Why is TCO so important?
High-Performance Computing for many years has always operated in the same manner, on-premise. HPC was a service where you bought hardware, stored that hardware on-site, and managed the service on your premises. People often think when looking at the cost of HPC it is only down to the cost of the machine, the “tin”. That all important capital expenditure is what people focus on.
However, with the progression of Cloud in the world of HPC, that all important debate of CAPEX vs OPEX remains a constant. Different people will have different views but what it does highlight is the need to understand your HPC solution in its entirety. What does entirety mean? Well you cannot just look at the expense of the tin and metal, the machine. There are significant overheads and additional costs which are paid for by your business to operate an on-premise machine. Whilst you may not see these costs in your PnL, they are very much there and need to be accountable.
Expenses such as: Electricity, Water Cooling, Maintenance, Facility Costs, Data Centre Costs, Insurance and Maintenance. These all need to be taken into account and included when looking at the Total Cost of Ownership of your on-premise HPC. What is also often left out when reviewing HPC is the commitment over time a solution could cost your business, such as:
The investments which may need to be made when looking at data centres or a data centre refresh.
The years of costs which you could be committed too.