# Measuring Reserved Instance profitability with the break even point

At the end of the day, the most important attribute of any RI purchase is its profitability. In other words, did purchasing that reservation save you money over running your instances on-demand?

You can determine this by referencing a value termed the ** total-cost break even point**. This represents the point at which you’ve used the reservation for enough hours to make up for the cost of the reservation, by accumulating savings over on-demand. If you use a reservation enough, you will most definitely pass this break even point and realise profit. But every Reserved Instance class will require a different utilization level to break even.

## Calculating the total cost break even point

There are only two numbers that really matter for calculating break even points: the effective hourly rate for on-demand, and the effective hourly rate under reservation. You can find these on the main AWS pricing page. Here’s an example for an M4.large running Linux in US East.

As you can see, the Partial Upfront payment option has an effective hourly rate of $0.074, compared to $0.126 for on-demand. As a thought exercise, let’s presume you just have one matching instance in your portfolio. If that instance runs for every hour of the year, then your total on-demand cost is $1103.76 (8760 hrs/year * $0.126) whereas your total reserved cost would be $648.24 (8760 hrs/year * $0.074), resulting in the 41% savings rate advertised by AWS in the screenshot above.

A key feature of RIs is that you commit to pay for every hour of the RI term, regardless of whether they are utilized—whereas for on-demand, you have the advantage of only paying for the actual hours used. If the instance had been turned off 50% of the time in the example above (such that the RI was only used 50% of the time), then the numbers come out vastly different: $551.88 (4380 hrs/year * $0.126) for on-demand* *compared to $648.24 (8760 hrs/year * $0.074) for the RI. In this scenario, the break even point has **not** been passed, and the RI has been unprofitable.

The break even point can actually be inferred directly from the savings rate. Running an on-demand instance only 59% of the time would save you 41% of the cost of running it 100% of the time. This means that at least 59% of the hours in a term should be used in order to justify purchasing an RI with a 41% savings rate—for any lower utilization rate, you’d be better off sticking with on-demand. If the RI can be utilized for a greater amount of time than this, then it can become significantly more profitable.

On the individual RI recommendation pages within Cloudability you’ll actually find what the overall savings rates are calculated to be (these take into account periods when RIs may not be fully utilized).

## Visualizing the break even point

Let’s again presume that you purchase one Partial Upfront reservation, and it indeed turns out that it is used 100% of the time (this means that for every hour of the year, there was at least one instance available to use the RI). Your cash flow situation would look something like this:

You’ll notice that initially, the reservation costs more than on-demand due to the upfront payment. You’ll also notice that at the 5 month point, the lines intersect. This is the ** cash flow break even point**, where the cash spent (thus far) with on-demand starts becoming greater than on the reservation. You aren’t actually guaranteed of profit though until just after 7 months—at this point you’ve already spent more with on-demand than with your reservation. This happens to be 59% of the way through year—which again makes sense, as of course the saving rate is 41%.

Another way of thinking about the break even point is in terms of how often your reservation has to be used for it to break even—in this case, 59% of the year. The instances could be used for 100% of each day, every day, for the first 59% of the year (until roughly August 1st) and then turned off and never used again. Alternately, the instances could be used for 59% of each day, every day, for a full year. In that case, the chart would look like this:

Both the cash flow break even point *and* total RI cost break even point occur at the same time: the end of the RI period. The salient point to note from both these ways of looking at it is that if you can be sure that your RI will be utilized for more than 59% of the hours of the RI term you’ll be ahead. This percentage value will be dependent on your reservation-instance type combo, but you’ll always be able to infer it from the relevant savings rate.

The good news is that Cloudability does all this hard work for you, calculating RI purchase recommendations based on your recent EC2 usage and working out whether you’d save money by purchasing reservations to cover your instance hours. Log in or sign up for a free 14-day trial to pick the right Reserved Instances for your infrastructure.