External embedded content is generally blocked here due to security reasons. Unfortunately, Discourse tries to show such content even in such cases (and it fails, because the Content-Security-Policy header is set to disallow it).
There’s a lot of similar devices on the market now that are way under $1k.
I’m sorry to say so, but I think we’re seeing issues with both software and hardware in the Turris project.
3 Likes
The situation is very different from ~nine years ago when the Omnia was launched: Back in the days the Omnia’s hardware filled a void between embedded routers and servers. This is different now with plenty of powerful and expendable devices available such as the BPI-R4 family and OpenWrt’s own One.
At the same time, CZ.NIC’s personnel resources seem to be much more restricted compared to then. From the outside, it appears as if it’s mostly one individual (Michal) who maintains both the ever-growing zoo of supported legacy hardware, Turris OS, the various software projects plus the development of new hardware. I’m not surprised to see projects like the SG being postponed one year after the other.
I’m not here to tell CZ.NIC what (not) to focus on, any of their work is much appreciated! From my user perspective, what I care most about are the fundamental advancements Turris OS brought to how OpenWrt can be used, i.e., BTRFS support, snapshots, package management and OS updates, Python/Go support. This is where I see most value in the Turris project and I hope that the team will be able to (somewhat) keep up with upstream to provide a very versatile OS for enthusiast hardware! Thanks for all the work you put in!
10 Likes
Can you please elaborate?
I don’t share these concerns. Even though the Omnia now isn’t the latest and greatest anymore, and as of recently there may be one or two alternatives on the market, it still runs steady and stable and receives frequent updates after more than 5 years of operation. I think that’s awesome. I manage three networks with Omnias, two of them remotely, and it’s been an absolute breeze.
I think there still is a gap for more enterprise-y yet highly customizable/hackable setups, so I’m very much looking forward to upgrading to the Turris Enterprise. I think that is exactly what we need in terms of driving forward Open Source networking too.
5 Likes
Totaly agree. nOOb here, and this Omnia is by far the best router i had (n=1) for almost 8 years. So the ROI is a big +.
Even though i use a very simple config, no containers and such, just stable openvpn & adblock.
i do like the whole FW /honeypots part. Since that is far to complicated to manage for me, and if i have issues, the forum is almost always very helpful to solve it.
Speedwise a 2.5 GB upgrade on the LAN/WAN part would be welcome, so i would be in the market for the next omnia. The corporate model is a complete overkill for me, but that one could be very good for the team, sales and support wise,
The only remark i would have is to have a normal WAN port, not just a fiber one? unless CZ comes with a fiber>RJ adapter, to avoid comp issues.
1 Like
The fiber port will not actually be a fiber port per se, it’ll be a SFP+ port, so even if none of the built-in RJ45 ports could be used for WAN one could just plug in a SFP RJ45 module into the SFP+ port.
2 Likes
ah ok, and any SFP RJ45 module will do? Or is that rather CPU/hardware/OS dependant?
Any compatible module will do 
In theory all SFP modules should be compatible with all SFP cages, in reality not so much…
That said, I like the current omnia’s design offering a RJ45 etbernet jack for the WAN side or an SFP cage…
But I know quite a lot of users are quite excited about the modularity that SFP promises, some are in for a disappointment I fear.
2 Likes
If it’s advertised as having Linux support, then yes I’d expect you could get it to work on a Turris too.
There’s also a (non-exhaustive) list of tested devices in the wiki: Supported SFP modules [Turris wiki]
Look, the way I see it is this. Omnia is still a strong and good enough platform. When it first came out it promised security and solid hardware. As an owner of three Omnias I can confirm that being still valid. For example I did a Wi-Fi 6 upgrade on one of them and it’s even better than before.
However the market in 2024 is different than what it was before. Reasonable hardware upgrade to Omnia would be at least 2.5gbit ethernet, 64bit cpu and Wi-Fi 7. That would need to launch right now to be competitive on the market at say ~500 EUR price point. There are devices with these features on the market that are under said price point.
Now the software part. Here is where I think the Turris team doesn’t seem to be planning to deliver any innovation. Except for maybe another rewrite of the GUI, which has little innovation value and doesn’t require any specialized skills and it is very specific to Turris and can’t be taken upstream or used anywhere else.
I think solid hardware platform is the basic requirement for a successful product. I think there is lots of opportunity and potential for added value in the software. Example - the AI boom makes coding way faster and easier. What about the Turris providing some AI features? AI-based privacy features like ad blocking or improved AI-based PaKon combined with a full-VPN for family mobile devices?
I watched the video where Michal talks about the enterprise version issues. While I appreciate the open approach, it didn’t spark confidence for me. I was very disappointed with Mox and I had to trash it. I think Turris is aiming to position itself also as an Enterprise device to be able to enter more expensive tier of devices and maybe get more margin. It sort of makes sense, however what happens when you compare Turris to MikroTik (cheap devices with absolutely terrible software!!!) or Cisco?
2 Likes
Hi,
thx @peci1 for sharing.
- While listening to the lecture, I got a bit worried about cut backs on the NG details. Now there are much more capable routers avaliable - where do you want to position the NG?
- @kixorz - I do concure in a way that AI should be integrated, but I think this should be applied to the sentinel network. AI does probability calculations quite well, and that might give an edge to block or identify bad IPs.
- Maybe somebody might give me a hint, why the hardware guys had time to get more on the NG model when on the other hand CPU RAM signal transfer was causing troubles. But I am greatful that they did it, so I get a NG next year.
Thx,
Vienna
1 Like