Compute Flashcards
What are reserved instances, to what do they apply and what’s the max. possible saving in %?
It’s a billing discount that is automatically applied to On-Demand instances when these instances match certain criteria (instance type, region, tenancy, platform). Savings can be up to 72% as compared to On-Demand rates.
Reserved instances can be purchased for a one-year or three-year commitment, with the three-year commitment offering a bigger discount.
What are Saving Plans and how do they differ from Reserved Instances in regards to the commitment type, flexibility and capacity reservations?
It’s a billing discount that’s applicable to AWS compute usage, saving up to 72%
- Commitment type is “hours of compute usage”, rather than the “per instance” type used by RI.
- More flexible as not only applicable to EC2, but also to Fargate and Lambda (indirectly also to ECS/EKS)
- Also more flexible as not dependent on tenancy or operating system
- Do not reserve capacity!
What are the two different types available for Saving Plans?
Compute Saving Plans and EC2 Instance Saving Plans
What’s the maximum possible saving in % with EC2 Instance Saving Plans and to what can they be applied?
Up to 72% (just like Standard RIs) in exchange for commitment to usage of individual instance families in a Region (for example, M5 usage in N. Virginia). This automatically reduces your cost on the selected instance family in that region regardless of AZ, size, operating system, or tenancy.
Besides cost savings, what’s another advantage of Reserved Instances?
They provide capacity reservation when used in a specific AZ.
Note: capacity reservation is optional - if not used, the 72% discount is not bound to a specific AZ, but automatically applies to any EC2 instance that meets the criteria in a given region. This is called a “regional RI” (vs. “zonal RI)”
What’s the maximum possible saving in % with Compute Saving Plans and to what can they be applied?
Up to 66% (just like Convertible RIs), applicable to EC2 regardless of instance type, region, tenancy and platform. They are also applicable to Fargate and Lambda.
What’s a Convertible Reserved Instance and what cost saving can they offer?
It’s a billing discount that offers up to 66% of savings as compared to On-Demand and allows you change the instance type, operating system or tenancy of an instance, without losing the benefit of the discount.
How do Standard and Convertible Reserved Instances differ from each other in terms of the following:
- Discount (up to)
- Changing AZ
- Changing instance size
- Changing networking type
- Changing instance family
- Changing OS
- Changing tenancy
- Changing payment options
- Benefit from price reductions
- Sellable via marketplace
- Discount: Standard=72%, Conv.=66%
- Changing AZ: Standard=yes, Conv.=yes
- Changing instance size: Standard=yes, Conv.=yes
- Changing networking type: Standard=yes, Conv.=yes
- Changing instance family: Standard=no, Conv.=yes
- Changing OS: Standard=no, Conv.=yes
- Changing tenancy: Standard=no, Conv.=yes
- Changing payment options: Standard=no, Conv.=yes
- Benefit from price reductions: Standard=no, Conv.=yes
- Sellable via marketplace: Standard=yes, Conv.=no
What’s the difference between “Zonal Reserved Instances” and “Regional Reserved Instances”?
For “Zonal Reserved Instances”, an AZ was selected, which effectively binds the RI to a single AZ. Benefit of that is that the capacity of the instance is reserved in that AZ.
“Regional Reserved Instances” are independent from the AZ, but do not guarantee any capacity.
For which operating systems can you change the Reserved Instance size?
Only for Linux/Unix Regional RIs with default tenancy.
NOT available for Windows, RHEL or SLES (Suse Linux Enterprise Server), Linux with SQL Server (and some more).
What’s the difference between Dedicated Instance and Dedicated Host in EC2?
Dedicated Host is a single, piece of hardware, dedicated to your account, so it’s not sharing hardware with any other customer. Typically used because of license requirements relating to usage of x number of CPU cores, or when the license dictates that the underlying host must stay the same.
Dedicated Instances are also using hardware dedicated to a single account, but the instance may move around from one dedicated host to another when instances are restarted for example. This flexibility on the underlying hardware may not always work for a given license requirement, which is why dedicated instances are generally cheaper (also because it saves AWS costs when being more flexible on the hardware).
What’s the difference between “Launch Configuration” and “Launch Template” in EC2?
Both define AMI, instance type, etc., but “Launch Templates” is the recommended choice as they are more flexible. For example, they support versioning and through that, having base versions of a configurations that is then used by other Launch Templates (think of them as children). They can also be used to launch EC2 instances, which isn’t possible with “Launch Configurations”.
