Was it really 80%? A deeper look into AWS’s latest price drop
Amazon announced today that it was “lowering the price the price of dedicated instances by up to 80%”. But was it really 80%?
We think there has been a misunderstanding in the coverage of this announcement, and we’d like to clarify what we think is at the very least misleading.
What are Dedicated Instances?
Some articles have made the confusion between Dedicated Instances and Reserved Instances. Let’s clarify this first.
Reserved Instances - as opposed to On-Demand Instances - are a payment strategy. Using Reserved Instances, you pay an upfront fee, and then you pay a lower hourly rate. If you actually use the capacity you reserved, you’ll have to pay a lower overall price.
Dedicated Instances (the one covered in the announcement) are instances that run on single-tenant hardware. Similarly to multi-tenant instances, they may be paid for using the On-Demand or the Reserved pricing model.
How are Dedicated Instances priced?
Since the hardware Dedicated Instances run on may not be leased to other customers, Amazon will bill you for the marginal waste.
Therefore, the pricing for Dedicated Instances is twofold. Here’s an illustration: in Red, your VMs, and in Grey, the Hardware. You’ll see that there’s unused capacity in each region, which AWS will charge you for.
The Instance Fee
The Instance Fee is charged hourly, and scales with the number of instances. This Fee also exists on regular, non Dedicated, AWS instances.
It is higher for Dedicated Instances than it is for regular instances.
This Fee was lowered in today’s announcement, by up to 37% for On-Demand instances, and 57% for Reserved Instances. We have to say “up to” as the exact value depends on the instance type.
The Dedicated Per Region Fee
The Dedicated Per Region Fee is how Amazon charges you for making hardware that you don’t use unavailable to other customers (since Dedicated Instances means that no-one else may use that hardware).
This fee is charged hourly, but does not scale with the number of instances. It only depends on the number of Regions in which you are operating Dedicated Instances.
The Dedicated Per Region Fee is the one that was cut by 80% and is now $2 (used to be $10).
How big a deal is this?
The important thing to realize is that the Dedicated Per Region Fee that was cut down by 80% from $10 to $2 does not scale with the number of instances.
It is therefore wrong to say that “AWS dropped the price of EC2 dedicated instances”, as some news outlets were misled into reporting.
For reference, this price cut is equivalent to the cost of running 4 on-demand m2.4xlarge instances - not insignificant, but likely to not matter if your business is heavily investing in the Cloud.
So, is this a non-event?
No, not exactly. Amazon also announced a significant price reduction on the Instance Fee.
Some instances, including the popular m1.large, saw their Instance Fee cost reduced by up to 37%.
If you were only running m1.large’s using Dedicated Instances, this would indeed mean that your EC2 bill would be coming down by 37% (excluding EBS fees).
Should you care?
Dedicated Instances are useful if you’re concerned with compliance and security and yet have a use case for the elasticity of the Cloud. They’re a product that’s typically targeted at Enterprises.
This move will particularly benefit Enterprises that understand the value of IaaS, and demonstrates yet again that Amazon wants to grab the lion’s share of that expanding market.
If you enjoyed this read - why don't you sign up for email updates?
We'll let you know when we publish new blog posts, we won't spam you.