It's truly inspiring to see how our community of innovators and tinkerers has pushed the boundaries of what's possible with the Brume 2.
From powerful drop-in routers and ad-blocking home servers to intricate IoT hubs and secure network gateways, your creativity has exceeded our expectations.
This brings us to the exciting news: the next Brume is pre-researching.
At GL-iNet, we firmly believe that the best products are built with our users, not just for them. Your real-world experience is invaluable. That's why we want to open the discussion early and listen to your voices on a key hardware decision.
1. The Vote: USB-A vs. USB-C
A central feature of the Brume is its versatile USB port. For the Brume 3, we're evaluating the connector type. Please vote in the poll above and consider commenting on your choice.
The Brume 3 USB Connector Type
A. USB Type-A (The current, rectangular standard. Best for backward compatibility with existing peripherals without needing adapters)
B. USB Type-C (The newer, reversible standard. Offers modern connectivity, potential for faster speeds, and is becoming the new universal norm)
Beyond the USB connector type, we want to understand what you plug in. Your use cases directly influence our design and testing priorities.
Please tell us:
What device(s) do you most commonly connect to your GL-iNet router's USB port?
What is your primary use case or scenario? (e.g., 4G/5G modem for internet failover, external storage for a NAS, a specific dongle for a project, etc.)
Examples from the community:
"I use a USB-A 4G LTE dongle in my camper van to create a always-on mobile WiFi network."
"I have a USB-C (C to A adapter) external SSD connected for network storage"
The Brume's flexibility has unlocked a world of applications we never imagined, and we are endlessly grateful for your shared scenarios and ideas.
Your insights are the blueprint for the Brume 3. We are listening, and we can't wait to read your responses.
What kind of question even is this? Type-A, of course. There's 'thumb drives' that will fit nicely. They're easier, cheaper to replace than solder-on eMMC when the flash finally wears out. This is all the more important now that GL.iNet has decided to drop microSD/TF card support on all current devices.
No USB modem is going to come close to USB 3's 5 Gbps max. through-output, anyway so USB-C 10 Gbps is a pointless premise given the real-world 2.3 Gbps Ethernet limit.
Then get a Slate Plus. There's no radios in a Brume.
How about you populate the UART headers & assign it a port, instead? 115200 bps serial is more important for hardware hackers than some esoteric, unsupported 'dongle.'
Here's the problems with the Burme v3 & it's not even on the drawing board:
Cortex A53, not A76
1GB where 2GB or 4GB is expected if not 8GB while 16GB is on the horizon
2.5 GbE when 5 GbE & 10 GbE is now mainstream while 40 GbE is falling in price every day
GL.iNet's primary wired devices for stationary use are poor comparisons, half-hearted imitations compared to FriendlyElec's year old offerings like their R6S. Only the GUI is 'better' but I personally don't dislike LuCI.
My 2017 launched Certa is more up to date than my Slate AX... yet my Certa is EOL'd. Real innovators can't use GL.iNet's SKUs with their stale kernels or let themselves be shackled to proprietary SDKs, Bruce. Pass that along to your PM(s), please.
root@certa:~# cat /etc/glversion && cat /etc/os-release | grep 'PRETTY_NAME' && uname -a
4.3.25
PRETTY_NAME="OpenWrt 22.03.4"
Linux certa 5.10.176 #0 Sun Apr 9 12:27:46 2023 mips GNU/Linux
If the brume 3 marketing still follows the security gateway road.
Then I think the hardware need to have much more processing power, the memory needs a increase the cpu is ok, but could be higher, on the other hand maybe arm is not the right cpu, but x86.
I'm aware IDS was on the feature notes, but then you also need to be more uniform with the hardware, snort for example asks alot of ram.
I think alot of consumers would expect this when they hear security gateway.
On the other hand, if that is not feasible, why not implement it as a extra matter controller for iot?
Then the marketing makes more sense, than just a few buzz words like iot, security to sound better, just my opinion
The thing with this is, it is a little misleading these extra words can make a consumer assume there is something more than just a router, which the brume2 basically just is, no extra firewall things like IDS, I think that would be a missing chance for Brume3.
Snort is a zombie project now that Cisco owns it. Everyone switched to Suricata.
That aside: what ever made anyone think the Burme series were ever anything but a gimmick? The Slate AX & Flint v1 blows the Burme v2 out of the water for pure VPN performance with less storage & half the RAM.
I’m very excited to see at least one more Ethernet port onboard. So, the Brume 2 specifications advise <2.6W. Do you know what the power consumption with these specs would be? I’m thinking it’ll be <6 or maybe 7W?
As for the USB port, nowadays most devices like Ethernet adapters, mobile phones, and MiFi devices have USB-C ports.
As for more processing power and stuff, I imagine that would probably increase its power consumption and heat ventilation design. I like lower power consumption devices since they don’t use as much UPS power during power outages.
Just because a port is physically & electrically USB-C doesn't mean it's logically USB-C. There are many phones 'out in the wild' wired USB-C but only transfer @ 480 Mbps (USB 2.0).
A UPS would still use Type-A, USB 2.0, for example. A SSD or NVMe with a 10 Gbps chassis will never reach more than 293.75 MBs, real world, best case because of the 2.35 GbE. 300 MBs is Serial ATA Revision 2 speeds (3.0 Gbps).
SATA II was standardized twenty-one (21) years ago.
Have you factored in the draw of your AP? The Certa (GL-AR750) draws 5W (5V, 1A) < 4W per its advertised specs. They're still in retail channels if one hunts around.
As much as I love USB-C, I think USB-A makes the most sense since many small portable drives will fit perfectly. No one seems to have solved that problem with USB-C.
Is cost a problem? I find the specs very underwhelming especially the RAM. I'm not asking for a $500+ product, but 1GB of RAM for a future product seems totally inadequate. Is the cost difference between 1 and 2 GB significant? 2GB should be the bare minimum and ideally 4GB. Please raise this issue with management. As for the CPU, it feels pretty weak.
I totally agree with you. If the Brume 3 doesn't have more RAM and potentially a better CPU, then I'm out. 1GB of RAM is simply unacceptable in a future-oriented device.
I currently have a RTL8156 2.5GbE dongle plugged into my Brume 2, which is running vanilla OpenWrt at the moment.
I would prefer to see the Brume 3 have a newer CPU than A53, which is a 2012 architecture. I understand that ARM has a strategy of licensing out newer core designs at a glacial pace for vendors who don't want to pay a price premium, but bro. There's just no escaping the fact that A53 is a 2012 core.
The FriendlyELEC NanoPi R5S is currently listed with a 2.0GHz quad A55 (a core which launched in 2017) and 4GB LPDDR4X memory for $109 on Amazon US, and that's with the markup involved in getting it from a reseller on Amazon as opposed to buying direct from FriendlyELEC (tariffs aside).
I'd rather see smaller flash, at least a quad A55 if not something a generation newer, and 2GB memory. FWIW I have a NanoPi R5S running vanilla OpenWrt and serving a 3.5" HDD over NFS, and that quad A55 is very busy just shipping data over the 2.5GbE, without any encryption happening. It's slow. A53 would be even worse.
I do like the low power consumption of the NanoPi R5S - the quad A55 in that device is fanless. The lack of moving parts is a major selling point for me. I want devices like this to last a very long time.
BTW... Rockchip support in vanilla OpenWrt has been improving by leaps and bounds ... I'd be curious to know if GLi is considering Rockchip SoCs for future devices. It seems like support might possibly be less painful than some of the SoCs used in other GLi devices which depend on a proprietary SDK. That's just me speculating, though.
I hear ya. Hopefully, GL iNet will reconsider. As it currently stands, I cannot imagine getting a Brume 3. The specs are horrible. IMO, either get it right, or don't release it at all.
... & that's still ignoring arm64 is now mainstream at ARMv9.
To you point re: Friendly Elec's SKUs, I've watched Van Tech Corner (Youtube) benchmark a NanoPi R6S using WireGuard at 1.1+ Gbps. That's a Cortex-A76 with a Cortex-A55 as a 'quad core by quad core' CPU. I'm sure we all know by now that WG is a multi-threaded process so more CPU cores can only help.
The ARM Cortex-76 is a six (6) year old microarchitecture (2018).
Agreed. This is all the more important when developing for industrial applications. Even the presence of a RTC, which requires a battery that will die regardless of how many years it runs, is enough of a liability to make it a 'non-starter'/completely unfit for purpose.
I just don't have the confidence in GL.iNet to release a compelling Burme v3 when they've already been out-lapped by Friendly Elec. I'll say it again: the only benefit to it is the GUI but that's of little consequence when one can just drop in a theme like Argon onto pure/vanilla OWRT.
(Side note: 18 votes for USB-C... so we know there's at least that many out there that clapped like the trained seals they are when Apple stripped the headphone jack from their phones. JFC.)
The R6S is a pretty good value for the money, but it doesn't have GL.iNet's GUI. There is a market for a Brume 3, but it's not just about the price. It has to provide good value.
A 'mere' VPN gateway stacked with LuCI, WG, PBR, DOH, SSH, SFTP, htop idles @ < 64 MB RAM, < 12 MB storage before any proprietary, third party trash like QSDK/MTSDK or GUI flash.
