The dog days of summer have done nothing to slow the pace over at Dell PowerScale. Hot on the heels of the OneFS 9.15 release comes the arrival of the latest and greatest PowerScale InsightIQ 6.4 monitoring and reporting release.
As a quick refresher, InsightIQ provides robust health, performance, and reporting capabilities that help maximize PowerScale cluster efficiency, including advanced analytics to optimize applications, correlate cluster events, and accurately forecast future storage requirements.
So what goodness does this InsightIQ 6.4 release add to the PowerScale metrics and monitoring mix?
New functionality includes:
| Feature | IIQ 6.4 Functionality |
| Top Talkers Report | New report type surfacing the top N clients slowing a cluster down, ranked by bandwidth, latency, and operation rate. |
| Performance Anomaly Detection | On-demand anomaly detection across the Performance Reports modules, highlighting unusual behavior directly on the existing performance graphs. |
| Dell ESE Integration | Secure, always-on connectivity to Dell support infrastructure via Connectivity Services, enabling remote troubleshooting without VPN or direct network access. |
| Telemetry | Automated, periodic collection of system health, configuration, and feature-adoption metadata, uploaded securely over the same ESE channel. No customer PII or cluster data. |
| Partitioned Performance Enhancements | New SmartQoS identification and workload-latency metrics, plus support for 1,024+ workloads per dataset. |
| AI Assistant Enhancements | Bring Your Own LLM (BYO-LLM) support, multi-session chat history, and response-feedback collection. |
| Platform & OS Support | Support for the latest OS and OneFS releases, plus N-2 version upgrade support for simplified upgrade planning. |
The new PowerScale InsightIQ 6.4 release introduces a healthy crop of enhancements aimed at improving performance visibility, supportability, and day-to-day usability. Let’s take a closer look at each of them in turn.
Top Talkers Report
Top of the bill in 6.4 is the new Top Talkers report, added to the Performance Reports family. Rather than trawling through graphs to work out which client is hammering a cluster, administrators can now see the busiest clients at a glance. The report presents bar charts of the top N clients across three brand-new modules — Bandwidth, Latency, and Op Rate — making it straightforward to pinpoint the workloads that are slowing a monitored cluster down.
Because Top Talkers lives within Performance Reports, it behaves just like its siblings: the usual filters can be applied (single-value filters are supported for these new modules), and the modules can be freely mixed and matched into custom user-defined reports.
Performance Anomaly Detection
InsightIQ 6.4 also gains on-demand Performance Anomaly Detection — a built-in capability that automatically flags unusual performance patterns on your monitored clusters. Anomalies are highlighted directly on the existing performance reporting graphs, with InsightIQ computing expected upper and lower confidence bounds for each metric and marking any data point that breaches them. It works across all performance report types, with the sole exception of multiline graphs.
Two detection modes are on offer, depending on how much history is available:
| Historic Mode | Current View Mode |
| Uses past trends and seasonality (hourly, daily, weekly) to establish a baseline. | Detects anomalies using only the data visible on the current graph. |
| Requires a minimum of one week of historical data. | Works for any selected time duration. |
| Configurable lookback window of 1, 2, or 3 weeks (default: 2 weeks). | No historical data requirement. |
| Best for steady-state workloads with predictable patterns. | Useful when historical data is limited or the cluster is newly added. |
Three sensitivity levels — Low, Medium, and High — control the strictness of the confidence bounds, with higher sensitivity flagging subtler deviations. Historic mode builds its baseline from up to three weeks of past data, learning the hourly, daily, and weekly seasonality unique to each cluster’s workload. To use Historic mode, at least one PowerScale cluster must be added, actively monitored, in a Connected state, and have a minimum of one week of collected performance data.
Dell ESE Integration (Connectivity Services)
Arguably the headline supportability feature of the release, InsightIQ 6.4 introduces integration with Dell’s backend through ESE Connectivity Services. This establishes secure, always-on connectivity between InsightIQ and Dell Technologies support infrastructure, allowing Dell support engineers to remotely access and troubleshoot an InsightIQ instance without requiring a VPN, direct network access, or customer-side coordination during the session. Once configured, a dedicated support user is automatically provisioned for those remote sessions.
Two core capabilities ride on this new channel:
- Remote Support — Dell support engineers can initiate secure remote CLI (SSH) or Web UI sessions to the InsightIQ instance, authenticated via a time-limited RSC passcode tied to an active Service Request.
- Auto Support Case Creation — InsightIQ can automatically open support cases with Dell when qualifying events are detected. This can be enabled in the 6.4 release, though it is not yet internally active.
Connectivity can be established in one of two modes, depending on whether the InsightIQ instance is permitted direct internet access:
| Connection Mode | Description |
| Connect Directly | InsightIQ connects out over the internet directly to Dell Connectivity Services. Requires outbound internet access from the IIQ instance. |
| Via Secure Connect Gateway | InsightIQ connects over port 9443 to a Secure Connect Gateway (SCG) appliance, which acts as a secure proxy to Dell Connectivity Services — useful when IIQ cannot have direct internet access. IIQ monitors only its connection to the gateway; the gateway-to-Dell link is not checked by IIQ. The SCG appliance must be acquired and set up independently. |
From a security standpoint, the auto-provisioned support user is deliberately constrained. Via the Web UI it holds a read-only role — able to view the InsightIQ interface but unable to modify any settings or data. Via CLI (SSH) it is limited to non-disruptive operations such as log collection and diagnostics; session-disrupting operations are blocked, with the sole exception of ‘system-reboot’. All remote sessions auto-terminate after 20 minutes of inactivity, and the support user’s credentials are managed internally by InsightIQ and the Dell backend, rather than by IIQ users. Prerequisites are modest: a machine-id (/etc/machine-id) must be present on Scale setups, and an Access Key and PIN are generated up front via the Dell support portal.
Telemetry
Complementing the ESE integration, InsightIQ 6.4 adds Telemetry — the automated, periodic collection of operational metrics and configuration data from InsightIQ, securely uploaded to the Dell ESE connectivity hub. Telemetry rides on the very same connectivity channel and runs quietly in the background as long as connectivity is enabled; no customer action is required after the initial setup. Data is gathered at staggered intervals, compressed into timestamped .tar.gz archives, and uploaded with MD5 checksum integrity verification. Crucially, no customer PII or cluster data content is collected — only system health, configuration metadata, and operational counters.
The collected data gives Dell health monitoring (IIQ health status, cluster connectivity, and critical alerts), deployment visibility (version, deployment mode, host specs, and cluster configuration across the installed base), and feature-adoption insight (which features such as FSA, Quotas, Dedupe, Performance Reporting, and the AI Assistant are actively in use). The collection cadence is pre-configured as follows:
| Frequency | Data Collected |
| Every 15 minutes | Overall IIQ health; clusters with data-collection failures; critical alerts in the last 15 minutes. |
| Every 1 hour | Cluster connectivity status; datastore usage %; node count; OneFS version; per-cluster node health (nodes up vs. nodes down). |
| Every 4 hours | Total node count across all monitored clusters; per-cluster feature availability (FSA, Quotas, Dedupe). |
| Every 24 hours | Alerts generated in the last 24h; storage usage (percent used, total capacity, per-cluster DB size); deployment info (version, Simple/Scale mode); host info (OS & kernel, CPU cores, RAM, IIQ storage used); alert configuration; performance-reporting configuration; and AI Assistant feedback. |
Partitioned Performance Report Enhancements
Keeping step with the latest PowerScale release, InsightIQ 6.4 extends its Partitioned Performance reporting to embrace the newest SmartQoS capabilities. Three enhancements land here:
- New identification metrics — The new Identification Metrics for performance datasets — Operation Class and File — are now displayed in the breakout values and can be filtered on as normal.
- Workload latency metrics — The new latency metrics ReadQoSDelay, WriteQoSDelay, and OtherQoSDelay appear as three new lines in the Workload Latency graph, with breakouts displayed as normal when selected from the Breakout By dropdown — a boon for SmartQoS latency analysis.
- More workloads per dataset — Support has been added for OneFS’s new configurable workload limit, so InsightIQ now handles 1,024+ workloads per user-defined dataset. Dell recommends bumping the deployment VM to 12 CPUs and 32 GB RAM to comfortably absorb the higher expected load.
AI Assistant Enhancements
The document-aware AI Assistant introduced in InsightIQ 6.3 — the intelligent companion that helps users find information, understand product capabilities, and troubleshoot InsightIQ and PowerScale issues directly within the interface — receives a trio of enhancements in 6.4:
- Bring Your Own LLM (BYO-LLM) — Enables integration with an external, OpenAI-compatible LLM. User queries can be securely routed to the configured external model for response generation, rather than relying solely on the built-in assistant.
- Chat history — Supports multiple chat sessions, letting users switch between conversations while maintaining context across interactions.
- Response feedback — Lets users rate the AI-generated responses. That feedback is gathered through telemetry to help improve the overall experience over time.
Enabling the AI Assistant is a prerequisite for all of the above. For BYO-LLM specifically, an OpenAI-compatible API endpoint, valid API credentials, and — where applicable — the required SSL/TLS certificates are needed.
Ecosystem & Platform Support
On the platform front, InsightIQ 6.4 keeps pace with the latest operating system and OneFS releases. The headline changes since 6.3 are the move to SLES 15 SP6 for Scale deployments and extended PowerScale coverage up to OneFS 9.15. The full qualification matrix is as follows:
| Qualified On | InsightIQ 6.3 | InsightIQ 6.4 |
| OS (Scale deployment) | RHEL 8.10, RHEL 9.6, RHEL 10.0, SLES 15 SP4 | RHEL 8.10, RHEL 9.6, RHEL 10.0, SLES 15 SP6 |
| PowerScale (OneFS) | v9.7 to v9.14 | v9.7 to v9.15 |
| VMware ESXi | ESXi v8.0 U3, ESXi v9.0.1 | ESXi v8.0 U3, ESXi v9.0.1 |
| VMware Workstation | Workstation 17 Free Version | Workstation 17 Free Version |
| Ubuntu | Ubuntu 24.04 Online deployment | Ubuntu 24.04 Online deployment |
| OpenStack | RHOSP v21 with RHEL 10.0 | RHOSP v21 with RHEL 10.0 |
As with prior releases, InsightIQ continues to offer the same two deployment models — the bare-metal / virtual-machine InsightIQ Scale flavour and the OVA-based InsightIQ Simple flavour on a VMware hypervisor — so existing customers can stick with the model that best suits their environment.
Upgrading to InsightIQ 6.4
One of the more welcome operational touches in this release is N-2 upgrade support, which simplifies upgrade planning by allowing a direct, in-place upgrade to 6.4 from either the 6.2.0 or 6.3.0 release — the process is identical on both Simple and Scale deployments. Beyond an existing InsightIQ Simple or Scale machine running 6.2.0 or 6.3.0, the only real prerequisite is at least 40 GB of free disk space.
At a high level, the upgrade is driven by a single script and moves through a precheck, the upgrade itself, and a post-upgrade cleanup:
- Precheck — Docker availability, an InsightIQ version check (6.2.0 or 6.3.0), free disk space, InsightIQ service status, and OS compatibility.
- Upgrade — EULA acceptance, extraction of the InsightIQ images, service shutdown, resource-limit updates, add-on and CIAM installation, and the InsightIQ service upgrade itself.
- Post-upgrade — InsightIQ metadata refresh, re-enabling of the AI Assistant (if opted in), removal of old Docker images, and clean-up of the upgrade and backup folders.
The mechanics themselves are refreshingly simple — download and uncompress the bundle, extract the upgrade scripts, and trigger ./upgrade-iiq.sh. Progress can be tracked at any time with the showupg –log and showupg –status helpers, with the full detail available in the insightiq_upgrade.log under /usr/share/storagemonitoring/logs/.
All told, InsightIQ 6.4 is a hearty release: Top Talkers and Performance Anomaly Detection sharpen the day-to-day performance story, the ESE integration and Telemetry step change the supportability experience, and the SmartQoS, AI Assistant, and platform updates round things out nicely.