The purpose of the Kazuhm Calculator is to illustrate the potential savings that could be achieved by migrating workloads from the cloud to Kazuhm when deployed on existing private infrastructure.
How it works
After entering a given set of inputs, the calculator then runs a linear optimization function that finds the lowest-cost combination of compute-optimized reserved instances on AWS and Google Cloud that produces an equivalent amount of compute power of the desktops and servers specified. The resulting output is an estimate on the potential savings that could be expected from migrating workloads from AWS and Google Cloud Platform to Kazuhm.
- Prices used to compute expected cloud expenditure are from publicly available prices.
- For a better comparison to the Kazuhm pricing model, reserved instance prices from AWS and Google Cloud Platform are used.
- Electricity price is $0.12 kWh.
- The benchmark instances/VMs are from the C5 instance family, the current generation of compute-optimized instances from AWS.
- For a better comparison between the two cloud providers, custom GCP instances were created to match the specifications (vCPU count and memory) of C5 instance family from AWS.
- Provisioning of instances/VMs are the only cloud-related cost considered in the calculation. Usage of other services like storage or networking in the cloud were excluded from the expected cloud expenditure calculation.
- The benchmark reserve instance plan used was the 1-year no-upfront plan from AWS and Google Cloud Platform.
- Calculation of electricity consumption at different levels of usage was derived from the United States Data Center Energy Usage Report (2016).
- Servers are hosted in mid-tier internal data centers (2000 to 20000 sq ft) with superior cooling systems and redundant power.
- Only the expected energy consumption of desktops and servers are used in the calculation Energy consumption of peripherals such as the monitor and keyboard are excluded.
- The benchmark desktop is a 2016 Dell Multiplex 7040.
- The total cost of ownership of Kazuhm is defined on an annual basis for better comparability with reserved instance plans from AWS and Google Cloud Platform.
- Calculation of expected cloud expenditure includes the volume discounts by offered AWS and Google Cloud for their reserved instances. Information on these discounts are available on their respective websites.
- There is a $0 floor imposed on the output of the calculator. Output can be negative when there are a high number of nodes used at low usage levels. This is a consequence of the Kazuhm fixed node pricing model.
- Calculation assumes that any application running on Kazuhm will perform the same whether it’s running on multiple small machines or a single large machine of equivalent capacity.
- Expected energy consumption of desktops and servers includes the energy consumed by the CPU, network and disk. Network and disk energy consumption are assumed to be fixed at all levels of usage while CPU energy consumption varies at different usage levels.
- Annual Kazuhm total cost of ownership (TCO) is defined as the price of Kazuhm plus the energy consumed by the desktops and servers running Kazuhm workloads. This assumes that there is no additional manpower/staff required to manage Kazuhm on the deployed infrastructure on an ongoing basis.
- The benchmark server is a theoretically constructed 12-core machine.
- Potential cloud savings is defined as expected cloud expenditure minus Kazuhm TCO. This savings is expressed on an annual basis.
- Assumes that customers can provision an unlimited number of instances/ VMs without being bound by the limits imposed by the cloud providers.
- AWS instances are assumed to be hosted in the US East Region, North Virginia, and are priced according to the rates in that region. Instances in this region are generally the cheapest compared to other regions.
- Google Cloud VMs are assumed to be hosted in the US Central-1 region, Iowa, and are priced according to the rates in that region. VMs in this region are generally the cheapest compared to other regions.