onto customers.
Establishing your provider’s PUE (Power Usage Effectiveness) ratio will give you a handle on this , describing just how efficiently a data centre uses energy. Those operating at a higher PUE are less efficient and more exposed to fluctuations in power costs. PUE is calculated by how much energy is needed to cool and provide resilience for a server, per one unit of power used.
If we take a data centre with a PUE of 1.8 for example, we’ll see 0.8 units of power are expended for every one unit used – it’s a typical figure for a five to 10 year old data centre.
A business in this scenario would be paying £578pm for a 4kW rack. Boosting efficiency to 1.5 would lower costs £480pm, until you reach a fully optimised data centre, with features including indirect cooling, a double conversion UPS and full generator pre-heating.
This will be in the 1.1 – 1.2 PUE region, averaging just over £350pm, half that of an inefficient data centre (a PUE of 2.2 equates to £700pm for a 4kW rack).
People and why automation matters
Efficient data centres automate wherever possible, particularly routine jobs. This reduces people costs, improves service delivery and the ready availability of expert advice.
Too many data centres still rely on their people to walk around 100,000 square foot facilities, gathering data on power usage per rack.
Or, they’ll operate their access control systems by printing huge spreadsheets, manually programming security passes when clients come on-site. Each of these activities ends up on the customer invoice.
An efficient operator prioritises tasking people with service and support roles, automating routine jobs. This releases teams to focus on doing the things humans do best –providing expert advice rooted in an understanding of customers’ businesses, responding quickly to specific requests and proactively addressing issues before they impact the service.
Focus on how your partner manages your infrastructure
Centralised control portals take care of a plethora of functions, from measuring and tracking performance across all clouds and the data centre, through to controlling the cooling of the data centre.
Used by both the customer and provider, they can centralise and simplify all tasks, requests and communications, delivering a potential 30% saving in human resources costs.
They can also streamline the client experience, with both an admin and customer interface into the system, allowing 24-hour access from anywhere, and any device. At no time was this benefit more apparent than during lockdown, when teams couldn’t access their kits.
We saw IT managers having to make hard decisions about how to fix servers, manage upgrades and trouble shoot. Businesses with centralised control portals were, on the other hand, able to operate almost seamlessly.
Consider your data centre’s scalability
Another lesson from Covid-19 was how some cloud providers held back resource for clients looking to deploy to new zones or locations as a result of the crisis. On occasion, we had clients come to us, unable to scale their cloud adoption quick enough to keep up with business needs, hampering growth at a stressful time.
Scalability will figure higher up in decision-making post-Covid-19, becoming as essential as compliance and cost-benefits.
The crisis has been a lesson in staying ahead of the curve when it comes to cloud deployment. Having the option to ramp up capacity can be the difference in being able to run your business – or not. And of course, efficient data centres will offer the flexibility to make this possible, affordably.
Make sure resilience is a given
You want your data centre’s technology to be set up to tolerate different failure modes without impacting the IT load.
Efficient data centres build infrastructure innovation into their strategies. You see this especially with new UPS (uninterruptible power supply) technologies. Alongside energy efficiency, UPS technologies offer crucial resilience. They can tolerate more components failing without impacting the IT load, delivering on uptime and minimised disruptions to your business.
Your organisation’s performance and growth are tied directly to your provider’s efficiency. Poor efficiency exposes you to unstable and inconsistent pricing models and, uncompetitive upgrade pricing.
Efficient data centres, on the other hand, will reduce running costs, put you in control of your infrastructure and deliver a more proactive service, as their teams aren’t consumed in admin and mundane tasks.
Ultimately, this saves you from worrying about an expensive, risky and distracting migration, helping make your business the success it can be.