Performance

Measured benchmarks for the 7 workflow tools, API latency baselines, and caching strategy. Tested against a Qualys tenant with 89K assets and 282 test cases.

Tool Benchmarks

Typical latencies measured against a production Qualys environment:

ToolTypical LatencyNotes
security_overview (quick)~1.7sCSAM-heavy, cached
security_overview (full)~8-10sAll sources, parallel
check_compliance (cached)~2msCached compliance data
check_compliance (cold)~5-12sPC API + CSAM
assess_risk (containers)~3sContainer image scan
assess_risk (all)~5-10sAll domains in parallel
assess_risk (cloud)~5-8sParallel AWS/Azure/GCP
assess_risk (certs)~2-5sCertView API
investigate (CVE, quick)~10sKB + basic asset check
investigate (CVE, standard)~15-20sKB + CSAM + threat intel
investigate (CVE, deep)~30-45sAll sources + summary
plan_remediation~3-8sPM + CSAM parallel
reports (list)~2-5sReport API
cache_status<1msMemory lookup

All standard-depth responses complete under 15 seconds. Deep investigations may take up to 45 seconds when querying all sources.

Caching Strategy

Tiered in-memory cache eliminates redundant API calls. Each tier reflects the volatility of its data:

CacheTTLImpact
Bearer token3.5 hoursEliminates auth overhead
KB entries1 hourInstant QID/CVE lookups on repeat
VMDR detections5 minutesFast threat intel queries
WAS findings10 minutesInstant web app queries on repeat
Scanner list5 minutesFast scanner health checks
ETM results1 hourInstant ETM queries on repeat

Concurrency

ThreadPoolExecutor(max_workers=8) runs independent aggregator calls in parallel within each workflow invocation. Cloud providers (AWS, Azure, GCP) are fetched concurrently. Typical parallel dispatch: 3-8 aggregator calls per workflow invocation.

Cache Warmup

On startup, _warmup_vmdr_cache() runs in a background thread to pre-populate the VMDR detection cache. This ensures the first real query hits warm caches rather than waiting for cold API calls.

Request Deduplication

The _get_or_fetch() helper with per-key locking prevents duplicate API calls when multiple aggregators request the same data concurrently within a single workflow invocation.

API Latency Baseline

Typical single-request latencies to Qualys APIs:

API / EndpointTypical Latency
CSAM v2 count0.2-0.5s
CSAM v2 search (100 assets)0.5-3s
VMDR KB lookup (single QID)0.5-1s
ETM report list/detail0.5-3s
WAS findings5-30s
Container images1-3s
TotalCloud connectors0.5-2s per provider
CDR findings1-5s
PM jobs1-3s
Scanner appliance list1-3s
FIM/EDR events1-5s
CertView certificates1-5s