Edge computing allows use cases – Mobility Report
The cost of the edge
To compare the cost of deploying compute resources at different scales, we convert capital expenditure into depreciation by dividing each asset category by the number of years it will be written down, and then add the resulting depreciation to the annual opex, providing a snapshot of the annual cost structure. For example, power and cooling systems are written down over 14 years, whereas COTS servers are typically written down over 3 years.
- Server capex is mainly the cost of COTS servers and virtualization software.
- Other capex consists of the cost of components such as power distribution and cooling systems.
- The electrical power required to run and cool the servers.
- Other opex, mainly the cost of operations and maintenance (O&M).
As an example, we estimate the cost of compute resources for a CSP in Sweden. Initially, edge compute rollout is expected to be on aggregation sites having power capacity installed up to 10kW, hosting an average of 8 server units, each with 4 cores. With approximately 8,000 access sites and 1 aggregation site per 10 access sites, there is a virtual processor (vCPUs) capacity of 25,600 (800 sites x 8 servers per site x 4 cores per server) for enterprise applications at CSP-owned edge sites. Capex depends on the required capacity plus redundancy in the edge hardware components to meet the reliability requirements for edge services or applications. The geographic distribution can also be leveraged to improve the system availability by avoiding a single point of failure. We categorize the capex into server capex and other capex due to the faster cycle of server performance improvement compared to others. Servers are typically depreciated over 3 years while investments in power and cooling systems are depreciated over 14 years. Upgrading aggregation sites with edge compute capability, with an average of 8 units of servers, can draw up to 1.6MW (800 sites x 8 servers per site x 250W per server) for running the servers. With an assumed power efficiency factor of 2, 3.2MW power is needed on average to power all the aggregation sites. The cost of compute resource at each aggregation site is estimated to be around USD 20,000. Hence the USD per critical watt for an edge site is USD 20,000/(8 servers x 250W/server) = USD 10/W. This cost is very similar to USD per critical watt for building a large-scale data center.
Opex is the sum of electricity cost and O&M. For the current study, we assume it to vary in the range of USD 0.10–0.15/kWh. For O&M, the cost of full-time employees required to manage and maintain the distributed edge servers is projected.
We constructed four different scenarios to estimate and compare the compute resource cost, based on USD per vCPU-hour.
- Scenario 1 is a base case with costs assumed for a small- or medium-sized enterprise handling its compute needs with its own IT infrastructure.
- Scenario 2 is an estimation of cost for a large-scale data center to provision the same capacity as the first case.
- Scenario 3 is built around provisioning the capacity used in the first two cases by deploying edge computing on the CSP network.
- Scenario 4 is an extension of the third case, with the addition of the cost to implement a set of measures to reduce power consumption. These include using renewable energy, dynamic usage of battery/power storage at peak times and advanced cooling technologies, including a heat exchanger for the server cabinets.
Server capex is the most significant parameter for all the scenarios except the base case where O&M (other opex) dominates due to the lack of scale. Electricity cost is the second largest factor in USD/CPU-hour for scenario 3. This leads to the significance of additional power efficiency elements in scenario 4. With an estimate of expenditure in use cases suitable for edge deployment, the cost of edge compute resources can be just 10 percent more than that of a large-scale centralized one. Capacity utilization is the most important parameter for increasing the cost efficiency of the edge resources.