Real-time Alerting

Radar's real-time Alerting feature provides instant notifications for critical events in your data pipeline, allowing you to act promptly on issues related to job performance, cost spikes, and data quality. This proactive approach helps maintain SLA targets and control costs effectively.

Prerequisites

• Completed the Real-time Alerting setup in Radar's Get Started guide.

Key Features

• Comprehensive Coverage: Built-in alerts for various scenarios for your pipelines: Snowflake, BigQuery, Bolt Schedules.

• Multi-channel Notifications: Receive on-time alerts via Slack, MS Teams, and email.

---

Available Alerts

Snowflake

Alert Name	Description	Recommended Threshold
Project Cost Anomaly Alert	Alerts when a project's cost exceeds a set percentage (e.g., 20%) above the daily average over the past 7 days.	0.2
User Cost Anomaly Alert	Alerts when a user's cost exceeds a set percentage (e.g., 20%) above the daily average over the past 7 days. Focuses on the top 10 costly users.	0.2
dbt Cost Anomaly Alert	Alerts when a dbt model's cost exceeds a set percentage (e.g., 20%) above the daily average over the past 7 days. Focuses on the top 10 costly dbt models.	0.2
Query Cost Anomaly Alert	Alerts when a specific query's cost exceeds a set percentage (e.g., 20%) above the daily average over the past 7 days. Focuses on the top 10 costly queries.	0.2
Warehouse Compute Cost Anomaly Alert	Alerts when a warehouse's compute cost exceeds a set percentage (e.g., 20%) of the daily average costs over the past 7 days.	0.2
Query Remote Spillage Ratio Anomaly Alert	Alerts when the ratio of bytes spilled to remote storage exceeds 5 times the total bytes scanned. Focuses on the top 10 queries with the highest costs.	5

BigQuery

Alert Name	Description	Recommended Threshold
Project Cost Anomaly Alert	Alerts when a project's cost exceeds a set percentage (e.g., 20%) above the daily average over the past 7 days.	0.2
User Cost Anomaly Alert	Alerts when a user's cost exceeds a set percentage (e.g., 20%) above the daily average over the past 7 days. Focuses on the top 10 costly users.	0.2
dbt Cost Anomaly Alert	Alerts when a dbt model's cost exceeds a set percentage (e.g., 20%) above the daily average over the past 7 days. Focuses on the top 10 costly dbt models.	0.2
Query Cost Anomaly Alert	Alerts when a specific query's cost exceeds a set percentage (e.g., 20%) above the daily average over the past 7 days. Focuses on the top 10 costly queries.	0.2
Small Query Waste Anomaly Alert	Alerts when the difference between billed bytes and processed bytes exceeds a set percentage (e.g., 5%) of total billed bytes for a project in a day.	0.05
High Shuffle Spill Anomaly Alert	Alerts when shuffled bytes spilled to disk exceed a set percentage (e.g., 20%) of total bytes processed. Focuses on the top 10 queries with the highest processed bytes.	0.2
Query Runtime Anomaly Alert	Alerts when query runtime increases by more than a set percentage (e.g., 50%) compared to the average over the past week.	0.5
Query Slot Usage Anomaly Alert	Alerts when average slot usage increases by more than a set percentage (e.g., 50%) compared to the average over the past week. Focuses on the top 10 costly queries.	1

Bolt

Alert Name	Description	Recommended Threshold
Schedules Time Guard Alert	Alerts when a schedule run takes longer than expected. The alert triggers when a schedule takes more than a set percentage (e.g., 20%) longer than the average over the last 7 days.	0.2

Threshold Range: All thresholds must be set between 1-100 for valid configurations.

---

1. How to Configure Alerts

1. Navigate to the Data Alerts section in Radar.

2. Select the alert you want to configure.

3. Set the threshold for the alert. For example:

   • For cost anomaly alerts, a threshold of 0.2 means the alert will trigger when costs exceed 20% above the average of the past 7 days.

   • For time-based alerts like the Schedules Time Guard, a threshold of 0.2 means the alert will trigger when a run takes 20% longer than the average of the past 7 days.

4. Click the toggle switch to enable/disable the alert.

---

2. Receiving and Acting on Alerts

When an alert is triggered, you'll receive a notification through your configured channels (Slack, MS Teams, or email). The alert will provide:

• A description of the issue

• Relevant metrics and comparisons

• Suggested actions to address the problem

How to Act on Alerts:

• Use this information to quickly diagnose and resolve issues, helping to maintain the efficiency and cost-effectiveness of your data operations.

---

Best Practices

1. Start with key metrics: Begin by setting up alerts for your most critical operations and costs.

2. Refine thresholds: Adjust alert thresholds over time based on your specific needs and patterns.

3. Regular review: Periodically review your alert configurations to ensure they remain relevant and effective.

4. Team alignment: Ensure that the right team members are receiving the appropriate alerts for quick action.