Diagnosing performance issues in a managed Redis instance requires visibility into metrics, logs, and key access patterns. Tair (Redis OSS-compatible) provides built-in observability tools — metrics, logs, and analytics — that go beyond what open-source Redis offers, so you can monitor instance health, investigate slowdowns, and identify problematic keys without external tooling.
Feature overview
Observability is built on three types of data:
Metrics: Numeric measurements collected over time that show the state and trends of a system.
Logs: Records of discrete events that occur during runtime, including errors, slow queries, and operations.
Traces: End-to-end records of a request's lifecycle.
Trace analysis typically requires middleware or client-side code changes on your client.
The following table compares observability features across editions.
| Observability | Feature | Redis | Redis Open-Source Edition | Tair (Enterprise Edition) |
|---|---|---|---|---|
| Metrics | Performance metrics | Supported | Supported (fine-grained) | Supported (fine-grained) |
| Logs | Active logs | Supported | Supported | Supported |
| Slow query logs | Supported | Supported | Supported | |
| Audit logs | Not supported | Supported | Supported | |
| Latency insights | Not supported | Supported | Supported | |
| Traces | ➖ | ➖ | ➖ | ➖ |
| Analytics | Real-time analysis of hotkeys | Not supported | Supported | Supported |
| Real-time analysis of large keys | Not supported | Supported | Supported | |
| Offline key analysis | Not supported | Supported | Supported | |
| Instance diagnostics | Not supported | Supported | Supported |
Metrics
Tair (Redis OSS-compatible) includes all standard Redis metrics — memory distribution, memory usage, memory fragmentation ratio, connection and command counts, network traffic, synchronization status, CPU utilization, and keyspace information — plus fine-grained read queries per second (QPS) and write QPS.
The fine-grained metrics enable three additional monitoring capabilities:
Real-time performance monitoring: displays performance metrics as they update.
Session management: shows active sessions between an instance and its clients.
Performance trends: shows how metrics change over a selected time window.
For the full list of available metrics, see View performance monitoring data.
Logs
Tair (Redis OSS-compatible) provides four log types for monitoring and troubleshooting instances.
Active logs
Active logs record persistence operations, synchronous replication events, debugging activity, and runtime error messages. Each entry appears as a row in the log.
To view active logs, go to the instance details page and choose Logs > Active Logs in the left-side navigation pane. For more information, see Query active logs.
Slow query logs
Slow query logs capture requests that exceed specific thresholds to execute. Execution time counts only the time spent running the command — it excludes time spent queuing or in network transmission.
Each slow query log entry includes:
| Field | Description |
|---|---|
| Execution timestamp | When the command was processed |
| Execution duration | How long the command took to run |
| Command and parameters | The command that was executed |
| Client information | The client that issued the command |
Use slow query logs to identify commands taking longer than expected, then optimize them to prevent service congestion.
To view slow query logs, go to the instance details page and choose Logs > Slow Logs in the left-side navigation pane. For more information, see Query slow logs.
Audit logs
Audit logs are built on Simple Log Service (SLS) and capture all online operation logs, including sensitive operations such as FLUSHALL, FLUSHDB, and DEL.
Each audit log entry includes:
| Field | Description |
|---|---|
| Log type | Category of the logged event |
| Execution duration | Time taken to execute the operation |
| Database number | The database the operation targeted |
| Client IP address | IP address of the client that issued the command |
| Account name | The account that performed the operation |
| Command details | Full command that was executed |
| Extension information | Additional context about the operation |
Audit logs let you search and analyze online operations, slow queries, and active logs, and export the results.
To view audit logs, go to the instance details page and choose Logs > Audit Logs in the left-side navigation pane. For more information, see Enable the audit log feature.
Latency insights
Latency insights records execution times for up to 27 events across all commands, and retains the data for the past three days.
To view latency data, go to the instance details page and choose CloudDBA > Latency Insights in the left-side navigation pane. For more information, see Use the latency insights feature.
Analytics
Tair (Redis OSS-compatible) integrates performance metrics, traces, and logs to provide analytics tools for diagnosing instance issues.
Hot key and large key analysis
A hot key receives significantly more requests than other keys. If not addressed, hot keys cause skewed request distribution and can lead to cache breakdowns. A large key contains many members or occupies a large amount of memory. If not addressed, large keys slow down the commands that access them and can trigger out-of-memory (OOM) errors.
The Real-time Key Statistics feature identifies hot keys and large keys as they appear, and retains data for the past four days. It shows the memory footprint and request frequency of each key, with high accuracy and minimal performance impact.
To view hot key and large key statistics, go to the instance details page and choose CloudDBA > Real-time Key Statistics in the left-side navigation pane. For more information, see Use the real-time key statistics feature.
Offline key analysis
Offline Key Analysis processes Redis Database (RDB) snapshot files without affecting running instances. It supports all data structures, instance architectures, and Redis versions that Tair (Redis OSS-compatible) supports.
Performance characteristics:
| Scenario | Performance |
|---|---|
| Mix of 10% large keys and 90% small keys | 4x faster than redis-rdb-tool |
| Medium and large keys | 20x faster than redis-rdb-tool |
| Memory usage during analysis | Under 1 GB (prevents OOM errors) |
The feature also lets you search for the longest sub-element within a key to pinpoint the source of large key issues.
To run an offline key analysis, go to the instance details page and choose CloudDBA > Offline Key Analysis in the left-side navigation pane. For more information, see Use the offline key analysis feature.
Instance diagnostics
Instance diagnostics combines performance metrics, slow query logs, and key analysis into a one-stop health check. Each diagnostic report evaluates an instance against criteria such as performance levels, request distribution skew, and slow query patterns, and provides actionable suggestions.
To run a diagnostic report, go to the instance details page and choose CloudDBA > Diagnostic Reports in the left-side navigation pane. For more information, see Create a diagnostic report.