mv-spot-viewer

/metrics-on-demand API Documentation

Description

This endpoint provides metrics for AWS Batch users running completely on-demand. It extracts data from DynamoDB to calculate various metrics, including job counts, retries, and costs.

---

API: /metrics-on-demand

Response Format

The API returns the following metrics in JSON format:

Field	Description
NumberOfJobs	Count of all job entries (JobID).
NumberOfRetries	Sum of the Attempts column.
TotalRunTime	Total run time in seconds (RunDurationSeconds).
JobQueueName	Unique job queue name(s) from JobQueue.
TotalOnDemandCost	Total cost of running jobs on On-Demand instances (USD).
TotalSpotCost	Estimated total cost if jobs were run on Spot Instances (USD).

---

Calculations

1. Total On-Demand Cost (USD)

Cost = Sum of (On-Demand Price (USD/hour) * (Longest Run Duration (seconds) / 3600)) for all instances.

• Extract unique InstanceIDs and their corresponding InstanceTypes from InstanceMap.
• Obtain the longest RunDurationSeconds for each InstanceID.
• Multiply by the On-Demand price for the region.

2. Total Spot Cost (USD)

Total Spot Cost (USD): Cost = Sum of (Spot Price * (Longest Run Duration (seconds) / 3600)) for all instances.

• Extract unique InstanceIDs and their InstanceTypes from InstanceMap.
• Multiply the longest RunDurationSeconds by the average Spot price for the region.

---

Example Response

{
  "NumberOfJobs": 150,
  "NumberOfRetries": 20,
  "TotalRunTime": 360000,
  "JobQueueName": ["HighPriorityQueue"],
  "TotalOnDemandCost": 120.50,
  "TotalSpotCost": 32.15
}

Notes

• All monetary values are rounded to two decimal places for better readability.
• Job Queue Name(s) are returned as a list of unique strings.

---

metrics-spot.md

# `/metrics-spot` API Documentation

## Description
This endpoint provides metrics for AWS Batch users running completely on Spot Instances. It calculates costs, savings, and other metrics based on DynamoDB data.

---

## API: `/metrics-spot`

### Response Format
The API returns the following metrics in JSON format:

| Field                   | Description                                                                                  |
|-------------------------|----------------------------------------------------------------------------------------------|
| `NumberOfJobs`          | Count of all job entries (`JobID`).                                                          |
| `NumberOfRetries`       | Sum of the `Attempts` column.                                                                |
| `TotalRunTime`          | Total run time in seconds (`RunDurationSeconds`).                                            |
| `JobQueueName`          | Unique job queue name(s) from `JobQueue`.                                                    |
| `TotalOnDemandCost`     | Estimated cost of running jobs on On-Demand instances (USD).                                 |
| `TotalSpotCost`         | Total cost of running jobs on Spot Instances (USD).                                          |
| `EstimatedCostSaving`   | Cost savings achieved by using Spot Instances over On-Demand instances (USD).                |

---

### Calculations

#### **1. Total On-Demand Cost (USD) (Estimate)**

Total On-Demand Cost (USD):
Cost = Sum of (On-Demand Price (USD/hour) * (Longest Run Duration (seconds) / 3600)) for all instances.

- Extract **unique `InstanceIDs`** and their corresponding `InstanceTypes` from `InstanceMap`.
- Obtain the **longest `RunDurationSeconds`** for each `InstanceID`.
- Multiply by the **On-Demand price** for the region.

#### **2. Total Spot Cost (USD)**

Total Spot Cost (USD):
Cost = Sum of (Spot Price * (Longest Run Duration (seconds) / 3600)) for all instances.

- Use Spot price history for each `InstanceID` to calculate the cost.
- Multiply the **longest `RunDurationSeconds`** by the **Spot price** for the region.

#### **3. Estimated Cost Saving**

Estimated Cost Savings = Total On-Demand cost - Total Spot cost

---

### Example Response

```json
{
  "NumberOfJobs": 120,
  "NumberOfRetries": 15,
  "TotalRunTime": 240000,
  "JobQueueName": ["DefaultQueue"],
  "TotalOnDemandCost": 100.75,
  "TotalSpotCost": 28.50,
  "EstimatedCostSaving": 72.25
}

Notes • Monetary values are rounded to two decimal places for better readability. • Savings are calculated as the difference between On-Demand and Spot costs. • Job Queue Name(s) are returned as a list of unique strings.

---

Additional Notes

1. Ensure to serve these markdown files alongside your API for better documentation.
2. The values for Spot prices and On-Demand prices should be dynamically fetched based on AWS Pricing APIs to ensure accuracy.

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details