What is Total Cost of Ownership?
The Total Cost of Ownership, or TCO, is an indicative figure used to quantify the financial outlay of a new product or system.
The TCO Calculator is the AWS tool which helps customers to compare their on-premises expenditure with that of a like-for-like AWS-hosted equivalent. Assumptions are made on numerous elements of a deployment at a high-level such as a three-year hardware lifespan and licensing which covers that period. Even at that early point, the figures produced can render subtle reactions from our customers such as raised eyebrows to more overt, audible reactions along the lines of “Why haven’t we thought about this sooner?”.
It can be tempting to think of the overarching estimated percentage of savings that the AWS TCO Calculator produces as being indicative of the final savings, but there are indirect costs associated with traditional infrastructure which can be cast aside or greatly reduced. There are also less obvious benefits such as greater agility (which we’ll touch on later) that you may be able to financially quantify as a result of some harsh lessons from emergency upgrades in the past.
The TCO calculator figure could be seen as the tip of the iceberg. Let’s have a look at the most common indirect costs associated with a traditional infrastructure and see how an AWS based solution can reduce or alleviate them.
It may be difficult to quantify exactly how much power an existing physical infrastructure uses as power consumption may be paid for from a facilities budget and may be indiscriminate. A move to AWS removes concerns over adequate power provision, cooling, UPS, with no need to over-provision when scaling up or scaling out. With a typical server adding up to £2000 per year to the electricity bill, this is a significant indirect cost.1
A move to cloud-based infrastructure would remove the need to rent or buy physical space to house physical IT infrastructure. Existing server rooms could even be repurposed for other business functions. No need either to continue paying rack space rental to a third party. Even bigger savings if the company IT infrastructure is collocated globally. There may also be some form of datacentre taxation applied to an existing rental cost, which can vary from country to country.
Since AWS takes care of the hardware, there is no need for extended maintenance contracts through hardware vendors or third parties, which can easily run into a 5 or 6 figure sum and, in some cases, won’t even cover replacement parts required due to wear and tear. There would be a reduced need for hardware support and existing employees can be retrained to work in cloud systems administration or with no need to back-fill roles requiring hardware maintenance skills.
It’s also worth bearing in mind that even less human resource is required to perform cloud-based infrastructure upgrades or changes. An AWS EC2 instance type upgrade can take the amount of time it would take for a reboot and a database replication can be carried out through just a couple of mouse clicks.
With the removal of your enterprise-level physical hardware, another knock-on effect would be no door access systems to server rooms or access controls to maintain. There’d be no specific workplace safety standards relating to electronic equipment rooms nor the need to devote time to shadow maintenance visitors. Video surveillance may not be necessary any longer (at least not covering a server room).
There would also be cost savings where hardware would have to be shipped to sites. New cloud-based infrastructure can be delivered relatively instantly. No more 60-90 day lead-times through distributors which are commonplace in hardware procurement through distributor channels. This would also bring an end to import/export taxes on infrastructure hardware at time of transit.
While we touched upon UPS provision whilst covering factors regarding power, there are other smaller hardware considerations to bear in mind, such as server or comms racks, consoles, KVM units and local high bandwidth cabling such as fibre channel or HBA. Some or all of these are usually bundled together with purchasing through distribution channels, but they are worthy of consideration and scrutiny.
Disaster Recovery / Business Continuity
At the turn of the century we would have talked about magnetic backup tapes in vaults or mirrored service architecture. More recently the conversation has been about infrastructure replication services between offices and VPN enabled remote working.
The overall goal has been to lower business continuity expenses or to build them into overall expenditure. However, with a move to an AWS cloud based infrastructure this cost can be almost eradicated. AWS performs replication between sites and they have taken great strides to promote their resilience. However, for those who are extremely risk averse, the ability to replicate worldwide would still cost a small fraction of having a physical replication node at each location.
There are also factors which are worth considering which don’t fall into direct or indirect costs, but should perhaps be taken into consideration using some historical context.
If, as an engineer or systems architect, you have endured a dramatic rise or fall in demand for system resources and wished that a significant part of the infrastructure be changed or upgraded as a matter of urgency, you could advocate taking agility into consideration when looking at TCO figures. You may be able to cite cases within the past few years where demand and resources were not aligned and significant additional expenditure was required for emergency measures.
Adopting a cloud-based infrastructure allows the provision scaling to be reviewed as often as you wish. Under provisioned? Just scale up the instance type. Development sprint finished? Shut down the instance. Project over? Terminate all associated assets. No pressure to recoup some of the CAPEX on the reconditioned hardware market or for old hardware to be a factor of any note in ROI figures. Purchasing from OPEX, and comparing it with the old CAPEX days, can give you real bargaining power if proposing a significant up-scaling.
Your Cloud-based Roadmap
Assuming that we at CirrusHQ have made a compelling case for you to migrate to a cloud-based infrastructure and you have taken flight into the cloud, what’s next?
Your previous estate may have been architected in an era where physical servers were ubiquitous and even virtualisation just resulted in the last hardware infrastructure refresh coming in fewer cardboard boxes. If that’s the extent of any progress you’ve made over the past few years, perhaps it’s time to allow us to bring you up-to-date with the latest prevailing technologies.
The initial tendency of cloud computing adopters is to favour on-demand instances, which offer greatest flexibility but not best price. However, by placing your production servers on reserved instances, a further saving of up to 72%2 over on-demand is possible.
Batch processing is also an area where immediate savings can be made. Taking some time to isolate and remove batch operations from always-on server instances can bring practical and cost saving benefits. Daily, non-time-sensitive batch processing can be run on EC2 Spot Instances at a time of AWS’ choosing within your defined set time range and subject to AWS’ spare capacity. In effect, they slip the job in when it’s quiet and you can save up to 90%3 over on-demand. Once fine tuned, it can be a good way to get your hands on some big cloud muscle for short periods whilst still making impressive cost savings.
Going further, modern technologies which have flourished in this age of cloud computing, such as serverless computing, microservices and containerisation can form part of a modern, architected solution which you might want to think about placing on your roadmap. We’re here to help you make that next leap too.
Let CirrusHQ help you get a good run up to the TCO hurdle. Feel free to drop us a line at firstname.lastname@example.org or via our contact page.
1 Calculated using the UK Average rate of 18.54p/kWh for a server with maximum power consumption of 1200W