Performance Efficiency - Compute performance options Flashcards
1
Q
Compute Considerations
A
- Understand your workload: gather and analyze data
- Understand AWS compute
- Test, test, and test
2
Q
EC2 - Characteristics
A
- It’s the default option. You own OS, and choose resources (IaaS)
- Resources: vCPUs, memory, storage and network
- Auto scaling options: based on metrics, based on schedule, and based on health to maintain capacity
3
Q
EC2 - Benefits
A
- Burstable: some instance types bank capacity so they can be used later when there’s excess workload
- Offers GPU focused instances
- Also provides FPGA focused instances (filled programable gate arrays): customized hardware accelerators
4
Q
EC2 - Instance types
A
- General purpose
- Compute optimized
- Memory optimized
- Accelerated computing: includes GPU and FPGA focused instances
- Storage optimized
- Bare metal to support licensing needs
5
Q
ECS - Characteristics
A
- Useful for similar workloads as EC2: long running apps, batch processing, and microservices
- Benefits: provides better use of resources, ELB, and Autoscaling
- AWS Fargate: lets you manage instances of containers for a serverless experience
6
Q
Lambda - Characteristics
A
- Useful for backend, event, stream, and data processing
- Types of triggers: AWS resource triggers and API endpoints
- Must identify memory, so AWS will provide CPU to process it
- Benefits: simply execute code, automatic scaling, fault tolerant, and pay only for the usage. In the case of EC2 and ECS you pay for instances even the system isn’t using it