EC2 Rightsizing recommendations not showing up for custom efficiency targets

Products

• CloudHealth

Issue/Introduction

Custom targets set to a narrow range example ( 40-60%) often generate fewer or no optimization recommendations compared to broader average targets spanning (25-75)%

Resolution

The rightsizing engine scans your EC2 instances and matches them against a predefined set of "candidate instance types" based on vCPU range filtering. The target utilization range (25-75% vs 40-60%) directly controls this vCPU filtering window.

Custom targets use a narrower vCPU range filter, resulting in fewer or no rightsizing recommendations.

Workaround: Widen the target utilization range (e.g., 30-70% instead of 40-60%) to expand the vCPU search window and generate more candidate matches.

Why it works: Broader utilization targets → Larger Elasticsearch (Greater Than or Equal to (≥) / Less Than or Equal to (≤)): range → More EC2 instance types qualify as candidates.

Custom target ranges establish a midpoint utilization target that drives instance type recommendations, with deviation tolerance determining candidate pool size.

Target Range(example): 40-60%

Midpoint: 50% ← Core utilization target for matching

Target Range: 30-70%  

Midpoint: 50% ← Same midpoint, wider tolerance (±20%)

Fixed upper bound (80%), varying lower bound:

Lower Target	Range Width	Midpoint	Recommendation Pattern
40%	40-80% (40%)	60%	Smaller instance recommendations
30%	30-80% (50%)	55%	Mixed instance sizes
20%	20-80% (60%)	50%	More larger instance recommendations

Lowering the range floor → Wider deviation tolerance around midpoint → Expands vCPU search window → More large-instance candidates qualify 

Different vCPU filtering ranges limit candidate instance types.

Target Type	vCPU Range	Recommendation Volume	Elasticsearch Range Filter
Average (25-75%)	~4-12 vCPUs	Multiple options	Greater Than or Equal to (≥):4, Less Than or Equal to (≤):12
Custom (40-60%)	~4-8 vCPUs	Few/none	Greater Than or Equal to (≥):4, Less Than or Equal to (≤):8

How It Works

1. Rightsizing scans current instance utilization
2. Applies target utilization range to calculate ideal vCPU needs
3. Filters candidate instance types by vCPU range (determined by target width)
4. Returns matches within 10% tolerance of target utilization

Troubleshooting Steps:

1. Switch to Average Target → Verify recommendations appear
2. Check current utilization → Very low (<20%) or very high (>80%) instances rarely match narrow targets
3. Review instance family → Some families have limited type options in narrow vCPU ranges
4. Test intermediate targets (30-70%) → Balance between coverage and precision