Back to Chapters

Cost comparison: cloud vs. private infrastructure

Cost considerations for cloud and private infrastructure

The costs of Cloud infrastructure

Low freight rates have brought cost management to the centre stage for the shipping industry. Yet, one often overlooked area of high expenditure is internal IT systems – and cloud may be a potential solution. In this post we will outline the major cost factors between self-managed IT and cloud alternatives, and key questions to consider when running a through cost-benefit analysis of the two.

For clarity, this post will be entirely referring to public Cloud Infrastructure commonly called Infrastructure as a Service (IaaS). Software as a Service (SaaS) companies often build on top of IaaS. If you’re unclear about some of this key terminology, or even what cloud really is, read our first post ‘What is the Cloud

Cloud IaaS SaaS

The Total Cost of Ownership (TCO) model can be used as a foundation for considering costs. Procurement of hardware (i.e. servers) vs Cloud subscription costs tend to be only the tip of the iceberg – the costs that are clearly visible and accountable. The main issues with on-premise servers come from the many additional costs that are less tangible – deployment and configuration, customisation, long term support staff, maintenance and upgrades.

What set up & replacement costs are there to consider?

When using private servers, the cost of both purchasing and storing the physical servers can be substantial. When using Cloud infrastructure, you’re essentially renting the use of servers as opposed to buying them. Whilst there can be substantial initial implementation fees with Cloud providers, remember to factor in replacement costs for private servers - they’ll need replacing every 3-5 years. With a longer-term outlook, the replacement cycle and falling performance increases the cost disparity between the two as time goes on.

It’s also worth mentioning that intense competition among Cloud providers has led to a phenomenon coined the ‘Cloud Price War’. This has prompted a 66% drop in entry prices from 2013-2016, according to Tariff Consulting. If Cloud infrastructure was considered an option a few years ago and deemed too costly, now may be the time to review.

What ongoing costs should be considered?

On premise servers need maintenance and, when issues arise, skilled professionals on the scene. Hardware failure can be caused by a multitude of reasons. As the servers reach the end of their lifespan problems occur more frequently – certain components malfunction or overheat. Other causes may be totally unpredictable i.e. a resident rodent chewing through a cord, or a power surge (we all saw what happened to British Airways).

Besides the cost of relying on someone to be around to handle these issues, the inability to access the system (server downtime) can have serious ramifications that don’t appear on the balance sheet.

In the fast-paced world of shipping, a minute lost can be the difference between a million-dollar deal or a ballast leg. Consider the fact that server downtime may stretch into hours – the opportunity cost can be huge. Cloud infrastructure providers are contractually responsible for maintaining uptime (the technical term for when you can access your data – the servers are ‘operational’. This is the opposite of downtime). Hardware failure may still occur, but it is not the customer’s problem, data is immediately moved to another server without a noticeable transition.

Downtime isn’t only caused by hardware failure, though. Operating system updates and upgrades require the servers to be rebooted. Those of you that are technically minded will know that ’rolling updates’ can be configured to continuously update servers without disturbance to the end-user, but require multiple servers running behind a load balancer. However, if done on-premise, this bumps up costs significantly. Within a cloud computing environment this is a default part of the service. Processes that may have taken hours to handle internally are no longer the customers problem, or cost.

A final, often ignored, operational expense is energy costs. Servers need to be in operation 24/7, are power-hungry and require constantly cooling.

What cost savings are there to consider?

Server capacity is the technical term for how much room there is for the server to manage the data and tasks users need – this may include storage of data or the ability to run computations. Server capacity is usually set at a level to cope with the maximum demand placed upon the systems, but how much of that capacity is actually used on a consistent basis is another concern.

With self-managed IT servers companies pay a fixed cost whether using it or not. According to McKinsey & Co “utilisation of servers in business data centres rarely exceeds 6%”, in other words these servers deliver no more than six percent of their maximum computing output on average over the course of the year. The elastic nature of cloud computing (i.e. easy to scale to match demand) can equip shipping companies with better control of resources across the globe, redistributing, ramping up and ramping down computing power and storage when necessary. Cloud gives you the ability to ‘pay as you go’. Gartner has long suggested that 80% of IT budgets get spent simply ‘keeping the lights on’. With an eye on operational efficiency, Cloud delivers the opportunity to only pay for what is in use.

What strategic company-level considerations should be taken into account?

Every shipping company is unique in culture and strategy. Some shipping companies consider internal IT capabilities a core strength and value driver. Others may already have the ability and scale to achieve IT scale efficiencies. However, even the biggest companies who place significant value on IT may choose to take a hybrid approach – looking item by item at which parts of the technology ‘stack’ should be managed in- house.

Taking a case study from the technology sector, WhatsApp was one of IBM Cloud’s largest clients. When Facebook acquired WhatsApp, outsourcing to the cloud no longer made financial sense – Facebook were more than capable of hosting WhatsApp on their own servers and had the internal IT knowledge, and size, to cost effectively manage them. Facebook’s on-premise server capacity is currently 300 petabytes of data – for shippers, this is the equivalent to approximately 15 trillion circular e-mails, or 4 trillion position list Excel files. Facebook are a technology business with 17,000 employees, most companies don’t come close to the capacity requirements, or core competencies, that make on-premise servers scale efficient in the same way.

For a number-crunching comparison, tailored to your specific situation try some of the TCO calculators for AWS or Microsoft Azure.

So, is the IaaS Cloud cheaper than on-premise?

Generally speaking, for a non-IT company, using a Cloud provider for infrastructure will be cheaper. This is particularly true with a longer-term outlook whereby utilisation, ongoing maintenance and replacement is factored in. There’s no longer a need for shipping companies to invest heavily in non-core assets.

Cloud Shipping Infrastructure

In the next post, we consider the costs and benefits of SaaS vs. desktop software vs. in-house developments. This is another cost area which is often deemed of strategic importance and can differ from company to company.

Cloud-based SaaS with Shipamax

Shipamax’s cloud-based platform is used by teams to work together and help keep track of the end-to-end process before it gets handed into your desktop ERP. Shipamax is built on top of Amazon’s AWS IaaS.