v1.1.2 — Azure Local LENS Workbook

Install

Download AzureLocal-LENS-Workbook.json from the assets below and import it into the Azure portal (Monitor → Workbooks → New → Advanced Editor → paste the JSON).

---

Azure Local LENS v1.1.2 is a quality and enhancements release focused on addressing customer feedback, improving reliability, and refining the workbook experience. It includes the following improvements:

TL;DR - Action required for existing LENS DCR deployments: Redeploy the revised ARM template using the same subscription, resource group, and DCR name to update your existing rule in place. This replaces the counter list, removing the five obsolete CSV counter paths and adding the four Cluster CSVFS paths. A new DCR is not required; existing associations continue to reference the same rule.

Update the workbook too; updating it alone does not change data collection. Back up and review any customized DCR before replacement, because settings absent from the template can be removed. See the DCR upgrade guidance for preview, update, and collection verification steps.

Adds machine-level capacity visibility for Issue #102, requested by @riemi92.

1. Physical Machine Capacity under Capacity & Performance > Hyper-V. Machine and Instance Name columns link to Azure resources; Instance Capacity Overview also uses the Instance Name heading. Physical cores, observed guest vCPUs, V:P CPU Ratio, and % of V:P Target sit together beside average host CPU usage and free utilization capacity, both displayed with % suffixes. Memory Used % uses the established Committed Bytes / (Committed Bytes + Available Bytes) * 100 formula, with a compact green tick below 80%, warning from 80% to below 90%, and red cross at 90% and above. Physical total, used (committed), and free/available GiB follow; the help text explains their meanings. These icons are visual guides, not health alerts.
2. Sortable numeric metrics, target comparison, and instance/ratio filtering. Highest V:P ratios appear first, displayed as 4:1 or 2.5:1 while retaining numeric sorting and filtering. Target vCPU:pCPU Ratio defaults to 4:1 and drives the colour-coded % of V:P Target bar; Filter by Ratio offers All and thresholds from 1:1 through 6:1. Both match Overview's controls without sharing its state. A machine-only Instance Name multi-select defaults to All and filters by instance resource ID without affecting VM inventory.
3. Conservative telemetry handling. Settled, matching snapshots reduce stale allocation counts. Missing data, changing or conflicting VM placements, ambiguous hardware mappings, and invalid memory values stay unavailable rather than becoming zero. Hardware lookup requires compatible qualified names and a unique match on both the Arc-machine and reported-machine sides before Instance Tag filtering, preventing same-short-name collisions from assigning another machine's physical cores or reported memory. Historic host selections resolve to Arc resource IDs without widening the settled metric window. The instance-level Capacity Overview table remains ARG-only.
4. Regression and live query coverage. Offline checks protect column order, filtering, identity, and formulas. The opt-in machine-capacity suite executes the exact workbook queries against live data and synthetic KQL cases for migration, duplicate samples, missing data, and sorting/filtering.
5. Consistent VM memory styling. Hyper-V VM Inventory's Avg Memory % now uses compact status icons instead of full-cell backgrounds, retaining its 80/100 thresholds. Help distinguishes Dynamic Memory demand/assigned-memory pressure, which can exceed 100%, from host memory usage and actual paging activity.
6. Corrected CSV storage sources (Issue #108). The DCR replaces five invalid CSV paths with four Cluster CSVFS latency/read/write paths: 26 paths in total. CSV and logical-disk performance series and coverage counts are separate, so OS-disk data cannot imply CSV coverage. CSV volume capacity uses validated, deduplicated Event 3002 snapshots rather than nonexistent percentage counters. Volume free-space forecasts describe logical CSV capacity, not physical S2D pool headroom. Existing deployments need both the workbook and DCR updated; see CSV storage sources for measurement scope and verification. Valid all-zero counter samples alone do not prove workload coverage.
7. Per-machine storage drill-down. Capacity & Performance > Single instance adds Cluster Shared Volumes - Per Machine Detailed Analysis and Logical Disk - Per Machine Detailed Analysis. Separate latency and IOPS charts for each source share one Machine selector, with no Storage filter. This selector does not change the existing comparison charts.
8. Clearer navigation and terminology (Issue #110). Capacity & Performance sections now appear as Overview, Multi-instance, Single instance, Hyper-V, matching the navigation help. Azure Local display headings and filters consistently use Instance Name, Instance Count, Affected Instances, and machine terminology. CSV cluster wording, AKS clusters/nodes, technical identifiers, and historical release notes retain their appropriate terminology. Active workbook help and documentation follow the same conventions.
9. VM inventory controls. Hyper-V VM Inventory defaults to descending Avg CPU % with unavailable values last. Independent minimum CPU and memory filters sit directly above the table; All retains unavailable values. These filters do not change Physical Machine Capacity or the trend charts. Single instance inventory VM names now link to the Azure portal when the existing guest-resource map has one unique matching Arc-enabled VM resource; missing or ambiguous matches remain unlinked without removing inventory rows.
10. Reviewed DCR deployment and upgrade workflow. The downloadable PowerShell deployment script provides explicit subscription selection, paginated reported-machine discovery, compatible-name and uniqueness checks, -WhatIf preview, conflict checks, intentional in-place replacement with -UpdateExistingDcr, and association rollback. Ambiguous short-name matches and duplicate reported rows remain ineligible for deployment and rollback. It does not install AMA or delete the DCR during rollback; follow the deployment guidance and review overlapping collection rules.
11. Capacity parameter and presentation fixes. Overview, Multi-instance, and Single instance safely parse unset shared maps instead of emitting invalid dynamic() expressions, including CPU/memory and storage forecasts. Hyper-V accepts both wildcard and literal All selections without dropping inventory rows. The forecast heading displays Current Avg %, percentage tables use compact icons instead of solid backgrounds, and observed vCPU counts remain neutral. Forecast icons follow the selected warning/critical thresholds; CPU and VM-memory icons retain their distinct thresholds. The six Single instance Hyper-V charts have a shared VM/virtual-disk heading and individual metric information tooltips.
12. Shared Capacity workspace selection and DCR setup guide (Issue #111). One Log Analytics Workspace selector applies across all four Capacity & Performance views, with Show DCR Setup Guide beside it. Workspace discovery is independent of the inventory subscription, resource-group and instance-tag filters, allowing telemetry workspaces in separate monitoring subscriptions. The guide is hidden by default and consolidates AMA prerequisites, host and Hyper-V counters, Event 3002, association safeguards, and maintained deployment links instead of repeating setup sections in Overview and Hyper-V. The scope note explains that the first Overview table is ARG-only and Managed Prometheus uses its separate Azure Monitor Workspace selector. Independently opened Capacity sub-templates retain the same guide and controls.
13. VM-level performance deep dive — no guest AMA required. Hyper-V VM Performance — Detailed Analysis uses Hyper-V counters collected by AMA on the Azure Local physical hosts, without requiring AMA inside the guest VM or guest Arc onboarding. These are host-observed performance metrics, not guest application or process telemetry. This final Hyper-V section adds one metric-selectable chart with independent VM and 1/3/7/14/30-day controls, defaulting to 7 days with 30-minute average buckets, metric information tooltips, and no VM selected by default. The chart defaults to an Average summary rather than summing percentages, rates, or latency across time buckets. A compact table beneath it shows Average, Min, and Max of the non-missing plotted bucket averages per series, in the selected metric units; these are not instantaneous extremes within each bucket. Investigate CPU, Dynamic Memory pressure, virtual-disk throughput/IOPS/latency, or individual guest virtual NICs without changing the Top 5 charts. Disk metrics support explicit selection of one or more virtual disks from the reporting physical machine, with a separate chart series and summary row per disk; disk ownership is operator-verified, not inferred. Machine-resource identity, complete counter pairs, sample deduplication, and null gaps protect the detail view.
14. Capacity & Performance tab. The main tab is renamed 🏗️ Capacity & Performance to reflect its capacity planning, forecasts, and performance drill-downs. Current navigation, setup paths, and documentation use the new name; internal IDs and gallery paths remain unchanged.
15. Clearer release support categories. Overview separates six-month-boundary releases into Nearing end of support, leaving missing, invalid, or future-dated release months in Unknown version. Boundary-month releases may already be unsupported depending on their official release date; the help and tooltip direct operators to verify that date. Supported and Unsupported age thresholds are unchanged.

The workbook header banner bumps from v1.1.1 to v1.1.2.

Overview storage charts occupy two separate three-column rows: CSV Volume Usage / CSV Latency / CSV IOPS, then Logical Disk Usage / Logical Disk Latency / Logical Disk IOPS. Every chart ranks only its named source and shows at most five instances, with matching coverage counts below it. CPU, memory and network share the preceding row.

Single instance uses the same three-column arrangement, with CSV usage By Volume and the other storage metrics By Machine, without a Top 5 limit. Multi-instance separates CSV and logical-disk latency/IOPS into two-column rows, followed by a full-width network chart; these charts retain all selected instances. Its storage forecast titles identify CSV Volume Free Space. Forecast calculations and Hyper-V VHD queries are unchanged by the layout split.