Micro SD! Is that really necessary? Feels like an extra that has little to no value especially since they are slow af lol. Nice touch with the M.2 connector. Would be baller to swap Wi-Fi Modules each time there is something new. Is there going to be any optional SaaS or Are you mainly keeping this Open? Also, on the Nano-ITX, is this something that you are designing from scratch or using some pre-existing template (hardware) that you are just bringing it up to date with newer components? Good job. Excited for this tool.
Fun fact: MicroSD makes it easier to develop as we can leave out NAND completely (in the prototyping phase) and have the operating system on the microSD card, which makes it much easier to swap to an Ubuntu machine for changes to the filesystem, for example. And yes, M.2 was selected for this very purpose, I want the router to be usable for at least 7-10 years, and with a new wifi standard coming out what seems like every week, this is the best way. We're designing everything from scratch!
Hate that 2.5gbe is missing, please include that since the cpu supports it. I know this is a router, not a switch, so only internet traffic is going over it, but if you can get a high-speed connection out of it without SFP modules you must, fibre to the home 1.5gbps is already here, upto 3gbps even I think with Bell in the area
@@brainsoftit being a router doesn’t mean that only internet is going over it. If you have separate vlans that traffic would need to go through the router. Having 2.5 is something I think would be fantastic.
Maybe something to look at since you are looking at PoE - is pick a port to allow PoE-IN So what that would allow someone to do is have a switch power it - 1 less power adapter ... or that and pwr adapter (in case the psu dies - happens - poe backup)
Please absolutely a 2.5gb with so many new wifi 7 aps, it makes sense. Dont want to use a converter for sfp for the AP. I think POE on just the 1gb makes sense if possible.
Thanks, I'm doing everything possible to get at least one 2.5 Gb port on the device, stay tuned as I'll hopefully get in touch with the people at NXP to check whether it's actually possible or not. It's bugging me as hell, too!
I am super hyped to see this in a solid form, hope it all works out. Can't wait for September. Also as a side note, I really enjoy your structure for you videos. They flow well, and the recap at the end is a nice helpful touch.
As most of the people already said it, 2.5 GB port is a must. But what really piqued my attention is mentioning of OPNsense, that would instantly make me a customer. Puno sreće u pothvatu, Tomaž, pozdrav iz Hrvatske. ;)
Very well thought out Tomaz. Points of consideration... Coreboot and ability to integrate with PFsense/ OpenSense as a means of "succession" upgrade ability. You are very much on the right path and I for one wish you every success.
Tomaz, I only stumbled on your channel a month or two ago, and I'm glad I found it as you were starting to present your plans for this product, as I've found watching the evolution of your plans very interesting. In reading the earlier viewer comments on this video, the "it needs to have added to sell" ones would be nice products to have, but it seems to me that they minimize just how difficult it is to start a new company , and that people are far more likely to give a new company a chance with lower-price products and you're more likely to get those products to market more quickly. The one thread that seems to come through well, and I agree with, is the need for 2.5 rather than 1.0 ports, given how fast the use of 2.5 has grown: I don't care that much personally about 2.5, but every motherboard I've purchased (mid-range models) in the last 3-4 years has at least one 2.5 port, and I suspect that the people who will be willing to give your company a shot are the folks who will value having 2.5 ports. Good luck with this.
Thank you! And yes, 2.5 seems to be the sweet spot currently. I was even thinking to dedicate a PCIe port to 2.5 GB ethernet, but then we have a problem that those would bypass the DPAA architecture which is one of the main benefits of the CPU. Compromises are hard 😅
Imho I would ditch all 1gbps ports all together and add 2.5gbps instead. They are backwards compatible anyway. For CPU, I like arm64 on some parts but for router, I don't know, i'd rather have AMD64 cause that means I can switch O/S whenever I want but with arm64 I would be pretty much locked to a single or two max options. Also one more thing, it has to be rack mountable or at least make the holes for it and we can 3D print something to mount it in a rack
Most Linux distros have an arm64 version, so you should have quite a few options especially if they get UEFI working, and I really like the CPU he chose for those networking optimisations he stated in the video.
@@SteveTech I am talking about pfsense for example that doesn't have arm64 at least not one that i can see in their downloads page. I am not talking about generic linux distros, this is not a pc but a router so please keep the convo in router o/s and not deflect on the topic
@@SteveTech en.wikipedia.org/wiki/List_of_router_and_firewall_distributions this is a list of router and firewall distros, there are a few that indeed support arm64 but I wouldn't use any of them in a device like that. Someone said about pfsense, it is listed as arm as well but I guess that is for netgate's devices. So indeed going with arm64 is limiting the options
@@toumpanis Yeah sorry, I'm happy to set things up the hard way through iptables, dhcpd, unbound, and whatever else, but I realise that is probably way out of the target demographic. Also he did say he hopes to port opnsense in the long run too.
@@SteveTech I don't see any point fighting windmills like Don Quixote when there is no need for it. As impressive as it might be to setup the hard way, I wouldn't go that path for various reasons, convenience is one of them. It is good for a project like this to be versatile so anyone can work it the way he wants :)
"it will fit into a rack" was the final checkbox. At last, the perfect router hardware I've been looking for. This series is really interesting in a lot of ways. Thanks for sharing.
Very cute, looks like it will be a little Brother to my Ten64 by Traverse here in Australia :D, Mate, i hope this takes off and its so enjoyable to see your enthusiasm and excitement about it all.. More open source, more consumer friendly, and above all, much love for the product you want to birth ! :D Keep it up and I look forward to seeing this come to fruition, I kinda wanna buy one myself regardless cause I can always find a purpose for it when it runs OpenWRT :D hehe (akin to the 'tiny mini micro' that serve the home do) :D KEEP IT UP!!!!!
We may fail, but not for the lack of trying. My train of thought goes like this: pfsense works on ARM, so opnsense, being a fork, should too. I guess it comes down to available resources and persistence 💪🤓
making sure usb cellular failover would be a must for me as well as 2.5g network port and possibly a redundant power solution along with locking power connectors
Just discovered your channel and this project... VERY EXCITING! A key point for me is Wifi (802.11be) but you mention that this could be addressed via an M.2 slot. Making it a standard connector with easy upgrade path sounds like a great idea vs onboarding it! Looking forward to seeing the progress, I'm tired of proprietary Wifi 5 / 6 routers that I can't flash OpenWRT on to. The trick of course is finding a Wifi 7 adapter that supports dual band AP mode
Your videos are great! You give a great overview on all topics around the router and just enough details to get us intrigued so we can search in depth on our own. As you mentioned the following video will have more detail but you are on the right spot of knowledge and entertainment! Great work! Can I pre order the router? 😂😂
Thanks for the kind words! I'm kinda torn on preorders, I wouldn't want to take people's money without knowing I can deliver both the highest quality and on time. Maybe once we're much closer to launch and I feel more confident. 💪
PoE is easy to add as an extra plug-in board, just need a header with the ethernet transformer centretaps and power, so having that as a user-installable option seems like the most sensible approach as not everyone will want it
Excited to see this come to together! My only concern is having 2.5GBe port. Internet speeds are only going to get faster (in general), future proofing to support these speeds would be great :)
Thanks for providing some much information about the process btw, it's been very interesting and I've learned a lot about the business side of making a product 💯
Im not sure 2.5g specially if it is rj45 is all that relavant for wan connector i think having sfp+(10g rated one) would serve much better in the long run
This is very exciting. I'm eager to see how this goes. We are switching from SonicWall to pfSense, but this might be even better. I told my I.T. dept about this project.
I feel the 1gbe ports are a waste of time at this point. Definitely need at least 1 2.5gbe port. Better yet, two SFP+ 10gbe, two 2.5gbe multigig, no point in 1gbe unless it's poe input.
Unfortunately, the CPU doesn't support two SFP+ AND two 2.5 gbe at the same time. I'm working to get in touch with an NXP engineer to check whether even one 2.5 gbe is possible - the datasheet is somewhat ambiguous in this regard.
Not really, most hardware that you connect to the network does not need high speeds. Think printers, security cameras, smart fridges, etc. And even in PC world we don't really need 2.5Gb/s when the ISPs don't support 1Gb/s. I think a single 2.5G port and two 1G ports would be a perfect mix.
Eh, I would say it's useful for normal devices that might be situated near/around the router (like a Home Assistant box, UniFi Cloud Key, etc) and plugged in as a convenience, then everything else would be connected to a dedicated switch over 10Gbe (which is probably better than the software routing on LAN ports on the router)
We should remember this is router not a switch which limits what will be going trough it so a 10g port is essentially to switch which is already overkill but you may have other ports to be do multinetwork routing so 10g for main line isnt that overkill but 1g i think is enought for next few years cause ik honestly not sure 2.5g will actually catch on specially on ISP side
@@hubertnnnas i said 1g in and 10g lan are probably enough even tho in perfect world you would have 2 10gs which would basically make from connection perspective this router future proof for next many years
I like the Layerscape processors capabilities. I think it is a great choice here. I think you mentioned in one of the early videos that this was targeting gigabit throughput with inspection capabilities. Is that still the case?
I get major Deus Ex: Mankind Divided vibes here, which is set in Prague and although not where you live, I get these strong "We're breaking out of what is on the market and making our own, good, stuff" vibe. Which is great - I'm looking for a router and if this project comes along well and great, I know what I'm going to buy. Besides: It is actually really watching along someone doing what they love and learning something from it, too. So thanks for doing it, I'll keep watching and hope for a very very good product. And one more thing: If at least one of the 3 RJ45 Ports can be Gigabit, that'd be great but what about making one of them POE+ for powering the device? Maybe not in the early stage but it' definitely be an awesome feature for skipping the powerplug/adapter. Anyway, Thank your for the videos and THANK YOU for this project, Alex' :)
Software features are often overlooked unfortunately. The internet would be in a better way if the following were legislated for routers: 1. First time boot has blank creds, thus no vendor defaults to be abused; 2. A "Verify" button that needs to be physically pressed to confirm any changes be active by default (CISA propose this also); 3. Minimum functionality exposed on WAN port by default, everything else must be explicitly setup manually/via wizard; and 4. Auto firmware update must be on by default, and support period disclosed by vendor before purchase. "... everything should default to most secure ..." source/credit: Steve Gibson (en.wikipedia.org/wiki/Steve_Gibson_(computer_programmer))
USB-C with PD is the way to go. You don't need to deal with gigh voltage certifications, have a world wide usable powersupply and one can use a usb-c powerbank with passthrough charging as a solid UPS at minimal setup complexity.
Look at the support for the DPAA of the CPU in the Linux Kernel. The biggest drawback of multi-port routers, which don't include switch chips, is the port2port packet forwarding performance. In these designs all packets need to go through the CPU which hammers the performance/increases power utilization. Also agree with 2.5GBe port - it's a must-have these days. I know you could offset this by a SFP+ -> RJ45 converter, but still a native port is better. Having PoE capability on copper ports will enable easy deployment with numerous WiFi APs. And do you plan to enable M.2 storage? Or just keyed for the WiFi adapter?
I have a form up in which I ask all the relevant questions and the majority of people "vote" for two SFP+ instead of RJ-45, reason being max flexibility. And as for the M.2 storage, I don't see why not, it's a firmware/software thing, once it's routed to the CPU properly.
Agree with the sfp+! What I meant is that if we don’t get a dedicated 2.5G copper port we’ll need to sacrifice one of sfp+ to it. If it comes to M.2 - WiFi and NVMe have different keys - so it’s either or (unless you use adapters)
M.2 storage and wifi tend to use the same interface (PCIe). Do be clear on what's supported, e.g. USB or SIM connections. As with type C, that's not evident from just the connector, though there is some keying.
@@0LoneTech They use the same PCIe Bus (yet that's not entirely true as well), but the physical slots are not compatible. M.2 WiFi card are key A+E, while NVMe M.2 drives are key M - they are physically incompatible due to the notch/key position. Key M slot has 4 lanes of PCIe, while A+E has 2 and additional signals to kill WiFi signal.
Even low-end consumer motherboards are now 2.5gig. We all keep 1g for "older hardware" and bmi, 2.5g is quickly becoming, if not already, the new 1g eth port. Perfect product, but make them all 2.5g ports that can be used at 2.5 or 1g , even for households as it future-proofs it a little more.
I'd say having the 2.5G port is a must have. Both of my local internet providers are set to introduce 2gib+ within the year, and are already upgrading existing CPE to have 2.5g. While using 1 of the sfp+ ports as an uplink from a poe+ 2.5g switch to provide wifi7 APs is acceptable, having to use one just to get a full speed link to your CPE would leave a bad taste in my mouth. IMO
Thanks for the feedback, I'm trying my best to get the 2.5 on the device. You see, the problem is that there has to be a compromise. A more powerful CPU would make the overall device more expensive (in a sigificant way). Or, if using the same one, we'd need to let go of one 10GbE to get more 2.5GbE ports. In every case, something has got to give.
@@tomazzaman Sure, that's understandable. I'm really only advocating to breakout the single 2.5g that the block diagram indicated was available from the CPU. However if losing an sfp+ port is needed to implement it, then may I suggest a mount w/blanking plate on the casing near the M.2 to support an M.2 A+E KEY 2.5G
@@invictus0x0yep, that too has crossed my mind. Or, adding an additional PCIe PHY chip on the board. This traffic wouldn't go through the "optimized" path (the DPAA), but 4 cores shouldn't struggle with 2.5Gb.
Super looking forward to it. I'd like the 2.5Gbe port just to be able to eek out that little extra but I feel like one SFP in and out should do the trick most of the time, but if it were possible with the current processor, full 1/2.5/5/10 Gbe RJ45 would have been great just to make it suit more needs and extend a bit further into the future, though I'm someone who bet on the jump straight from 1Gbe to 10Gbe rather than the market all going to 2.5Gbe, so I'm set for it. I'm not expecting it soon but this is definitely enough that I'm putting off upgrading my pfsense box to see how this pans out just because it checks the boxes so nicely. Best of luck and looking forward to the future updates!
The market is too saturated. This will cost more or the same as the other devices on the market and offers no benefits than what is already available: Firewalla GLINet N100/N300/Older Celeron Boxes - Running OPNSense/PFSense. R86S Devices The list goes on. Full DIY: Old PC. Then you have other players with more robust features, ie: TP LINK and Ubiquiti. I don't see how this can compete.
A lot of what you mentioned are custom devices that require a lot of tinkering. Some people just want to buy a product and be done with it. Plus he already has an audience (a growing one due to these videos) so I can see it doing well because of that. There's not doubt it will be hard to compete with the likes of Ubiquiti but he's going for the home/enthusiast market not enterprise as far as I can see.
@@phizaal the UniFi line is for home as well. I run it myself. There’s no tinkering needed for the other generic devices either. You install pfsense or opnsense and everything just works, I have a bunch of these boxes running at different locations. They can route full 10gbit with IPS on. They also have sfp+ cages and a bunch of other features too. I dunno, we shall see.
Good luck on your quest in making good looking and capable router. I have few queries, why not consider Banana Pi BPi R4, open source hardware which matches 60% of your spec, can be customised further and assemble it with your outer design enclosure. would you consider partnering with GLi net under white label service to make the router board cheap and assemble with customised outer casing.
Banana PI I can just buy, where's the fun in that? 😂 In all seriousness I believe there's a place in the market for a router that isn't just a router, but rather a community of enthusiasts that builds stuff together, shares their findings and hopefully inspires the next generation of makers and thinkerers. It's not the hardware. It's the people. And yes, partnerships are of course always at least considered. Or they will be, once we're far enough.
This sounds like a great product, but I think it will be really important to know the market and who actually want such router. Since the product is not for the regular user and not for the really advanced users. Average Joe will use a generic router and the advanced users will use more powerful hardware with pfSense or openSense. From my point of view the product will fit for the advanced home users and for simple homelabers etc. For me personally the most important features will be rack mounting, upgradability and more powerful hardware so I can run demanding software like Suricata on openSense or pfSense.
Take a look at VYOS as an alternative OS. Yes, it's not everyone's cup of tea (no GUI) but it's definitely a more rounded router operating system for the professionals.
This is a very good start. I'd really like to see it grow as a lot of users used a bridged setup, but then with a hardware firewall. afaik here is no a Pfsense or an Opnsense ARM firewall, so if you can design an similar (Linux) based (so not BSD) Bob's your uncle. (ofcourse it would mean "need more memory")
offering a version with just 2.5 GbE would be a good fit for most users (not everyone is a home-laber); fiber internet isn't widespread yet (at least in USA)
Hi Tomaž! How do you find the motivation to keep working on an open-source project? I was designing a CM4-based ADS-B receiver, but lost motivation, and abandoned the project.
By having this RU-vid channel. Positive thoughts motivate me, negative ones keep me in check and me making videos keeps me accountable. Win win win. :)
What’s with an discreet power port like a 12V barrel jack? The USB-C option is great but when it comes to integrating it in existing stuff a barrel jack or something similar is easier because you don’t need to deal with USB PD communication and just can add a 12V source.
I'd argue is the other way around. USB Type-C is pretty much everywhere these days. Everyone has these cables lying in a drawer somewhere. And USB PD is also standardized, so there's no issues with compatibility. The device will come with a GaN charger and a cable anyway, so in reality, it makes no difference - in fact, PD wasn't even taken into consideration initially, but enough people "voted" for it on the form (2000 responses to this day) that we decided to add it.
@@tomazzaman Yeah USB PD is great and an Allrounder, and pretty much works with everything. But would it be a lot of work and higher cost to add a DC barrel jack?
Excellent initiative. There is definitely a major gap in the market for a device like that. If it had both a 2.5Gb port and built-in Wi-Fi 7 I'd buy it in a heartbeat. Best of luck to you and your team.
This is so awesome! How can we pre-order or pledge? 2 small inquiries: SFP+ compatibility is a pain and each ISP uses different modules. Is it possible to maximize compatibility somehow to avoid needing a media converter? Similarly, is it possible to make the SFP+ support all multi-gigs: 10, 5, 2.5 (and 1)? Thanks for sharing. Really looking forward to more updates! Edit: I also believe Type-A would be better as the host for a cellular modem or printer. What sort of Type-C devices would fit there?
I'm not a huge fan of preorders, as many have been burned by them in the past and I'd hate to be in this position. Maybe, just maybe, when we're close to launch and if the demand is significantly higher than the supply. As for the multi-gig support, that's up to the SFP+ module that you use, we plan to test as many as possible to make sure the widest selection is possible.
Thank you for your attention to detail in the development and inclusion of crowd feedback while providing good analogies. If I were wealthy, I'd invest, and i don't even have a computer at the moment! I'm curious, or maybe just suggesting a compatability ( running a bitcoin full node or other cryptocurrency of which the samurai dojo was what I was just watching.) And tax laws in Deleware are good. California has better weather ( that's where I am).
All ports UTP should be 10Gbe since that would be a minimum for power users in some years ahead. NAS loaded with SSD"s/M2's will be the norm, and you want at least 800-1000Mb/s throughput to use those NAS boxes for video editing in 4K. Even 8K is around the corner. And for Mac, since local storage is extreme expensive, this becomes a very good use case!
I agree. But there's this small issue of,... price :) NXP actually does produce a CPU that supports 8x 10GbE ports, but that CPU itself costs around $300, and if we add PHY chips, connectors and all the other necessary stuff for 10GbE networking, it would cost well over $1000.
I want a small display even if it's just a small cheap one. I'd like to see a spi based display with known libraries like a st7735 or an st7789 adafruit has the small 240x240 display that's almost identical to the one on unifi units. Or add an eyespi header so we can mod. If you do use poe why not just make it poe pass through rather than add internal poe. Setup the 1gb ports to take in poe+ power the device and pass on single pair poe.
Did you consider to build an architecture of a "hub" based on Raspberry Pi Compute Module? This would support a modulare architecture that allows consumers to size CPU and RAM based on their preferences and needs.
Just 1 M.2? Full spectrum Wifi 7 AP modules take 2 M.2's. Cellular as well. Maybe a PCIe multiplexer with support for modular applications? It' sounds like a BPI with a restricted feature set otherwise. SFP+ is a standard from 2006.. That CPU will be old news also by the time September comes around. I may have to wait for a Gen 2 :).
We're making a router. Even a single M.2 is an overkill for most, haha. And fun fact, ARM Cortex-A72 is now 8 years old, still being mass produced. Just shows that old doesn't mean bad. Like myself. 😂
With you wanting it to run on OpenWRT, is there room for orchestration software? Or at least a good gRPC implementation? This could really take become something of a self-hosted Meraki-style solution... I'm no dev but open to support it
He said m.2 wifi in the video and m.2 wifi and m.2 cellular have different keying, also m.2 cellular requires a sim slot on the motherboard wired to the slot
Hey Tomaž, do you have any rough estimations on how it will perform as a router (in term of speeds)? Or a speed comparison with other most common used CPUs? Did you check BananaPi BPI-R4 which is quite similar to your board and also supported by OpenWRT? Pozdrav iz Slovenije :)
I really like the idea of this project but I think having PD on 2 ports instead of 1 would be great for having a possibility of redundant power sources or even using a powerbank as a ups for the router. I would also prefer 2.5gbps ports over 1gpbs
Now that i think about it more from router perspective a single 1GbE RJ-45 jack is a must as it can be used as a management port or as a networking port for those of us still slumming at rj-45 based 1GbE lan (not that i need anything more). but i also think 2 SFP+ ports are a must as well. SFP+ is very versatile connector that is native to fiber which is what most faster then gigabit WAN(and even slower then that) connections would already be either sfp SFP+ or a fiber connector that you connect into SFP+ module therefore having RJ45 2.5 GbE port would be kinda bad and wastefull specially if SFP+ port was eliminated also using 2.5GbE RJ-45 makes sense from UX perspective, but doesnt from any kind of expandability or usability perspective as sfp+ can be adapted to do rj45 so if user had a rj45 2.5GbE network they could use a media converter module to go from sfp+ to RJ45 at which point they can connect to that. and still allows those with 10 Gbps network to use SFP+
Specialized hardware will always be better for specialized devices. Gotta love that efficiency. Just don't use FPGAs, that would (to my knowledge) bloat the cost like crazy. And I'd love to see your studio! Sounds nice :)
I Think Its a cool project but i have a few concerns. I ve recently upgraded my router to 25G using a minisforum ms-01 and a mellanox 2x sfp28 nic. It runs opnsense and has 2*SFP28,2*SFP+ and 2*2.5GBase-T and cost me around 800€ which is an incredible price for a device this fast and capable of pushing 25G... I sadly dont see where the Router your Team is designing would make sense for somebody like me... Cool project nonetheless
Thank you for the feedback. I like how some of you are pushing the limits, but what I need to keep in mind is the majority. And the majority doesn't use or need 25G.
@@tomazzaman and the Router definitely has its place... I used to work for an ISP and we had ekinops OneAccess devices which were arm based devices just like this one using some BSD/Unix firmware and they destroyed every other manufacturer in terms of flexibility and performance. If done right yours is gonna be a beast with even lower latency than my Software based machine (using a Opnsense Software bridge for the SFP+). Also i like the machined Aluminium housing alot! I dont remember wether you plan on putting a fan in it, but if you do, i definitely wouldnt cheap out on that. My previous Asus AX89X had a Fan which was ridiculously annoying... Dealbreaker for me by now. And for the WiFi i have a wird idea: you could do a tray like a disk bay on a Server Option one: a tray to put something like an Intel be220 in with SMA antenna plugs on the Outside of the case Option two: a tray with an oculink Interface to connect on the outside so one could build an external WiFi AP module with high gain antennas or whatever using the pcie over oculink to connect to the rozter... Similar to a framework Laptop modules
@@legendaryz_ch Thanks! We do plan to put a fan header on the board, but aren't sure yet whether it'll be necessary, at least not for the alu version. Regarding the WIFI, there will be an M.2 connector on the PCB (Key E), which should give us the flexibility to put one on after the fact. What we're also exploring is out own access point, but nothing conrete yet, we'd rather focus our limited resources on the router to develop something great.
@@tomazzaman that might be true, but the majority also does not want a full metal enclosure which costs more than wlan routers with NAS and phone and iot and cable modem and ap. So if going for the high price, also go for the high-end
I have experience with these NXP Layerscape CPU's and UEFI is quite hit and miss there is some work that needs to be done to get a "full" bios. (Currently use a LX2160A has DPAA2)
Came in wanting to defend the N100 (the N200 has an embedded option btw) but I see no issue with this. Just make sure to get somebody who knows NXP and Layerscape on your side. Also, on openWRT NXP says you can do 20 Gbps with 5% CPU load, but you bought 100% of the CPU, what do you wish to do with the rest of it?
As a gamer I want a low latency router first and foremost. I would like to have easy installation for pihole(easily set up local dns) and some form of smart queue management. Automatic updates would be really important and an option to set your timezone and time format to be able to make it update at a set time. I do not care about POE unless you can get it to work with Ubiquity products.
I build custom networks for very strange places, usually with extreme weather. Often using the $250 Mikrotik RB5009UPr+S+in and loving them. Not quite the same specs but I'm interested in more info on the durability specs.
How will your router compare to a high-end server (Sapphire Rapids) running pfSense? I'm guessing the server would easily beat it, but maybe the specialized processor of your router instead of a general purpose one would give it an edge?
Good project, wish you luck. I've got a technical question: there are NOR flashes up to 256MBytes (not cheap though), but to my experience, it's more than enough for a router, at least for OpenWRT. Why to add NAND, especially if you plan to have USB and M.2?
Because in the long run, I'd like to port OPNsense to this device, and based on their documentation, 40GB is recommended. And NAND is there because I want the device to be plug and play, I don't want customers to have to deal with M.2 and/or USB unless they want to.
All the newer pc motherboard (AM5, LGA1700) that are not 10$ have a 2.5gbe port, so I think it could be really useful for the router! Also im wondering if i can somehow get DD-WRT working cuz im too trash for openwrt
So basically you are building something like turis. It would be probably best to build it so it is affordable but also can do wifi and has easy to use interface for non technical people with an option to see more advanced. Building a device that maily meets requirements of enthusiasts is a mistake I think. Try to make one that is more down to earth and one that is for enthusiasts. Will this router be available in the entire EU? Do you also plan to get in contact with local sellers or will you mainly handle everything yourself ? Btw i just discovered you and it seems like interesting project. Good luck.
I'm curious what kind of bandwidth this CPU will be able to do with SQM (CAKE) enabled on a 10-gigabit link. If you use an SFP+ module to link to a cable modem with a 2.5GbE port, you'll definitely want SQM so it won't try to send 10 gigabits through the 2.5-gigabit link.
Unless this is some sort of enterprise firewall, SFP+ really isn’t needed. Especially if the majority of users will be connecting this to Cat Ethernet. Copper Ethernet SFP+ modules are known to get hot. That smells a lot like a future headache you probably don’t want to deal with.
A N305 with 4 x 2.5Gb ports is enough for me. Running FreeBSD 14 native with opnsense in a bhyve vm. This means I can get a lot of use out of it with multiple VMs.
Ideally, this router will not be the only device in a home network, I personally want it just as a router, and then then have it connected to a switch, which then can break out the 10Gb SFP+ into multiple 2.5Gb ports, for example.
Will there be the option to add some M.2 SSD into it? IMO it would make it a perfect NAS without the need for external one. Just my 5 €c. AFAIK WiFi and SSD have a different keying on connector. Maybe adapter could be used for different keys.
So, the current plan is to use Key E for the M.2 socket, which isn't compatible with NVMe drives. If you want additional storage, there will be a USB3 port. But, that's not the focus, we're making a router, not a NAS :)
I'd like to hear from you why this router is better than alternatives from your competitors at the same price point. I think it's wise to consider it while making your specifications, because if you're going to be roughly the same specs at the same price, I think your competitors will have all the advantage.
@@tomazzaman I did watch that video, but it is rather vague in terms of specifications when talking about your competitors. But now you have your specifications, your price point, there's still competition to look at. For example, Mikrotik's CCR2004-16G, which also has 2x 10G SFP for about the same price. Having your hardware open and customisable OS is a big selling point, but you might need a little extra to push fence-sitters to your side, even if it may be just advertising the ways the hardware can be customised (upgrade RAM? / Add Wifi? / Expandable Storage?), or on the software side like being capable of running a small mail server for your family on the router itself. You made a video about COGS vs Perceived Value, and mentioned in it that you can increase the perceived value simply by flaunting your product's capabilities.
is the POE+ going to be a power sink or source? Regardless there are POE modules that can be added if needed , provided that you added pinheaders on the needed signals. OR u can simply pin pinheaders and develop the POE module later on another 2-4L PCB if there is enough demand ( if there is not enough space for a standard module or you simply want something built better/ that fits better overall )
