Burstable Instances
- Burstable instances have baseline performance.
- They are used for low usage with high peaks ocassionally.
AWS Burstable Instances
| Instance Type | vCPUs | Baseline CPU (%) | CPU Credits/hour | Memory (GiB) | Notes |
|---|---|---|---|---|---|
| T2.nano | 1 | 5% | 3 | 0.5 | Older gen |
| T2.micro | 1 | 10% | 6 | 1 | Free tier eligible |
| T2.small | 1 | 20% | 12 | 2 | |
| T2.medium | 2 | 20% | 24 | 4 | |
| T2.large | 2 | 30% | 36 | 8 | |
| T2.xlarge | 4 | 40% | 96 | 16 | |
| T2.2xlarge | 8 | 40% | 192 | 32 | |
| T3.nano | 2 | 5% | 6 | 0.5 | Uses Nitro |
| T3.micro | 2 | 10% | 12 | 1 | Free tier eligible |
| T3.small | 2 | 20% | 24 | 2 | |
| T3.medium | 2 | 40% | 48 | 4 | |
| T3.large | 2 | 60% | 72 | 8 | |
| T3.xlarge | 4 | 60% | 144 | 16 | |
| T3.2xlarge | 8 | 60% | 288 | 32 | |
| T4g.nano | 2 | 5% | 6 | 0.5 | ARM-based (Graviton) |
| T4g.micro | 2 | 10% | 12 | 1 | |
| T4g.small | 2 | 20% | 24 | 2 | |
| T4g.medium | 2 | 40% | 48 | 4 | |
| T4g.large | 2 | 60% | 72 | 8 | |
| T4g.xlarge | 4 | 60% | 144 | 16 | |
| T4g.2xlarge | 8 | 60% | 288 | 32 |
- Burstable is also available in ebs volumes
Azure Burstable Instances
| VM Size | vCPUs | Baseline CPU (%) | CPU Credits/hour | Memory (GiB) | Notes |
|---|---|---|---|---|---|
| B1ls | 1 | 5% | 3 | 0.5 | Cheapest Azure VM |
| B1s | 1 | 10% | 6 | 1 | |
| B1ms | 1 | 20% | 12 | 2 | |
| B2s | 2 | 40% | 48 | 4 | |
| B2ms | 2 | 60% | 72 | 8 | |
| B4ms | 4 | 70% | 168 | 16 | |
| B8ms | 8 | 80% | 384 | 32 | Highest in B-series |
Scaling
- Elasticity refers to scaling up/down/in/out (increase or decrease)
- Scaling is of two types (computing)
- Vertical Scaling:
- increasing/decreasing the instance cpu/ram resources (sizes/types)
- Scale up and down refers to vertical scaling
- Generally will have a down time.
- Horizontal Scaling:
- increasing/decreasing the number of instances.
- Scale in and scale out refers horizontal scaling.
- doesn’t have a down time.
- Vertical Scaling:
Horizontal Scaling
- AWS enables horizontal scaling using Auto Scaling groups where as Azure enables this by Virtual Machine Scale Sets (VMSS).
- Generally when we use Horizontal scaling we need custom images where the application gets started
- We need to figure out
- How to create custom images
- Using image how do i configure horizontal scaling
- How do i rollout new versions (Without downtime) and go back to previous versions in cases of failure.
