Overview

Cloudability is enhancing the resource inventory metrics to improve data accuracy and customer analysis capabilities. We're introducing null value representation and a new IS NULL filter operator to distinguish between "no data available" and "actual zero values". This is a critical distinction that may impacts the accuracy of customer analytics.

Key Problem Solved: Previously, metrics with no data were indistinguishable from zero values, causing customers' exported data and calculations to be systematically inaccurate. One documented case showed this issue artificially reducing average memory utilization calculations from 28% to 13%.

The Problem: Zero vs. Null Ambiguity

Current Limitation

When the resource inventory API returns { "cpu_utilization": 0 }, it's not possible to know if this means:

• Actual measurement: CPU utilization is genuinely 0%
• Missing data: No CPU utilization data is available for that resource

This ambiguity cascades into customer workflows:

1. Resource Inventory Data Export: Resources without metric data appear as 0 in CSV exports
2. Incorrect Calculations: Customers calculate averages that include these artificial zeros, skewing results downward
3. No Filtering Option: No way to identify resources with data quality issues
4. Business Impact: One customer's analysis showed memory utilization calculations dropping from ~28% to ~13% when zero values for missing data were included in the average

The Solution: Null Values + IS NULL Operator

Part 1: Null Representation

In API: Metrics with no available data will now return null instead of 0:

Before:

{
  "cpu_utilization": 0,
  "memory_utilization": 0,
  "unblended_cost": 0
}

After:

{
  "cpu_utilization": null,        // No data available
  "memory_utilization": 25.5,     // Actual measured value
  "unblended_cost": 0             // Actual zero cost
}

In UI: Metrics with no available data will show the value as 'N/A'. 

In CSV Export: Metrics with no available data will show the value as 'Not Available'. 

Part 2: IS NULL Filter Operator

A new filter operator lets you explicitly search for null or non-null metrics. This operator applies only to metrics, not dimensions (string fields) and available on both UI and API. 

metric IS NULL true - Find Missing Data

Returns only records where the specified metric has no data:

cpu_utilization IS NULL true

Results: All resources where CPU utilization data is unavailable

metric IS NULL false - Find Complete Data

Returns only records where the specified metric has a value (including zero):

memory_utilization IS NULL false

Results: All resources with available memory utilization data (whether measured at 0% or higher)

Use Cases

1. Data Quality Analysis

Identify resources with incomplete metric coverage:

memory_utilization IS NULL true AND cpu_utilization IS NULL false

Results: Resources with CPU data but missing memory data (potential data collection issue)

2. Accurate Aggregations

Calculate average CPU utilization only for resources with data:

cpu_utilization IS NULL false

Export these results → Calculate average → No artificial zeros skewing the result

3. Cost Analysis Filters

Combine IS NULL with existing operators:

(unblended_cost IS NULL false) AND (unblended_cost > 100)

Results: Resources with actual cost data > $100 (excludes unknown/unmeasured costs)

Technical Highlights

Backward Compatibility

• Existing comparison operators unchanged: metric==0 still matches only actual zeros, automatically excluding nulls
• Sorting behavior preserved: Default descending sort shows resources with data first, null values appear last
• Dimensions unaffected: String fields work exactly as before

Error Handling

The system provides clear error messages:

Scenario	Error Response
Use IS NULL on a dimension field	400 Bad Request: "IS NULL operator is only supported for metrics, not dimensions"
Missing value parameter	400 Bad Request: "IS NULL operator requires value: true or false"
Invalid value (not true/false)	400 Bad Request

Sorting Behavior

• Ascending (ASC): Null values appear first, then sorted values
• Descending (DESC): Sorted values appear first, null values at end
• Default UI behavior: Metrics sorted descending by default, so resources with complete data display first

Impact & Benefits

Availability

This enhancement will be available in the resource inventory UI and API on Aug 18th. 

------------------------------
Alok Jain
------------------------------