Proxmox goes 24/7 on 19 October: what those two hours actually buy you

On 2 September, Proxmox announced that its enterprise support becomes 24/7 on 19 October 2026. Until now, tickets were handled Monday to Friday, 07:00–17:00 CET/CEST, on Austrian business days. It is the best thing that has happened to Proxmox inside a board meeting in two years. And the small print deserves the same attention the headline got, because what you buy there is two hours to first contact, not two hours until your cluster is running again.

Before we go on: we are not Proxmox resellers. We do not sell their subscriptions and we take no commission on them, just as we take none from VMware. We say it now so the rest reads as what it is: we think the announcement is good, and we still think the question it answers was the wrong question.

What the announcement says, unadorned

Proxmox Server Solutions GmbH, the Vienna company behind Proxmox VE, published the statement on 2 September 2026 with two announcements in the same text. First: enterprise support moves to continuous coverage from 19 October, with engineers across every time zone. It covers Proxmox VE, Proxmox Backup Server and Proxmox Datacenter Manager, and includes offline updates and key activation for regulated, air-gapped environments. The rollout is not uniform across plans:

The second announcement is Proxmox North America Inc., a subsidiary based in Kingston, Ontario, led by Bill Hughes. It handles sales and account management for US and Canadian customers and partners, contract administration, invoicing and procurement in US and Canadian dollars, and technical support during business hours across the Eastern, Central, Mountain and Pacific zones. Hold on to that last sentence; it comes back below.

Why this is more than a change of opening hours

In almost every migration project there is a meeting with the board where someone non-technical asks the only question that matters: "and what if it goes down on a Saturday at three in the morning?". Until 19 October, the honest answer was that the vendor replies on Monday morning, Austrian time. In the meetings we have sat in, that sentence has stopped more migrations than any performance benchmark, and not because it is unfair: it just reads terribly in a set of board minutes.

A couple of weeks ago we wrote about when NOT to migrate from VMware to Proxmox, and one of the reasons on that list was precisely this one. We have also looked, with numbers, at what really happened to the exodus after Broadcom bought VMware. On 19 October, one of those reasons drops off the list.

The arithmetic almost nobody does before the meeting

Proxmox VE subscriptions are not priced per node or per virtual machine: they are priced per CPU socket per year. These are the prices and terms published on their own page, consulted on 5 September 2026:

The last column is the arithmetic for a small, very common cluster: three two-socket nodes, six sockets in total. The 24/7 tier with a two-hour response sits in the bottom row, at €6,600 a year. Almost every migration calculation circulating online is done with the first row or the third.

We are not going to put a figure from the other side next to it. VMware-ecosystem licensing is negotiated case by case, and any number we wrote here would be a headline, not your invoice; that comparison has to be made against your own quote. What we will say is this: €6,600 a year is real money for a small company, and in the mid-sized businesses we work with it is less than a single day of downtime costs, which runs into tens of thousands of euros. That, and not the budget line, is the right frame for the decision.

Three asymmetries sitting inside the announcement itself

1. Response is not resolution

The Premium tier's two hours are to first response: someone reads your ticket, classifies it and replies. Nowhere does the announcement say —nor could it— that your cluster will be up two hours later. That is not a criticism: no serious vendor commits to a resolution time for a fault it has not yet seen. The problem is not in the contract, it is in the gap between what the contract says and what the person signing it understands.

2. Global support goes 24/7; local support stays office hours

Here is where that sentence comes back. In the same statement where support becomes continuous, the new Kingston subsidiary offers "business-hours technical support" across the four North American zones. It is not a contradiction: the 24/7 part is the global engineering rota, and the subsidiary is local presence for sales, contracts and invoicing in the customer's currency. But anyone who reads the two headlines and merges them into one —"Proxmox opens in North America and gives 24/7 support there"— is understanding something the text does not say. An on-call engineer on another continent is still an on-call engineer, and that is a good thing; it is just not your account manager at four in the morning.

3. The scope is three products, not your system

The coverage is Proxmox VE, Proxmox Backup Server and Proxmox Datacenter Manager. Read that as a closed list, because it is one. Outside it sit the array or the JBOD, the switch, the uplink, the UPS, the Windows running inside the virtual machine, the database, the ERP and the firewall your people come in through. Do the exercise against your own history: take the last five times somebody on your team got up in the middle of the night for an infrastructure problem, and mark how many of those pieces were on the list above. Our experience says few of them are, and that when one is, it usually arrives alongside a power cut or a change somebody made on a Friday.

What actually decides whether you get up at three

A failure is a component breaking. An outage is your company stopping. The failure is inevitable; the outage is a design decision, and it is taken months before the incident. If a node dies in the small hours and the cluster has quorum, high availability configured and replicated storage, the machines start on another node and what waits for you in the morning is maintenance paperwork. If the same node dies and the data lived in one place, no support contract on earth gets your service back before lunch.

The arithmetic is unforgiving, and worth keeping in mind when the conversation is about opening hours: 99% availability is 87.6 hours a year in the dark, more than two working weeks. 99.9% is 8.76 hours. 99.99% is 52.6 minutes. Each extra nine is not bought with an open phone line: it is bought with architecture. And that architecture has details that cost dearly when ignored; we wrote about it when we went through the small print of the 650 milliseconds corosync adds for every node you put in the cluster.

Vendor support comes in after the architecture, never instead of it. It is the piece that saves you with the rare kernel bug, the ZFS regression, the Ceph behaviour documented nowhere. What stops a dead disk from stopping the company you put in place yourself, the day you decided how many machines your data lived on. And the hour at which you stop waiting for the vendor and start restoring is not set by the vendor's contract either: it is the decision that costs most on a bad night, and the one that is almost never written down, as we described when we wrote about continuity plans nobody has rehearsed.

Six questions for reading any 24/7 support contract

They work for Proxmox's, for any other vendor's, and for your managed services provider's. Ours included: if you hire us for maintenance, ask us these six questions and demand the answers in writing.

When Premium is worth every euro

That said, there are three cases where we would not argue about the price for a minute:

The first is regulated or air-gapped environments. The announcement explicitly mentions offline updates and key activation; if your cluster never touches the internet, that is exactly what you need, and a forum will not solve it for you. The second is large Ceph clusters, where the hard problem is usually the software's own behaviour rather than a broken component: there, prioritised escalation to the engineers who wrote it is the only thing that shortens the clock, because no amount of your own redundancy fixes a bug.

And the third, the most common of the three in small companies: having nobody on call. If your company has no night shift and no provider on a contracted rota, having someone to call stops being a budget line and becomes the only line of defence there is.

What we do with this

We run Proxmox VE with Ceph storage in production across several data centres, and we go back to the 3.x branches. We have migrated companies from VMware to Proxmox and we have also recommended staying on VMware when that made sense. We sell licences for neither, which is why we can write this article without anyone having to wonder whose side we are on.

When we put together an IT maintenance contract, the vendor subscription is one piece, not the contract. The other pieces are the ones that almost never make it into a press release: who watches the telemetry and at what thresholds, who wakes up and how fast, who gets to the data centre and with which spare, who has your documentation open before picking up the phone, and who has the authority to say "we stop waiting and we restore". That is what we sell as 24/7 support, and in the consulting that comes first, half the job is writing all of it down before it is needed.

From 19 October, the board's question —"and what if it goes down on a Saturday at three?"— has an answer from the vendor. It still does not have yours.

Sources: the announcement of the move to 24/7 support on 19 October 2026, the per-plan rollout (Premium from day one with unlimited tickets, two-hour prioritised response and SLA-backed escalation; Standard during an onboarding window in Q4 2026; Basic with no SLA change and coverage extended beyond Austrian business hours), coverage of Proxmox VE, Proxmox Backup Server and Proxmox Datacenter Manager including offline updates and key activation, and the creation of Proxmox North America Inc. in Kingston, Ontario, led by Bill Hughes with business-hours technical support across the North American zones: press release from Proxmox Server Solutions GmbH dated 2 September 2026, with reporting by heise and StorageReview. Prices per socket per year (Community €120, Basic €370, Standard €550, Premium €1,100), included ticket counts and first-response times (1 business day, 4 hours and 2 hours): official Proxmox VE pricing page. A note on that page: on 5 September 2026 it still described ticket support as Monday to Friday, 07:00–17:00 CET/CEST, on Austrian business days, with first-response times "within a business day"; that is consistent, since the change takes effect on 19 October, but worth knowing if you check it today. The figures in the table's last column (€720, €2,220, €3,300 and €6,600) are our own multiplication by six sockets, not a published rate. Availability nines, the failure-versus-outage distinction and the cost of a day of downtime: everyWAN's own material. We have not independently verified the service's actual response times; we report what the vendor publishes.