0%
Tracing accuracy
close
d
e
e
p
s
h
i
e
l
d
s
APM
Content-Based Non-Intrusive Distributed Tracing
Deepshield's distributed tracing offers eBPF-powered, non-intrusive distributed tracing capabilities, providing full-link monitoring from backend components, databases, to closed-source components, with support for over twenty common application-layer protocols. It achieves full-link tracing without modifying a single line of code by analyzing the semantic correlations within request content.
Cross Protocol Tracing
- Non-intrusive eBPF-based monitoring captures raw request/response data at the kernel level without code instrumentation, eliminating deployment complexity.
- Extensive multi-protocol support (>20 protocols) via RFC-compliant templates enables accurate parsing of diverse application-layer traffic beyond basic HTTP/RPC.
- Efficient request boundary detection using length-field jump parsing avoids computationally expensive full-payload inspection.
High-Precision Request Correlation
- Dual-phase transaction affinity analysis (API traffic correlation + TF-IDF field similarity) leverages semantic context to prioritize causally related spans, reducing false mappings.
- Probabilistic fusion model adaptively weights transaction features with causality metrics (e.g., temporal proximity), maintaining accuracy even when transaction fields are absent.
- Entropy-driven automatic weight adjustment dynamically prioritizes reliable metrics, mitigating the impact of fluctuating or invalid features.
Query-Driven Trace Assembling
- Tag/metric dual-indexing compresses span storage by deduplicating shared attributes.
- On-demand iterative trace retrieval collects only spans relevant to user queries (e.g., high-latency traces), slashing unnecessary data transmission.
- Flexible query API supports complex operator-defined filters (tags/metric thresholds)
0%
Transmission overhead
decrease
decrease
0X
Increase trace assembly
rates
rates