Hello,
Is hEX refresh efficient enough for internet 1000 Mb/s / 100 Mb/s?
Not modified much default firewall. NAT.
“Ethernet test results”
Thanks, I know that, but I’m more interested in real world throughput.
Look at the results for the 512 kbp packet size and there are basically your real world results. Just depends on your config as to what results you’d see.
Thanks for confirming the advice I found earlier - to look at 512B column in the last line.
Nowdays, Mikrotik has again created a device that cannot handle currently popular consumer fibre connections.
RB5009 is too powerful and too expensive. Will Mikrotik give something relatively cheap, for home use, that will handle a 1Gbps connection?
You’re demanding a lot from a 50 dollar router. You’re not going to get that kind of performance from any brand for that price or else it’ll be totally stripped down with no firmware/software updates and capable of doing nothing more but straight routing with no customizable firewall rules, vlans, etc. You could go with a hap ax2 which can deliver those speeds + wifi and can be found for less than 100 dollars.
https://mikrotik.com/product/hap_ax2
https://multilink.us/mikrotik-hap-ax-wifi-6-wireless-router-us/
Maybe poke around and look at Mikrotik’s various offerings before claiming Mikrotik has missed the mark.
You’re just ranting. Have you had the device in your hand and made a benchmark? No of course not. You came here to complain only. 50$ is apparently too much of a high investment to try out. Shops like getic have a 30 day return policy. No risk. Or even better: ask your favorite distributor whether hex refresh can handle gigabit ffth.
Regarding the results: I don’t know who came up with the 512byte column is the one to look at. It reflects the packet size used in the benchmark. MTU on any typical FTTH connection is around 1490 +/-. So read the benchmark results again. You don’t even need to extrapolate and have a degree in statistics. Take the 1518 column and reduce it by 10%. Then you have your values for your FTTH. IMHO. Anyone can come here and prove me wrong. Taking the 512byte column is the worst of worst case scenario I don’t know when and why this scenario should apply. Of course this is an all port benchmark and single ports are 1gigabit (full duplex). So these numbers may need to be cut in half to get throughput of one direction.
Ziter’s mention of the RB5009 as if it were the next alternative to the Hex R was a giveaway. The hAP ax2 has nearly twice the performance of the Hex and WiFi 6 radios as well for less than twice a Hex R, less than half the price of an RB5009.
All of the above are great value in their niches.
AX3 is better for you
TPLINK TL-R5010PE-EN with the same soc can route at 3.594Mpps @ 64Byte with HWNAT enable.
FWIW … from PC using wAP AX as AP and Hex Refresh in between towards RB5009 as router with iperf3 container … I was able to get 900Mbps.
Hex wasn’t even sweating.
But that’s with Hex Refresh in managed switch mode (read: VLANs being used) NOT using ether1 as uplink !
In that same config using ether1 as uplink, speeds go down drastically (peaking sometimes towards 500-600Mbps but rarely more, average 363 Mbps).
So pay attention to block diagram to decide what you connect where and what you use it for.
I can’t believe. That would make this “refresh” completely use-less for anyone living in 2024.
I’d like to quote Viktors from the YT video: https://youtu.be/Zrzq_zPWoQ4?si=Yv26gyvw3A5GUWmY&t=134
When I buy a router I really want to use it as a router not as a switch. But also I really want to squeeze all the speed out of those ISP cables.
And I can’t believe that 363Mbps is the “squeeze all the speed out of those ISP cables”. This must be some kind of misconfiguration, wrong testing setup or Viktors is talking marketing bla bla.
Could be a config issue but I don’t think so.
These results are a pure consequence from having that ether1 port directly connected to CPU and not to switch chip.
Keep in mind: this was a MANAGED SWITCH config so everything VLAN not being HW offloaded to switch, has to pass CPU.
Again:
I used the EXACT same config when testing only swapping ether1 and ether2. Nothing else.
Only difference: how ports are handled (or not) by switch chip.
I never tested anything about that device as a router. I trust the test results they post are realistic.
Rule of thumb: 25 filter rules for 512Kbps packets
Hex Refresh = about 500Mbps
Old Hex = about 265Mbps (and that one also had some odd config with 1/3/5 on 1 bus and 2/4 on another)
That was a software problem (no HW NAT offload, which the SOC could do). The Ubiquity EdgeRouter X & -X SFP, which are identical to the hEX & hEX S hardware-wise, can do a gigabit just fine with NAT offload enabled. Not in both directions at the same time, though; the internal connection is a 1Gbps bus,
In case used as a router it would have to pass CPU anyway. An c’mon, with these few ports who would buy such a device as switch? You can buy dumb unmanaged switches in this form factor for 10$.
Unmanaged switch ? Not for me, thank you.
I wanted a managed switch running ROS on arm 
Serious
purely for home lab purposes. To replace old Hex.
And W?BIC.
Why ? Because I Can.
HEX refresh is a router. It does not even have Switching results in the Tests tab. So it is not primary intended to be a switch. Yes, it has a switch chip built in (EN7562CT) and can do HW offload. But again: It is a router.
You write that I complain and that what do I want from a router for $50?
Only I wrote that Mikrotik’s offer lacks a relatively cheap router for gigabit internet. I did not write that it should cost only $50. Why create hEX refresh with such parameters in 2024 for $50, maybe it should be a bit more expensive, but efficient enough for 2024.
Some of you propose hap AX2 for $80-100, but I do not need wifi - I want an even better price for equipment that can handle gigabit internet without major investigations.
Also, thanks to some of you for your clarification on throughput.
Well
We use the “512 bits, 25 rules” as a standard not because it’s the best case: We use it EXACTLY because it’s the worst. The idea is to have some room for growth.
I had on hEX Gre3. It did 980Mbps up or down (never tested both at same time) - with NAT and PPPoE. There wasn’t any trick: just a lean firewall, using fast track as much as possible. It’s successor should be even faster. Well, it has gigabit ports, so won’t go faster than that - but one could use a more elaborated firewall, or be sure to reach better speeds with really fragmented traffic.
Here, a list with the ethernet routers Mikrotik sells. One of them must be good for Your needs.
https://mikrotik.com/products/group/ethernet-routers
But:
Do You want to “just” route up to 1Gbps? No complex firewalls, no weird queues, no acessing a NAS through wireguard? If so, the hEX refresh will be enough. Bear in mind: there will be not much resources left on the router to do more than this.
Do You want more? Then I’d suggest an RB5009 or better - depending on your needs.