Optimizing Virtualization Costs with Smarter Storage? (Insights from the TrueNAS Team)

Announcements
1

Hey all!

One topic that’s been on our minds at TrueNAS lately is how the underlying storage infrastructure can really impact the overall cost of virtualization, especially given some of the recent shifts in the licensing landscape.

We recently published a blog post that dives into this, exploring how TrueNAS, particularly with the latest 25.04 release (“Fangtooth”), offers features like high-performance NVMe and intelligent deduplication to help consolidate your virtualization hosts and potentially reduce licensing overhead.

Here are a few key points from the article that you might find interesting:

  • Achieving sub-millisecond latency with TLC NVMe drives for improved performance.
  • Implementing deduplication that’s specifically designed for virtualized environments.
  • Utilizing Fast Copy for faster and more efficient VM provisioning.
  • Leveraging the benefits of an open, ZFS-based architecture, avoiding vendor lock-in.

The blog post also shares a real-world example of a customer who’s projecting substantial long-term savings by adopting this approach with TrueNAS. You can check out the full article here:

Slash Your Virtualization Costs with TrueNAS

I’m really interested to hear if anyone in the community has already experimented with similar approaches using TrueNAS, or if you’re thinking about how storage efficiency could play a role in your virtualization setup. Always keen to learn from everyone’s experiences!

image
image1200×1200 100 KB

2 Likes
2

Welcome @Ruth_Jose to the forums.

Preface: I am not an IT person. I have been building, troubleshooting, and programming computers since 1977, the most recent system being built is the weapons system going onboard the next SSBN Class submarine in a few years. It is a complex computer which takes up a considerable amount of space, so not your typical corporate computer, and not a quantum computer yet, although that would be very cool. Please temper my comments knowing this about me.

I read the article and I personally have not given LXC a solid try, I was waiting for the actual release version to come out as I found bugs in RC.1 so I thought to myself “Self, I can wait another month to play with LXC”.

The statement Users with production VMs on TrueNAS 24.10 should not upgrade to TrueNAS 25.04 until after this experimental feature stabilizes in a future TrueNAS release. This statement is apt to put off some of the experimentation from a corporate perspective. I come from a government background so I tend to think this way, and the Govt IT folks would never be allowed to play around until a solid product emerged and has proven reliability and passes all the security (lots of that stuff) requirements. Of course the Govt can and does buy volume licensing for ESXi, but with Elon slashing the budget, that is a possible cost savings. Not $200 Million or Billion but still, if TrueNAS could get there, that would be a rather good savings. With that said, I would expect to purchase a support contract from iXsystems so that needs to be factored in (I have no idea what that cost is or will be in several years).

As for the ultra fast NVMe response times, I suspect you would get that as well with TrueNAS on top of ESXi for example (how I personally run my tiny home lab). Yes, I have the all-mighty free license. If I had to pay, I’m am 100% sure that my tune would change and I’d be using something else. I have not run TrueNAS on bare metal in a very long time.

As for Deduplication, most of the home users would never be advised to use deduplication in the first place, however I realize that your posting is really aimed for the small to large business user, not the typical home user.

I guess my question back to you would be something like this:
When comparing TrueNAS VM to ESXi VM, How easy is it to deploy and configure VMs in TrueNAS vs. ESXi?
This will of course be a key part of a users concern when choosing a Level 1 Hypervisor.

With all of that said, I think your Announcement is a bit early. I don’t even see this being an announcement level posting. Maybe I’m being too critical.

I’m certain a few IT users will chime in and provide more valid comments than mine.

Cheers

3

I think you misread the intent of the blog. In the Enterprise It domain, we recognize that ESXI is the gold standard, but it also comes with a gold price. Unless you have inherited a free license or have unlimited budgets, reducing license costs can be necessary.

4

I may have for my summary but I think I covered it from my narrow perspective/experience which is US government related. I tried to provide honest feedback, even if it isn’t flattering. Let’s not forget my disclaimer of not being an IT person.

Maybe I got this blog/announcement completely wrong, sorry if I did. I would expect @Ruth_Jose to apply the appropriate weight to my comments. If I missed the mark then give it a weight of 1. If I’m spot on (probably not) then weight it with a 10. It is all part of the analysis.

I know that @Stux has deployed TrueNAS (according to what I recall about the TTT) as a Hypervisor for his business and it sounds like it is working well for him. Can it be done, of course.

Anyway, not trying to ruffle any feathers.

1 Like
5

Hey @joeschmuck, thank you so much for the warm welcome and for taking the time to read the blog.

I really appreciate your response. I’m not an IT person either, so that makes two of us. I’m here to connect with the community and highlight how folks are using TrueNAS in the real world. I’m also here to learn from all of you, and it’s always great to hear different perspective.

Thanks again for sharing!

6

Yes, TruenAS can be used as a hypervisor with all the next Incus capabilities.

However, we are not recommending it (yet) for Enterprises that need high uptime and automated management. Fo those customers, many are staying with VMware and just want to reduce their annual bills. TrueNAS Enterprise can help with that.

7

I read the article too. With all the best intentions on my part to try to be helpful, it very much comes across as a sales pitch with some numbers thrown in but it fails to explain how and when those would become relevant in real world scenarios.

Yes, ESXi licensing is a big concern for many. But reading the article I don’t understand how TrueNAS proposes to address that. By altogether replacing ESXi as the hypervisor? Not anywhere near ready - virtualisation in TrueNAS is still experimental and breaks with every major release, let alone missing many of the enterprise features in ESXi. By deploying fast storage hardware to reduce latency thus reduce server count? That’s a moot conclusion, and NVMe can anyway be deployed in most storage solutions, not only TrueNAS. Fast dedup? Yes maybe interesting, but from my understanding is the jury is still out on real-world gains from running VMs - in addition, it would slightly negate the previous point on ultra-fast NVMe storage, as it adds latency and overhead.

The article is very light on any sort of detail of how all this ties together into an attractive value proposition for real-world scenarios towards a particular customer group with particular needs. Rather, it looks reverse-engineered from a bunch of technical features.

My thinking would be you need to work more on who you are actually targeting with this and get more specific on what you can actually do for them, beyond headline ZFS feature level.

Large enterprises with large ESXi clusters and storage from Dell or IBM? Cost-sensitive but performance oriented academic or government agencies, e.g. research labs, currently running things like Ceph clusters? Mid size or micro companies who are running just a handful of servers and would like to simplify and consolidate? Or those who are concerned about keeping data safe and encrypted? Or those who are currently mostly in cloud but increasingly concerned about geopolitics and wants to bring the data home?

Once the overall picture of where you’d like to go is clearer, it would also be easier to explain the value proposition.