CALCULATING THE
BREAK-EVEN POINT:
AN AWS EXAMPLE
Instance: m5.large (2 vCPU, 8 GB Memory), running Ama- Formula:
zon Linux Reserved Instance cost ($/month)
On-Demand cost ($/hr)
Region: US East (northern Virginia)
On-demand (pay-as-you-go): $0.096/hour
Standard 1-year Reserved Instance, all paid
upfront: $501 ($41.75/month)
= Break-Even Point (hrs/month)
Calculation:
$41.75/month
$0.096/hour
= 435 hours/month
Interpretation:
• If this instance is expected to run for more than 435 hours each month, consider a 1-year reservation.
• If this instance is expected to run for less than 435 hours each month, use an on-demand payment model.
Use Cases:
Scenario A: This m5.large instance is being used for user-acceptance testing of application updates. New updates are
released on average once every 4 weeks, and testing takes about 5 days for each release. We can assume that
this instance is needed for 120 hours/month (1 release/month, 5 days/release x 24 hours/day = 120 hours/
month)
Outcome: Since the expected usage is less than the break-even point (120 < 435), using an on-demand payment model
is recommended.
Scenario B: This m5.large instance is being used as a development environment for an application whose developers are
located in India (GMT+5.5), Ireland (GMT+1) and California(GMT-7). Due to the differences in working hours
for the development teams, this instance is in constant use, except for a period of 36 hours each weekend. We
can assume that this instance is needed for 586 hours/month (730 hours/month - 36 hours/weekend x 4
weekends/month).
Outcome: Since the expected usage is more than the break-even point (586 > 435), purchasing a 1-year reservation for
this instance is recommended.
SUMMER
SUMMER 2019 | THE DOPPLER | 39