Overview
Go beyond simple data stream lookups. This agent analyzes your Amazon Kinesis data to answer complex questions like "Why are we seeing data processing delays?" or "What patterns are emerging in our stream data?"
**Capabilities:**
- Analyze stream throughput and identify processing bottlenecks
- Surface data quality issues and anomaly patterns
- Compare stream performance across shards and partitions
- Identify latency spikes and their root causes
- Synthesize insights from stream records and processing metrics
[{"type": "user", "content": "Why are we seeing increased latency in our Kinesis data streams? Processing times have doubled in the last week."}, {"type": "thinking", "content": "Analyzing stream performance patterns...", "thinkingSteps": ["Retrieving stream metrics from last 14 days...", "Calculating latency by shard and partition...", "Identifying throughput bottlenecks...", "Comparing to baseline performance..."]}, {"type": "assistant", "content": "I analyzed your Kinesis stream data and found the root causes of the latency increase:\n\n## Performance Analysis\n\n| Metric | Current | Baseline | Change |\n|--------|---------|----------|--------|\n| Avg Processing Latency | 2.8s | 1.2s | +133% |\n| Records/Second | 4,200 | 8,500 | -51% |\n| Error Rate | 3.2% | 0.8% | +300% |\n\n## Root Causes\n\n**Primary Bottleneck:** **Shard 3** is experiencing 85% capacity utilization (vs. 40% target). It's processing 12,000 records/sec (limit: 14,000), causing throttling. The shard is handling 3x the volume of other shards due to uneven partition key distribution.\n\n**Secondary Issue:** **Error rate** increased to 3.2%, causing retries and backpressure. 68% of errors are 'ProvisionedThroughputExceeded' exceptions from Shard 3.\n\n**Pattern Identified:** Latency spikes occur during **peak hours** (10 AM - 2 PM), correlating with increased traffic from the mobile app release.\n\n**Recommendation:**\n1. Reshard immediately - split Shard 3 into 2 shards to reduce capacity to ~40%\n2. Redistribute partition keys to balance load across shards\n3. Implement auto-scaling for peak hours\n\n<ENTITY_ANNOTATION id=\"STREAM-USER-EVENTS\" title=\"User Events Stream\" integrationType=\"AMAZON_KINESIS\" />\n\n<CTA label=\"View Stream Metrics\" action=\"Show detailed stream performance dashboard\" />\n<CTA label=\"Generate Optimization Plan\" action=\"Create a stream optimization action plan\" />"}]
