SUPPORT ON V12

Monitoring that runs in your network. Proactive support from day one.

VDURA SentinelTM is part of the V12 software package, running inside your environment. It polls your VDURA storage with read-only credentials and gives your team a full local dashboard: fleet health, capacity, alerts, and per-system detail.

Talk to VDURA How it works

✓ Runs entirely inside your network ✓ Read-only, outbound-only ✓ Every category can be switched off

WHERE IT LIVES

Inside your network, on your terms.

Sentinel polls your storage read-only and drives a dashboard your team owns. The only path out is a single outbound push, filtered by your category toggles, and any category, or all of it, can be switched off at any time.

THE SERVICE

Sentinel running inside the network, polling read-only.

THE DASHBOARD

Fleet health, capacity, alerts, and per-system detail, all local.

THE TOGGLES

Per-category sharing controls and the payload inspector.

THE PAYOFF

A fault becoming a service request on its own.

ALWAYS WATCHING YOUR NETWORK

PART OF V12

Sentinel

YOUR TEAM SEES

Local dashboard

READ-ONLY POLL ↓

Realm A

Realm B

Realm C

VDURA support
cloud

Automatic service requests & Slack notifications

OPTIONAL, FILTERED, TOKEN-AUTHENTICATED HTTPS →

One outbound push, on by default and yours to switch off. No inbound path, ever.

HOW IT IS BUILT

Read-only, outbound-only, and reviewable.

No way in

Sentinel authenticates with read-only credentials and never modifies your systems. There is no inbound access, no VPN, and no listener into your network.

Every category, its own switch

Capacity, node inventory, alerts, and deep-dive data are each toggled individually. Sharing is on out of the box so support starts on day one, and any category can be switched off at any time.

See the payload before it leaves

A payload inspector shows the exact JSON of the next outbound push before it sends, and every push, successful or failed, is logged with its payload retained for review.

Read the source yourself

The collector is dependency-free, readable Python. Your security team can review it directly rather than taking our word for it.

File data, credentials, user and workload information, and logs are never transmitted under any setting.

YESTERDAY'S SUPPORT MODEL

You notice. You file. You wait. Repeat.

The customer is the monitoring system

Something breaks, and the clock starts only when a human notices. On a busy fleet, that can be hours after the telemetry knew.

The ticket starts from zero

You open a request, describe the symptom, and wait for a technician to ask for the logs you could have sent on day one.

Diagnosis by correspondence

Each round trip adds a day. The fix was often known on the first day, three emails ago.

PROACTIVE SUPPORT

Support starts before the ticket.

Out of the box, VDURA sees current health, capacity, and alert data across your fleet and can flag issues before they become tickets, with no waiting for you to pull logs or describe symptoms. Fleet-informed analysis reads what arrives against every failure pattern we have resolved across thousands of systems and more than 1,000 tests per software release, and returns a summary, a trend view, and recommended actions in plain language.

1 OPTIMAL DIAGNOSTIC

IN YOUR NETWORK

Sentinel polls your VDURA storage with read-only credentials and keeps the local dashboard current: fleet health, capacity, alerts, and per-system detail.

2 ANALYZE

ON BY DEFAULT

What is shared is read against every failure pattern we have resolved across thousands of systems, with a summary, trend view, and recommended actions returned in plain language.

3 ALERT

AS IT HAPPENS

The service team is alerted with the diagnosis already attached, and alerts can land in your Slack and email as they happen.

4 RESOLVE

NO BLANK TICKETS

A service request opens on your behalf, with parts dispatch where hardware is at fault. The first human to look starts from a diagnosis, not a blank ticket.

YOUR DATA, YOUR CHOICE

Shared by decision.

Sentinel ships with sharing on, so proactive support works from day one. Every category is individually controllable, and switching any of it off, or all of it, is one setting, not a support call.

THE DEFAULT

Proactive support on

Health, capacity, and alert data flows to VDURA so support starts before a ticket exists. Filtered by category, inspectable before it sends.

YOUR CALL

Trim the categories

Switch off any individual category, capacity, inventory, alerts, or deep-dive data, and keep the rest flowing.

YOUR CALL

Go local only

Switch sharing off entirely. The local dashboard, alerts, and history keep working inside your network, nothing transmitted.

Pause sharing or stop the service at any time. Local monitoring keeps working, and your storage is unaffected either way.

WHY THE ANALYSIS IS GOOD

Your failure has almost certainly happened before. We were there.

The analysis behind Sentinel is built from decades of production history: thousands of systems in the field and more than 1,000 tests run against every software release, including deliberate corner-case failure injection. When a shared fault arrives, it is read through the lens of every fault like it we have ever resolved, and the summary, trend view, and recommended actions arrive written in plain language rather than five thousand lines of log.

Decades

of production history behind the analysis

1,000+

tests run against every software release

Thousands

of systems in the field

LIFECYCLE AND SUPPORT

One contract, and the watch never ends.

One Contract, 24×7

Hardware, software, and support on one contract, with 24×7 expert response and OEM support for compute nodes.

10-Year, Risk-Free Coverage

VDURACare Premier combines 10-year, risk-free coverage with 24×7 expert response, hardware, software, and maintenance under a single contract.

Independent Upgrades

Software moves on its own cadence; release notes and a proposed window arrive in advance, and you approve the window.

Custom Terms

Customized offerings where a deployment has specific requirements.

WHAT THIS MEANS FOR THE BUSINESS

Proactive support is an SLA you can resell.

Your Tenants Never See It

Operators see Sentinel; tenants never do. Faults are found and fixed behind the SLA you sell.

A Better SLA Than You Could Offer Alone

Fleet-informed analysis and proactive dispatch back the availability commitments you make to customers.

The First Look Is a Diagnosis

By the time anyone on either side reads the ticket, several levels of analysis have already run.

A Lean Team Stays Lean

Nobody staffs a screen. The watch is continuous, and people engage only when there is something worth engaging.

Security Review Passes

Read-only, outbound-only, category-level off switches, and source your team can read. The questions that stall deals are answered on the page.

Works With Your Stack

Everything Sentinel sees, your dashboards can see too. It is all API-driven.

Observability →

THE WORKLOADS ON WATCH

Sentinel watches the platform. Here is what runs on it.

Proactive support is one part of the platform. The rest is what your accelerators are actually doing with it.

AI SOLUTIONS AI Training Proactive dispatch matters most mid-run. A failed node on day nine of a training job is the failure Sentinel is built to catch first. Storage for training → AI SOLUTIONS AI Inference Serving runs forever, so the watch has to as well. Latency SLAs you sell are backed by faults found before your tenants feel them. Storage for inference → AI SOLUTIONS Persistent Context Context that survives restarts depends on a healthy flash tier underneath it. Sentinel is what keeps that tier honest. Persistent context →

See what your fleet looks like on watch.

Talk to VDURA Book a Briefing