API HealthCheck Dashboard

Overview

This guide covers the API Uptime/Healthcheck Dashboard — used to view API performance and send automated alerts to clients during performance degradation.

Dashboard URL: https://ind.hvstatus.co/

Problem

Failing to send timely alerts to clients during vendor downtime was a major pain point in NPS surveys, likely to worsen with new vendors in the API marketplace.

Solution

The API Healthcheck Dashboard provides:

1. Monitor API performance historically
2. Monitor API performance in real-time
3. Send automated alerts to clients during outages
4. Send automated alerts once service is operational
5. Send communication regarding scheduled downtime
6. Configure alerts for APIs specific to a client based on usage
7. Configure client email IDs for notifications

Default thresholds: 20% error rate in 5 minutes (configurable per API).

Access by User Type

User Type	Access Flow
Platform Clients	HV One Dashboard → Dev Hub → API Health → Select Country → Dashboard Opens
Non-Platform Clients	Credentials generated in backend → Client login → Access via CTA in emails
HV Members	Login via HV email credentials

info

One-time login to HV Dashboard is mandatory to access the downtime dashboard. After that, it can be accessed via CTA in emails.

The Dashboard: Deep Dive

Understanding Downtime Thresholds

The dashboard provides a high-level view of API uptime but does not reflect precise real-time accuracy.

Example: If we set a performance threshold of 80% for an API:

• Performance drops below 80% → dashboard shows downtime
• Performance above 80% (even at 90%) → dashboard shows 100% uptime

Components

Component	Use Case
Instatus	Communication platform to display uptime and enable client subscriptions
Pingdom	Service that calls healthcheck APIs to detect if a service is up or down
Healthcheck API	API that monitors endpoints and triggers notifications based on failure percentage

Playbook

Calculating Thresholds and Min Volume

Steps to calculate threshold:

1. Identify a day where actual downtime occurred
2. Analyze API calls per minute and failure percentage within a fixed interval (e.g., 5 or 10 minutes)
3. If failure percentage usually stays below 30%, set 30% as the threshold
4. Define downtime as failures exceeding the threshold for more than the fixed interval

Steps to calculate min volume:

1. Analyze API call volumes during low-traffic periods (nighttime)
2. Calculate average volume within the fixed interval
3. Establish minimum volume threshold — traffic below this level is not considered for downtime detection

FAQs

Q: How many APIs are currently supported? A: Currently 26, with plans to scale to top 50.

Q: How far back can we access data? A: As far back as needed, but accurate historical data starts from June 2024.