Amazon EC2 provides virtual compute capacity in the AWS cloud. Each instance type is a combination of CPU, memory, storage, and networking capacity, and AWS groups them into families based on their intended use. Selecting the right instance type from the start reduces cost, avoids scaling problems, and ensures your application runs within its expected performance envelope. Official documentation: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/instance-types.html and https://aws.amazon.com/ec2/instance-types/
The Instance Type Families
AWS organises EC2 instances into six primary categories, each designed for a different class of workload. Understanding which category your application fits into is the starting point for any sizing decision.
General Purpose instances (M and T families) provide a balanced combination of compute, memory, and networking. They are suited to web servers, code repositories, small-to-medium databases, and applications where resource consumption is roughly proportional. T-series instances are burstable — they accumulate CPU credits during periods of low utilisation and spend those credits during spikes, making them cost-effective for workloads with variable demand.
Compute Optimised instances (C family) are designed for applications that need high-performance processors. Typical uses include batch processing, media transcoding, high-performance web servers, scientific modelling, and gaming servers. Memory Optimised instances (R and U families) are built for workloads that process large data sets in memory — in-memory databases, real-time big data analytics, and high-performance enterprise applications. Storage Optimised instances deliver high sequential read and write access to large local data sets, suitable for data warehousing and log processing. Accelerated Computing instances use hardware accelerators such as GPUs for machine learning inference, graphics rendering, and high-performance computing.
- General Purpose (M, T): balanced CPU, memory, networking — web servers, dev environments
- Compute Optimised (C): high-performance processors — batch processing, transcoding, game servers
- Memory Optimised (R, U): large in-memory workloads — analytics, enterprise databases
- Storage Optimised (I, D, H): high I/O throughput — data warehousing, log processing
- Accelerated Computing (P, G, Inf): GPU and hardware accelerators — ML inference, rendering
Naming Conventions and Generations
EC2 instance names follow a consistent pattern: instance family, generation, processor family, additional capabilities, and size. For example, m7g.large means: M family (general purpose), 7th generation, powered by AWS Graviton (g), large size.
AWS recommends using current-generation instance types for new workloads. Current general purpose types include the M8g (Graviton4), M8i, and M7i families. AWS continues to support previous generations for existing workloads but does not recommend them for new deployments.
Graviton-based instances (denoted by the g suffix, e.g. M7g, C8g, R8g) are powered by AWS-designed Arm processors and are frequently recommended for workloads that can take advantage of their price-performance characteristics. Reference: https://docs.aws.amazon.com/ec2/latest/instancetypes/ec2-instance-type-specifications.html
- Instance name format: family + generation + processor + capabilities + size
- Current generation types are recommended for all new workloads
- Graviton (g-suffix) instances offer strong price-performance for compatible workloads
- Burstable T-series instances use CPU credits — appropriate for variable workloads
How We Approach Instance Selection
In our standard AWS baseline, we prioritise workload-specific instance sizing from day one rather than defaulting to a large instance and hoping for the best. Oversized instances waste budget; undersized instances create reliability problems that are harder to diagnose under load.
For most application services we start with current-generation M or T family instances, then move to C or R family instances when profiling shows the workload is CPU-bound or memory-bound. We pair instance selection with health checks, monitored service uptime, and security group hardening as part of a baseline deployment standard.
AWS provides a Find an EC2 Instance Type tool in the console and documentation to help match your requirements to available instance types: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/instance-types.html