EC2 Image Sizes on Amazon Web Services
Amazon Web Services provides a wide range of image sizes, which should make it possible for you to meet your application performance needs by tuning the EC2 infrastructure it runs on. If you think that the variety of instance types makes it difficult to decide what to do, the variety of image sizes will make your mind reel.
The original instance type (Standard) aims for a good mixture of resources to meet the requirements of, well, standard applications. The other instance types contain a larger amount of one type of resource in terms of the other resource types of the instance; one particular instance type can then better support a particular set of application requirements than another.
This table illustrates the range of resources available across the instance types, just to give you an idea of the flexibility you have in choosing them for your application.
|Compute unit||1 (Standard M1.Small)||88 (Cluster Compute cc2.8 x Large)|
|Virtual core||1 (Micro, M1.Small, M1.Medium)||16 (2 x Intel Zeon 8 core Sandy Bridge architecture)|
|Memory||615MB (Micro)||68.4GB (High-Memory Quadruple Extra Large)|
|Instance store volume||None (Micro)||3360GB (Cluster Compute Eight Extra Large)|
|Network I/O||Low (Micro)||10 Gbps (High I/O Quadruple Extra Large, Cluster Compute Quadruple Extra Large, Cluster Compute Eight Extra Large, Cluster GPU Quadruple Extra Large)|
Amazon documentation lists 18 instance sizes, spread across the 7 instance types. That doesn't mean each type has two sizes, however. There are 4 standard instance types, though several of the more exotic types (Cluster GPU, for example) come in a single very large size.
For exotic types with a single size, Amazon's thinking is that the type of computing you're likely to do with them requires such high resource capacity that they'll go ahead and provide the largest possible numbers of resources within its overall infrastructure constraints.
Another way to say it is that the users who are likely to use these exotic instance types are so demanding of resources that they will want only the largest possible type that can be delivered, so Amazon hasn't bothered creating smaller sizes of these instance types.
AWS defaults to delivering the M1.small version if you fail to explicitly choose an instance type and size. The most common use case for most users is the Standard instance types, but they end up moving to the Large instance sizes after first beginning with sizes lower on the scale.