Performance & capacity
Utilization versus saturation analysis, capacity planning and reporting, latency diagnosis, and workload characterization.
Systems performance · Capacity planning · Observability & telemetry
David Graham
Systems performance and capacity engineer with 25+ years spanning capacity planning, real-time performance monitoring, and large-scale telemetry platforms. Built monitoring that protected multi-billion-dollar transaction paths, led capacity programs for distributed systems, and operated analytics pipelines across AWS, Snowflake, Splunk, and ELK. Now focused on observability, performance tuning, and capacity engagements for engineering and platform teams—method-driven diagnosis, instrumentation strategy, and measurable cost and latency outcomes.
- Focus
- Performance, capacity, observability
- Available for
- Consulting engagements
- Based
- Minneapolis
01 — Practice
How I work
Observability & telemetry
Metrics pipelines, logging platforms, high-volume event processing, and dashboard and alert design. Instrumentation that supports diagnosis, not vanity dashboards.
Platforms & delivery
Splunk, Elasticsearch/ELK, Kafka, NiFi, Snowflake, and the AWS analytics stack. Prometheus and OpenTelemetry are in progress. Prior consulting-practice ownership and cross-team platform ownership.
02 — Selected impact
Selected outcomes
-
Outage response: ~1 hour → minutes
Real-time transaction performance monitoring (~$2B/day)
-
Platform compaction: 12 → 5 servers
Telemetry platform re-architecture
-
10,000+ devices centralized logging
Syslog-NG
-
>1M hits/day web cluster
Apache, Oracle, and F5
-
Multi-team telemetry platform ownership
Splunk, AWS, and Snowflake analytics
03 — Study
Current study
- In progress: Linux Foundation LFS241 (Prometheus); Prometheus Certified Associate (PCA)
- Next: OpenTelemetry path (LFS148 / OTCA)
Credentials CISSP ((ISC)²) · GCDA
04 — Labs
Labs and notes
Labs and notes coming soon. Public code is on GitHub.
05 — Contact
Send a message
Consulting inquiries or anything else — use the form. LinkedIn and GitHub are also fine. I will reply by email; my address is not published here.