Outbound-Only Telemetry0 Inbound PortsZero Client Media Bin Access
SESSION-CONTINUITY OBSERVABILITY

Storage Observability Engineered for Commercial Edit Suites

Protect $800–$1,000/hr finishing sessions from dropped frames, eliminate uncoordinated scrub slowdowns, and track ingest runway across Linux, TrueNAS, and Unraid.

Agent v1.4.2Sub-35MB RAM<0.5% CPUZero Inbound PortsRead-Only Agent

Native Support for Creative Storage Arrays & White-Box Hardware

45Drives Houston / StorinatorTrueNAS SCALE & COREOpenZFS & Linux MDRAIDUnraid & Debian Server
FACILITY VULNERABILITIES

The 3 Fatal Weaknesses of Creative NAS Arrays

Standard NAS alerts fire when a drive is already dead. In commercial post-production, silent degradation during active client review passes ruins deliverables.

Silent Sector Decay & Parity Crashes

Enterprise drives accumulate uncorrectable read errors (UREs) and bad sectors over weeks. Too often, the first time a controller flags them is during a grueling rebuild—when a secondary drive drops offline and corrupts the entire pool.

Machine Room Thermal Creep & Platter Wear

Server rack ambient temperatures surge when finishing suites kick off overnight render queues. Generic monitoring tools either wake sleeping platters 50+ times a day or miss nocturnal 50°C+ heat spikes completely—accelerating drive degradation.

Uncoordinated Scrubs & Dropped Frames

There is zero tolerance for I/O stalls hours before master delivery. When an uncoordinated ZFS scrub or drive resilver cuts pool read throughput by 60% during a live 4K client review, playback stutters and edit bays freeze with zero warning.

ARCHITECTURAL COMPARISON

Generic Monitoring vs. Pillar Fleet Control

Generic IT tools treat post arrays like database clusters. And custom cron scripts wake sleeping platters 50 times a day while remaining blind to block queue stalls.

The “I Can Script This with smartmontools” Trap

Why Custom Python & Bash Scripts Break Down in a Post Facility

Your Custom smartmontools Script:
  • Wakes cold nearline platters every hour, wearing spindle bearings and inflating power draw.
  • Spams Slack/email with raw hex logs whenever an irrelevant attribute fluctuates.
  • Completely blind to real-time block-layer queue depth and 100ms I/O await spikes.
  • Cannot model ZFS copy-on-write 80% allocation collapse cliffs before playback freezes.
Pillar Fleet Control:
  • Invokes -n standby to preserve sleeping disks without waking platters.
  • Surfaces p99 service time spikes against the 41.7ms 24fps client playback deadline.
  • Calculates rolling 30-minute velocity deltas to catch scrub collisions with morning sessions.
  • Predicts exact hour of ZFS metaslab fragmentation collapse with non-disableable safety floors.
Scroll horizontallyPillar Fleet Control
Diagnostic VectorDIY smartctl CronDatadog / PrometheusPillar Fleet Control
Zero-Latency Outlier Isolation (p99 Tail)
Identifies 140ms disk service stalls during 24fps/60fps playback
✕ Blind to I/O queue! Averages mask p99 spikes✓ 15s Outlier Isolation
Platter Spindown & Standby Safety
Preserves low-power sleep on cold archive without waking heads
✕ Polling wakes platters✕ 50+ spin-ups/day✓ -nocheck standby safe
Smart Scrub Windowing Guardrails
Prevents monthly parity checks from choking morning client reviews
✕ Unaware of schedule! Naive linear math✓ Rolling 30m delta warnings
ZFS CoW 80% Degradation Runway
Forecasts performance collapse cliffs before 35TB multi-cam drops
✕ Only alerts at 90% full! Linear disk alert✓ 80% CoW cliff modeling
Facility Security & TPN Footprint
Auditor-approved read-only container with zero footage bin mounts
! Needs root shell cron✕ Heavy ~250MB daemon✓ Read-Only Root / Port 443
THE SESSION-CONTINUITY ENGINE

Four Pillars Engineered for Post-Production Storage

Eliminate finger-pointing between editors, network engineers, and storage leads when playback stutters during client reviews.

24fps Budget: 41.7ms

Edit Suite Drop-Frame & Latency Attribution

When 4K ProRes playback stutters, editors immediately blame the storage array. Pillar measures kernel block-layer await times and isolates the critical p99 tail latency. Within 15 seconds, you know whether a single disk is choking on sector retries (140ms await) or if the SAN is clean—exonerating 10GbE network switches and client Thunderbolt docks.

Telemetry: /proc/diskstatsZero-Root Sampling
Scrub Conflict Engine

Predictive Scrub Windowing: Morning Session Conflict Alerts

Automated scrubs consume up to 70% of array read bandwidth. Naive linear ETAs fool engineers into assuming an overnight scrub will finish before 9:00 AM client reviews. Pillar models variable scrub velocity to forecast maintenance collisions with active edit hours—giving your team hours of advance notice to throttle or pause parity checks before sessions begin.

Schedule: Active Edit HoursPredictive Intelligence
CoW 80% Allocation Cliff

Ingest Velocity & 80% ZFS Degradation Forecasts

Post-production ingest is notoriously lumpy—a 3-day multi-cam shoot can unload 35TB in 48 hours. Copy-on-write filesystems like ZFS suffer severe throughput collapse once pool capacity crosses 80% due to block allocator fragmentation. Pillar models multi-window velocity (24h/7d/30d) to project the exact date your array hits the 80% performance cliff.

Runway: Multi-Window (24h/7d/30d)Tiered Storage Breakdown
Non-Disruptive Standby

Safe Pre-Failure Detection with Zero Spindown Disruption

Detect raw SMART bad sector growth curves, SAS/SATA CRC bus errors, and machine room heat spikes before a drive drops offline during a conform pass. Built-in standby checks guarantee sleeping nearline platters stay at rest. The dual-drive URE stress calculator forecasts secondary rebuild failure risks before you pull a physical drive.

Drive Microcode: SAS/SATA/NVMeZero Platter Wakeups
ZERO-TRUST FACILITY ISOLATION

“Don't Touch My Footage Until I Know What It Does”

Auditor-approved read-only container architecture designed to pass MPA & Trusted Partner Network (TPN) zero-trust security audits without touching project media bins.

Data Flow & Isolation Architecture
01. Studio Storage Host
TrueNAS / ZFS / Linux
/proc/diskstats :ro
/sys/block :ro
/mnt/tank (ZERO ACCESS)
Outbound Only (Port 443)
Encrypted Telemetry Push
0 Inbound Ports
Zero Listening Sockets
HTTPS Metric Heartbeat
02. Pillar Control Plane
Real-Time Fleet Dashboard
p99 Frame Spikes
Scrub Window Alerts
80% Ingest Forecasts
🛡️ZERO FOOTAGE ACCESS

Audit Our Container Privilege Footprint Before You Run It

Every flag is locked to read-only interfaces. No media dataset mounts, no elevated root capabilities, no listening network ports.

# Security audit preview: read-only root, dropped caps, zero footage mounts<YOUR_STUDIO_API_KEY> issued in dashboard during pilot setup
$
docker run -d --name pillar-agent \ --restart unless-stopped \ --read-only \ --cap-drop=ALL \ --cap-add=SYS_RAWIO \ --tmpfs /tmp:rw,noexec,nosuid,nodev,size=64m \ -v /dev:/dev:ro \ -v /proc/diskstats:/proc/diskstats:ro \ -v /sys/block:/sys/block:ro \ -v /sys/class/block:/sys/class/block:ro \ -v /sys/class/enclosure:/sys/class/enclosure:ro \ -v pillar-agent-data:/etc/pillar \ -e PILLAR_API_KEY=<YOUR_STUDIO_API_KEY> \ ghcr.io/pillarinteractive/pillar-agent:latest
Sub-35MB RAM<0.5% CPUOutbound TLS 443 OnlyZero Media Bin Access
PRE-FLIGHT READINESS ENGINE

Test Your Storage Bus Before Deployment

Run our 100% non-destructive audit script directly in your storage terminal. Validates kernel VFS mounts, SES chassis bays, and zero-inbound connectivity.

Pre-Flight Validator (100% Read-Only Audit)
$curl -sSL https://pillarinteractive.com/test | sh

=== Pillar Fleet Control :: Pre-Flight Readiness Audit ===

Running non-destructive inspection of local storage bus and runtime...

[•] Host Platform: Linux (x86_64)

[✓] User Privileges: Root / Host Administrator (Docker launch permitted)

[•] Workload Engine: Active (Docker daemon ready)

[✓] Security Posture: No docker.sock mount required (--cap-drop=ALL --cap-add=SYS_RAWIO)

[✓] Kernel VFS Diskstats: Available (/proc/diskstats)

[✓] Sysfs Block Tree: Available (/sys/block)

Storage Bus Topology & Physical Media:

[✓] Enclosure Services: SES Active (/sys/class/enclosure detected for bay mapping)

[✓] Topology Symlinks: Found (/dev/disk/by-path)

• /dev/sdb -> Samsung SSD 870 (~465 GB)

• /dev/sdc -> Samsung SSD 870 (~465 GB)

• /dev/sdd -> WDC WD60EFAX-68J (~5589 GB)

• /dev/sde -> ST8000VN004-3CP1 (IronWolf) (~7452 GB)

• /dev/sdf -> ST8000VN004-3CP1 (IronWolf) (~7452 GB)

• /dev/nvme0n1 -> Micron 7450 Pro Enterprise NVMe (~1920 GB)

Excluded 3 virtual device(s) (ZVOLs / DeviceMapper / MDRAID)

Hardware Safety & Spindown Preservation:

[✓] Spindown Policy: Active (--nocheck=standby prevents waking sleeping platters)

[✓] Storage Invariant: Strictly read-only (zero writes to media pools or footage datasets)

Outbound Network Connectivity:

[✓] Outbound Fleet API: Connected (HTTPS :443 Handshake verified)

[✓] Firewall Policy: Zero inbound ports or port forwarding required

Audit Complete: 6 physical drives detected. System is READY to run Pillar Agent.

The Unifying Principle: Radical Transparency Over Hype

Pillar observes, models, and alerts. It will never issue an unsolicited write command, kill an active process, pause a scrub autonomously, or touch your pool configuration. You remain 100% in control of your storage.

Why TPN and Studio Auditors Love Pillar: The container runs with --read-only root and strictly zero filesystem access to editorial media bins, client masters, or DaVinci databases.

Inspect the Exact JSON Telemetry Leaving Your Server

Review the exact payload sent over HTTPS (Port 443) every 10 seconds. Notice what is present—and more importantly, what is physically absent.

Outbound Diagnostic Payload (TLS 1.3 / Port 443)
{
  "schema_version": "2026-03-01",
  "agent_semver": "1.4.2",
  "timestamp": "2026-09-07T22:20:00.142Z",
  "node_id": "nas-01.chicago.pillar",
  "kernel_source": "/proc/diskstats:ro",
  "telemetry_type": "block_io_queue",
  "disks": {
    "sdc": {
      "reads_completed": 1849201,
      "read_await_ms": 148.2,
      "service_time_ms": 140.0,
      "queue_depth": 14.2,
      "util_pct": 99.8,
      "p99_spike_flag": true,
      "outlier_risk": "HIGH_DROP_FRAME_RISK"
    },
    "sdd": {
      "reads_completed": 921402,
      "read_await_ms": 3.8,
      "service_time_ms": 3.2,
      "queue_depth": 1.1,
      "util_pct": 24.5,
      "p99_spike_flag": false,
      "outlier_risk": "NOMINAL"
    }
  },
  "security_audit": {
    "media_mounts_accessed": 0,
    "project_files_read": 0,
    "payload_bytes": 1280
  }
}
✓ Zero Project Data:Only anonymous kernel counters and block latencies leave your server.
Full Schema Spec
STORAGE CONTINUITY PRICING

Transparent Pricing. Unlimited Terabytes on Every Tier.

In post-production, expanding from 100TB to 300TB happens because of 6K/8K RAW ingest—not because your team tripled. Zero per-terabyte penalties. Zero surprise metric overages.

The Editorial ROI: Continuity Insurance vs. Billable Suite Downtime

A commercial finishing suite with agency creatives and directors bills at $800–$1,000/hour ($8,000/day). A single 2-hour playback freeze or uncoordinated scrub crash costs more in lost client confidence and overtime than 2 full years of Pillar.

1 Prevented IncidentPays for 24+ Mo.
Monthly BillingAnnual Billing2 Months Free (~17% Off)

Boutique

For small editorial shops & suites

Starter
$149/ mo

Billed monthly • Up to 3 nodes

Unlimited Terabytes
Up to 3 Storage Nodes
Up to 48 physical disks
30-Day Telemetry Retention
Drop-Frame Attribution
Slack, Discord & Email alerts
Up to 3 team members
Recommended for Commercial Facilities

Studio

For multi-suite post houses

Popular
$299/ mo

Billed monthly • Up to 10 nodes

EVERYTHING IN BOUTIQUE, PLUS:
Up to 10 Storage Nodes
Up to 150 physical disks
90-Day Telemetry Retention
Smart Scrub Windowing
80% CoW Runway Velocity
Up to 10 team members

Facility

For enterprise media groups

High-Scale
$599/ mo

Billed monthly • Unlimited nodes

EVERYTHING IN STUDIO, PLUS:
Unlimited Storage Nodes
Unlimited physical disks
1-Year Telemetry Data Lake
Unlimited team members
Founder Architecture Review
Multi-Tenant Console

Consultant / MSP

For IT integrators & TDs

$899/ mo

Billed monthly • Up to 10 clients

INTEGRATOR & MSP POWERS:
Up to 10 Client Studios
Centralized Fleet Switcher
Co-Branded Client Health Reports
1-Click Client Onboarding
Direct Integrator Slack Channel
FREQUENTLY ASKED QUESTIONS

Quickstart Technical FAQ

Does Pillar support SAS controllers and JBOD expansion chassis?

Yes. Pillar natively supports direct SATA, SAS, and NVMe block devices across LSI, Broadcom, and Microsemi Host Bus Adapters (HBAs) flashed to IT mode (Initiator Target). In JBOD topographies, our collector tracks both physical enclosure slots and OS device identifiers (/dev/sd*, /dev/nvme*n*).

How do I deploy Pillar on TrueNAS SCALE or Unraid?

Pillar is packaged as a standard, multi-architecture OCI Docker container image. On TrueNAS SCALE, deploy it via the “Custom App” interface with read-only host path volume mounts for /proc/diskstats and /sys/block. On Unraid, paste the 1-line Docker command into the web terminal or use the Unraid Community Apps container template. For step-by-step guides, systemd unit files, and persistent init scripts, see our Platform Deployment Guides.

How do I cleanly uninstall the agent if we decide not to proceed?

Complete uninstallation takes 5 seconds: run docker stop pillar-agent && docker rm pillar-agent. Because Pillar runs inside a strictly read-only container without host root filesystem hooks or cron modifications, zero residual files or background daemons remain on your server. See Troubleshooting & Lifecycle for de-provisioning details.

Why don't you bill by terabytes like cloud monitoring vendors?

In post-production, a single 3-day multi-cam commercial shoot can generate 35TB of camera RAW footage. Charging by the terabyte penalizes facilities for doing their work and creates unpredictability. We charge exclusively by the number of monitored storage appliances and team capabilities—your terabytes are always unlimited.

Can Pillar execute commands or modify my active storage pools?

No. Pillar is engineered strictly as a non-interfering observability engine. The container runs with --read-only, dropped capabilities (--cap-drop=ALL), and strictly read-only volume bindings. Pillar will never issue an unsolicited write command, kill an active process, pause a scrub autonomously, or touch your pool configuration. Your storage administrators remain 100% in control. Read our Security & Network Whitepaper for complete container audit specifications.

Deploy on Your Secondary or Ingest Array in Under 60 Seconds

Eliminate dropped frames in client reviews and uncoordinated rebuilds. Start your 14-day free pilot with zero credit card required.