HA Enterpise

If you are scared of eBay, you can just contact companies directly. There are many hardware resellers out there that offer warranties and support for used/off-lease hardware. In the US, for example, I've used Vibrant for 20+ years, primarily for Dell hardware. I've even had them do ownership transfers then added or extended the warranty and support through Dell directly. Dell will typically provide hardware warranty support for seven years past the last sale date. You can also purchase technical support separately if you want to handle hardware repairs yourself. For example, they offer per-incident support.

Used versus new is a fine argument, but anytime someone is concerned about cost, I will suggest used. And with regard to reliability, there just isn't a difference between the two. You can have high failure rates with new hardware too. Redundancy is required to be built in regardless.

The sealed, OEM Windows licenses on eBay are legit. I've been buying them for decades. Far better than the online key sellers since you get a physical disc and genuine MS sticker with the activation key.

Ultimately, my reason for mentioning Windows Server 2025 Datacenter was due to Storage Spaces Direct, which can be used with Microsoft's virtualization platform to build a hyper-converged cluster. The haters will chime in, but the fact is there are plenty of people running this and it's a valid solution.
 
  • Like
Reactions: Johannes S
@janus57 and @alexskysilk

Perhaps I wasn't clear.
The question in this thread is: "Can I replicate the functionality offered by a SAN using Proxmox today, given that SAN vendors stubbornly refuse to provide direct support for Proxmox?"

While running VM disks over NFS on an external ZFS server is technically possible, it lacks the throughput and low IOPS latency required to compete with a SAN.
To get comparable performance and resiliency, ZFS pools must reside directly on the hypervisor hosts using local storage and cross-node replication. Consequently, achieving HA this way demands a much higher overall disk footprint than a traditional centralized array.

This is why, compared to a traditional SAN, ZFS-based setups generally require more disks.

Since you are compareing Ceph to a SAN. I feel that you dont realize the price of a good speed all flash SAN with full PSU redudancy, switch redudancy, controller redudancy a-s-o.
If you have 4+ hosts with 24x 2.5" in the front... then instead of buying for example 24 slot SAN, with a price of what? 25-50.000dollar ex disks? or something. Get 4x extra network cards and put 8 none-oem disks in each host and woops... you have the SAN for free?

How much storage and what speed of the SAN are you looking for?
 
  • Like
Reactions: Johannes S
Hi,

The question in this thread is: "Can I replicate the functionality offered by a SAN using Proxmox today, given that SAN vendors stubbornly refuse to provide direct support for Proxmox?"
I would say : yes, but I don't know how far the integration goes because some vendors like NetAPP or DELL officially support Proxmox through their doc and support.

This is why, compared to a traditional SAN, ZFS-based setups generally require more disks.
I would say you have compared 2 things that are not equal so your comparison is flawed.


Can you tell more about your actual plan and/or cost for your infrastructure ?
If I'm not mistaken, you have, at no moment, tell us if you are going to use your existing hardware or go to buy a new one. Which can effectively make a huge difference in price.

Given the actual pricing, I don't see where a new full-HA solution (at hypervisor level) costs less in SAN than in CEPH, because for a real HA, you need at least 2 SAN (with a dual controller), which is close to a CEPH solution.

Or the real alternative between can be "blockbridge" as you mentioned.

Best regards,
 
Last edited:
  • Like
Reactions: Johannes S
The question in this thread is: "Can I replicate the functionality offered by a SAN using Proxmox today, given that SAN vendors stubbornly refuse to provide direct support for Proxmox?"
short answer- yes. longer answer- Proxmox doesnt provide a CAF the way vmware does, which means the workflow will be different. the SAN vendors arent "refusing" to directly support Proxmox, and many already provide documentation on usage within linux environments generally and pve environments specifically.

While running VM disks over NFS on an external ZFS server is technically possible, it lacks the throughput and low IOPS latency required to compete with a SAN.
Again, apples and oranges. A zfs server with an iscsi target IS a SAN. It has pros and cons just like every SAN.

To get comparable performance and resiliency
THATS the magic point. if you leave one side of the comparison ambiguous, the whole comparison doesnt mean anything. "ZFS" and "SAN" arent mutually exclusive. In context of the conversation, the storage device internal filesystem has almost no bearing on the conversation- you're evaluating features, performance, failure domains, and price. Once you define your needs you can make rational comparisons.
 
Given the costs associated with competitors, more and more companies are considering Proxmox.

But as things stand - aside from Blockbridge and Ceph (the latter being hard to recommend nowadays due to prohibitive disk costs) - are there enterprise-grade solutions available for new infrastructure deployments that guarantee stability, High Availability (HA), and snapshots?

Everyone would like at least the same guarantees as the classic nodes+SAN model..

Just for curiosity. Whare are we talking about?
How many sites? Nodes? Cluster ? Disc capacity?
If we forget licensing and money for a while. What is it you would like to accomplish?
 
I think this discussion is going off-topic.

I’m an IT consultant, and I’ve been running Proxmox in production since version 2.
I’m asking a very practical business question, not looking for a definition academic debate.

Yes, technically a ZFS server exporting iSCSI is a SAN. But in the real world:

iSCSI / LUNs on Proxmox: You lose native snapshot integration, thin provisioning management, and the seamless workflow that VMware offered with storage integration APIs.

NFS with qcow2: It simply does not deliver the same IOPS, latency, or throughput efficiency that enterprise NFS/VMDK integrations provided under vSphere.

When enterprise clients migrate away from VMware, they are looking for production-grade, vendor-supported, and certified solutions with equivalent capabilities—not operational downgrades or custom workarounds.

As of today, that 1:1 drop-in replacement for the classic Node+SAN model in Proxmox simply doesn't exist without trade-offs. That was the core point of this thread.
 
@ AkUsP
o satisfy your curiosity, we are talking about the classic SME baseline:

  • Topology: 1 Single Site, 3 Compute Nodes (HA Cluster), ~30 TB usable capacity.
What we want to accomplish for the client:

  1. High-Performance Shared Storage: Block-level performance with native, instant VM-level snapshots and thin provisioning.
  2. Zero-Impact Live Migration: Flawless VM displacement between nodes without IOPS degradation.
  3. True High Availability (HA): Instant automatic failover if a host dies, with storage redundancy and zero single points of failure.
  4. Operational Simplicity & Vendor Support: A certified, production-grade stack that allows maintenance and updates without complex workarounds.
In VMware with a traditional SAN, this setup was a "set-and-forget" industry standard for 15+ years.

Ceph is definitely a mature and viable alternative today (even NAS vendors like QNAP integrate it now), and it solves the shared storage problem natively in Proxmox.

My original question was simply: Aside from Ceph and Blockbridge, are there other enterprise-grade, shared storage solutions (like traditional SAN integrations) that deliver this exact same seamless experience in Proxmox today?
 
Last edited:
Operational Simplicity & Vendor Support: A certified, production-grade stack that allows maintenance and updates without complex workarounds.

If you had asked that first we could have saved the entire thread :) you know the answer. why bother with the rest of the conversation? As to why- ask your suppliers. You're in the wrong forum- Nutanix is that way.
 
If you had asked that first we could have saved the entire thread :) you know the answer. why bother with the rest of the conversation? As to why- ask your suppliers. You're in the wrong forum- Nutanix is that way.
Luckily, not everyone thinks the way you do.... In fact, the thread is becoming interesting.
I repeat... I’ve been using Proxmox in production since 2012, and I’ve also been an Authorized Reseller—since 2013, I believe... you too?
 
Last edited:
@ AkUsP
o satisfy your curiosity, we are talking about the classic SME baseline:

  • Topology: 1 Single Site, 3 Compute Nodes (HA Cluster), ~30 TB usable capacity.
What we want to accomplish for the client:

  1. High-Performance Shared Storage: Block-level performance with native, instant VM-level snapshots and thin provisioning.
  2. Zero-Impact Live Migration: Flawless VM displacement between nodes without IOPS degradation.
  3. True High Availability (HA): Instant automatic failover if a host dies, with storage redundancy and zero single points of failure.
  4. Operational Simplicity & Vendor Support: A certified, production-grade stack that allows maintenance and updates without complex workarounds.
In VMware with a traditional SAN, this setup was a "set-and-forget" industry standard for 15+ years.

Ceph is definitely a mature and viable alternative today (even NAS vendors like QNAP integrate it now), and it solves the shared storage problem natively in Proxmox.

My original question was simply: Aside from Ceph and Blockbridge, are there other enterprise-grade, shared storage solutions (like traditional SAN integrations) that deliver this exact same seamless experience in Proxmox today?

Feels like this is a moment 22 discussion. As for nr4: Even if you get an awnser here. No vendor vill support and maintain something you "got the awnser for" in a forum. Vendors wan't you to come with the preq. list, then they give you an offer that makes you fall of your chair.

I have yet heard of a vendor who says "Aha so you printed the awnser from some guys @ a enthusiast forum, well set that up and well happy".

The hole point of an vendor is no "setup-and-forget". The point is "setup-and-forget-someone-else-can-help-you-but-us"

The TrueNAS plugin JTY gave seems like a perfect player if it workes.
https://github.com/truenas/truenas-proxmox-plugin -
 
Last edited: