Tag: Observability
-
Incident Prediction and AIOps, Honestly Assessed (Infra to Data Science Series, Part 24)
Incident prediction on your own telemetry usually fails on arithmetic, not modelling. A base rate check, an honest look at AIOps, and the narrow cases where prediction actually pays.
-
Log Analysis and Clustering at Scale for Infra Telemetry (Infra to Data Science Series, Part 23)
Clustering raw log lines fails because ids and timestamps make every message unique. Mine templates with Drain3 first, then cluster the structure, and a day of logs collapses to a triage table you can read in a minute.
-
Anomaly Detection on Metrics and Time Series for Infra Telemetry (Infra to Data Science Series, Part 21)
Static thresholds on seasonal infrastructure metrics measure the time of day, not trouble. Deseasonalise with STL, score the residual with a median and MAD based z, and reach for an isolation forest when several metrics move together.
-
Monitoring Models in Production for Drift and Decay (Infra to Data Science Series, Part 18)
A served model decays quietly. Here is how to catch it with input and prediction drift checks, a KS test, and a PSI threshold you can page on, all on your own telemetry.
-
MLOps for Infrastructure Engineers, the Operations You Already Run (Infra to Data Science Series, Part 16)
Most of MLOps is the operational discipline infrastructure engineers already have. This part maps your runbooks, CI and monitoring onto a model, and names the two or three things that are genuinely new.
-
Getting Data Into Python From SQL, APIs and Monitoring (Infra to Data Science Series, Part 8)
Turn last part’s hand export into a repeatable pull. How to read infrastructure data straight into pandas from SQL, an HTTP API and JSON monitoring, parameterised and correctly typed.
-
Infrastructure Telemetry as a Dataset for Data Science (Infra to Data Science Series, Part 7)
Your monitoring data already is a dataset. This part reframes infrastructure telemetry as rows and columns, and shows why a counter has to become a rate before it can be a feature.
-
Observability for LLM Applications: Tracing and Debugging a Non Deterministic System (AI Engineering Series, Part 25)
A request level trace is the only artifact that explains why an LLM answer was wrong, slow or expensive. Here is how I instrument one with OpenTelemetry GenAI conventions and Langfuse, which attributes actually earn their storage, and what tracing costs once traffic is real.
-
Vertex AI Observability and Tracing, from Dashboard to Span (Google Cloud Gen AI Series, Part 25)
A green status code on an eight second request tells you nothing. Here is how Cloud Monitoring, Cloud Trace, and Cloud Logging on Vertex AI tell you which call was slow, what it cost, and what to never log.
-
Azure Monitor Observability for GenAI, from Metrics to Traces (Azure Gen AI Series, Part 25)
Azure GenAI observability comes in three layers: free platform metrics, per request diagnostic logs in Log Analytics, and OpenTelemetry traces in Application Insights. Here is what each one sees, what it costs, and the order I turn them on.
-
Azure AI Foundry Evaluation and Observability, from CI Gate to Live Traffic (Azure Gen AI Series, Part 23)
Evaluation scores catch a bad agent; tracing tells you why it went bad. Here is how Microsoft Foundry runs the same evaluators at dev time, in your CI gate, and against live traffic, and what continuous evaluation actually costs.
-
Amazon Bedrock Observability with CloudWatch and Invocation Logging (AWS Gen AI Series, Part 25)
Bedrock ships almost no history by default. Here is how I turn on model invocation logging, pick the CloudWatch metrics worth an alarm, and pull token cost per model straight from the logs.
Architect’s Toolkit
PJ’s Tools
- Infra 360 Hub – All Series
- VCF 9 Interactive Walkthroughs
- VCF Design Cheatsheet
- VCF Upgrade Planner
- VCF 9 Series Hub
- VCF Deployment Hub
- AI Stack Hub
- AI Infra Sizing & Cost Calculator
- LLM & RAG Cost Calculator
- DrJhaGPT – Ask Pranay
VMware Cloud Foundation
- VCF Documentation
- VCF 9 Planning & Preparation Workbook
- VCF Bill of Materials (BoM)
- VMware Compatibility Guide
- VMware Interoperability Matrix
- VMware Configuration Maximums
- VMware Ports & Protocols
- VMware Hands-on Labs
- RVTools Download
Nutanix
AI & Cloud-Native Platform
- NVIDIA Build (Model Catalog)
- NVIDIA AI Enterprise Reference Architecture
- NVIDIA NIM Performance Benchmarking
- NVIDIA NGC Catalog
- NeMo Microservices Helm Chart
- Helm Charts Repository
- Hugging Face Models
Architecture & Design
About the Author
Dr Pranay Jha
Dr. Pranay Jha is a Cloud and AI Consultant with 18+ years of experience in hybrid cloud, virtualization, and enterprise infrastructure transformation. He specializes in VMware technologies, multi-cloud strategy, and Generative AI solutions. He holds a PhD in Computer Applications with research focused on Cloud and AI, has published multiple research papers, and has been a VMware vExpert since 2016 and a VMUG Community Leader.
You May Have Missed
DrJha