You missed the s (in https)normis:
https://mt.lv/News107

Love to see this hAP ax2!

Why not use Noctua or similar fans with a lower noise factor as standard? I mean, are they really that much more expensive? Then you wouldn't need the JayzTwoCents mod for a normal office server room..

Why is ax2 slower in routing over ac2? ax2 has powerfull CPU...Is it some mistake in test result?

finally ax! but sadly no 6ghz / 6e

Why is ax2 slower in routing over ac2? ax2 has powerfull CPU...Is it some mistake in test result? tests.jpg

RoS 6 uses an ancient Linux kernel that still have route caching. RoS 7 uses a fairly new kernel (5.6.x, if I'm not mistaken), and routing cache was dropped from Linux some time ago. It was done for security reasons, and will not come back.

So, RoS 7 is slower than RoS 6 in SOME use cases - especially when routing lots of traffic to a single host. Sadly routing speed tests are one of the use cases that are affected by this.

When will we get AX devices that support 802.3af PoE? I have a huge demand for these with standard PoE input (aka not 24v passive)...

Do we have any anticipation of feature-set to be included with CAPsMAN for AX devices? Any expected date that Wave2 will be fully integrated to allow my APs to be replaced with AX hardware?

完全同意,默认的球迷在CRS和CCRseries are terrible - also going to replace mine with noctuas
- not a big fan of the front port labeling on the hap ax2, I II III IIII IIIII, just confusing

It is just so frustrating how you guys are releasing new products while neglecting CAPsMAN development.

With all due respect, as a corporate user and home prosumer of your products I don't care about running docker containers on a router board with 256MB ram rather have good stable wireless coverage using my multiple access points.

Wave2 hardware been in your boards for years.
Is it supported by CAPsMAN? NO
Is it fully developed and has all the usual advanced features? NO
Would it work on limited resources boards with wave2 hardware nonetheless? NO

Then why did you release it. AX will suffer the same issue and this is why it took you guys ages to release it to market. I bet AXe won't come till 2025 maybe if I'm hopeful. With all the same restrictions.

The CAPsMAN feature has been forgotten. Neglected. It barely functions. What are you guys waiting for? The tiniest example about issues with it is pressing reselect channel for example. Does nothing. All caps default to the same channel sometimes. Speeds are disappointing with multiple users on HAP AC3.

Just why? Can some senior exec at Mikrotik share with me why?
You guys are over diversifying your product line without a solid base nor clear vision. This is what I feel.

I'm willing to endlessly chat with your top level management. Heck I'm even willing to join the team and help you guys out.

Competitors are thousands of kilometers ahead of you and still manage to get things right on time. Ubiquity, Aruba, TPLink, NetGear to name a few.
Go buy those products you might say and pissoff after you read my reply. I tell you NO. I love this brand. I love the hardware and routeros been using it for decades and will do for decades more BUT MOVE YOUR ASSES GUYS! We need you to do better!

A shiny youtube channel and polished videos won't do you better than streamlining your portfolio and narrowing down your focus.

This is not a negative feedback. I just want better up to my expectations. Again I'm even willing to help as much as possible!

Cheers.

Can't have a hAP in the ceiling. When is the cAP/wAP ax ready?

I'd appreciate a bit more detailed info about power consumption (as standard part of product specification). It's nice to know that if I get e.g. hAP ax2 as home router, it won't need more than 27 W, while routing over 2 Gbit/s, pushing 0,5 Gbit/s over IPSec (guess) and feeding 17 W to another device using PoE. It's not bad for all that. But it will do this, roughly, never. What it will actually do, for half of all time, will be sitting and idling. With few connected cables and occasional packet here and there. It would be nice to have some idea how much will that require. There's a difference whether it's 1 W or 10 W. With electricity getting annoyingly expensive, people start to care about that.

I understand that there are too many possible combinations, number of connected cables, wifi on or off, clients connected or not, etc. But if nothing else, at least some "connected idle power consumption", defined as (from top of my head, I welcome better suggestions) ethernet ports connected to something and alive, enabled wifi with no clients, and no or near zero traffic, would help. If there were few more examples, even better.

Glad to see POE out. Sad to see it's only 2-pair 28V. That alone eliminates its usefulness with any of the CPE I use to provide more than 150Mbps of bandwidth, where AX would really shine. Those all need 48V (or 4-pair 28V).

I hope we see 48V POE In/Out on the (yet to be announced) hAP AX^3 with beefy (1000mA-1500mA) power supplies.

Why is ax2 slower in routing over ac2? ax2 has powerfull CPU...Is it some mistake in test result? tests.jpg

RoS 6 uses an ancient Linux kernel that still have route caching. RoS 7 uses a fairly new kernel (5.6.x, if I'm not mistaken), and routing cache was dropped from Linux some time ago. It was done for security reasons, and will not come back.

So, RoS 7 is slower than RoS 6 in SOME use cases - especially when routing lots of traffic to a single host. Sadly routing speed tests are one of the use cases that are affected by this.

But routing lots of traffic to a single host is what I want to use the AX2 for

No USB, so nice otherwise, why? :(

Maybe a soon to be released Plus version with USB and 802.3af PoE :)

When outdoor AX products for wisp?

CAPsMAN is being worked on. By the time we have cAP ax, we will also have CAPsMAN.

Good news !
Any chance to be a bit more specific on the timing ?
September ? Still 2022 ? 2023 ?

Never ask for the timing :-)
Also don't get too enthousiast when "it is being worked on"... BFD for RouterOS (a minor feature compared to CAPsMAN) is being worked on for almost a year (viewtopic.php?p=877008#p877008), yet still no sign of it in the beta...

hAP ax² - Passive PoE ?
Do routers have Passive PoE?
As far as I know, routers always support Active PoE

CAPsMAN is being worked on. By the time we have cAP ax, we will also have CAPsMAN.

Excellent. Looking forward to this. Thanks :-)

@normis
This newsletter has very exciting devices especially the ax stuff ... IMO the Chateau LTE18 ax AND the Chateau 5G ax will be a real winners assuming RoS7.x and the ax drivers are truly ineffectivesync when out the door and in the hands of the consumer. BTW, the Linux Kernel is now in v6 RC1 and I hope that your dev team are nee deep etc. ......

But routing lots of traffic to a single host is what I want to use the AX2 for

Yes. You, me and 95% of the SOHO market. But this change was made at kernel level, and will not be reverted - too many headaches and security issues.

Although there is a bright side: we can trust the numbers on the results page - this makes easy to choose what we need.

By the way: I upgraded my home router (one hEX) to RoS 7, and didn't see any changes in speed. So far, so good.

Why should they? This is why RB4011 and similar devices exist. I'm almost certain that there'll be something like a RB5011 in the near future. (RB5009 + ax)

Nice to see ax, I was forced to go with Grandstream for the last year or so just because some customers required ax...
Although this is more home device and features like SFP etc just do not apply there...
What I would like to see next is CAP ax but with one input and other output PoE and preferably 2.5Gbps Ethernet ports because 1Gbps is already too slow for Wi-Fi 6...

Good afternoon. Do you know what's missing from this great product? SFP or SFP+ port! Many ISPs run optical cable in apartment buildings. Direct connection is a nice bonus. Do you plan to add these ports?

On the other hand, more and more ISPs run GPON/XPON technology and the use of SFPs or other equipment not provided by the ISP is becoming more and more of a can of worms...
The local ISPs here recommend to use their supplied NTU that converts the fiber to 1Gbps ethernet and wire it with UTP. May seem unclean, but you have a clear demarcation point between network and router (the 1Gbps ethernet) that you can trace and connect to another router to troubleshoot intricate connection problems.
When you use direct SFP+ in the MikroTik router and it happens not to work (or stops working), it will be impossible to debug for them and you will usually get no support.

It is a bit like the DSL SFPs that we were discussing a while ago. No line status display on MikroTik routers so impossible to really use.

Good afternoon. Do you know what's missing from this great product? SFP or SFP+ port! Many ISPs run optical cable in apartment buildings.

What kind of SFP/SFP+ module are you expecting, then ? Some GPON or similar ones ? Ethernet ones ? Others ?

I think in real-world scenarios is very unlikely to saturate a 1gbps ethernet interface with wifi traffic on a small AX access-point like this.

Beyond specific benchmarks, under ideal conditions, an access-point like this will be far from pushing 1gbps or considering 1gbps a practical real limitation or flaw.

Maybe in a bigger Access-point with multiple 5ghz/6ghz capable interfaces but that's another tier completely.

Right now I'm using the AC2 to max out my gigabit internet (to a single PC). I really hope AX2 won't be a downgrade for me...

Does Mikrotik have any plan to fix the ROS7 performance regression? I mean, patch the kernel if you have to right?

Does Mikrotik have any plan to fix the ROS7 performance regression? I mean, patch the kernel if you have to right?

No. Only a more capable hardware will be able to offer similar performance. It's as simple (or as complex) as this.

Guess I'll just skip the AX2 for now until the performance regression is fixed.

For my 50-200Mbps customers, I deploy hAP AC^2. For those 300-700Mbps ("gigabit"), I deploy hAP AC^3.
At my home office, I've used RB4011 (~1Gbps), CCR2004(~2Gbps), and now I have a CCR2116 (have to test 2-10Gbps).

Guess I'll just skip the AX2 for now until the performance regression is fixed.

The ax²isthat more powerful hardware mkx was referring to.Compared to the ac²,它有一个更快的时钟频率,20.6%运行ning on the more efficient 64-bit ARM instruction set besides. That should be more than enough to buy back the performance loss relative to ROS 6.

我有2 x hAP ac2在多个安装。我是first happy with the ax2 announcement, but having synology nas at home due to upgrade, I wished for a 2.5 Gbps ethernet there. Maybe my next future solution is the RB5009 and some ax2 CaPs, once available ....

That should be more than enough to buy back the performance loss relative to ROS 6.

If that were true, it would show in the performance specs... but it doesn't.
Apparently the performance loss of v7 (in such tests! that does not mean it is also true in normal usage...) is more than 20%.
That should be measurable by using a v6-capable device and measure it before and after upgrade to v7.

Agree. And IMO re-testing should be done by Mikrotik. Apparently they have an established test infrastructure and protocols and would be fairly easy to re-do tests. This way the v7 tests would actually compare apples to apples, any other tests, done by individual customers, would highly depend on actual configuration of test devices.

Minimum would be to re-do tests on all v6 devices that are still being sold (to avoid buyers' complaints about sub-par performance). Courtesy to all their customers would be to re-do testing of all devices supported by v7.

I can imagine MT would be reluctant to do it as test results will be notable lower than they are now. But this way they could avoid lots of negative feedback after customers discover performance degradation them selves.

Yes, adding a separate "test results under RouterOS v7" section to the specs of every router originally released and tested with RouterOS v6 and still being sold would be the best.

... and more people will stay on 6 and complaints will diminish :lol:

Good afternoon. Do you know what's missing from this great product? SFP or SFP+ port! Many ISPs run optical cable in apartment buildings.

What kind of SFP/SFP+ module are you expecting, then ? Some GPON or similar ones ? Ethernet ones ? Others ?

Actually it would be nice if MT could consider implementing EPON support directly in the products. Symmetric 10G/10G-EPON would look like the best choice for the future at this point. No special SFP with its own CPU running its own OS inside, just an optical transceiver for the right PON Tx/Rx wavelengths over a single fiber, and do all the PtMP Ethernet TDMA magic in the device running RouterOS. Implement both the PON client (ONT) and ISP (OLT) side, to replace the proprietary devices (but also inter-operate with them if needed). I'd suggest EPON rather than GPON because it's a simpler Ethernet based open standard.

¡Paquete自由!✊

What kind of SFP/SFP+ module are you expecting, then ? Some GPON or similar ones ? Ethernet ones ? Others ?

Actually it would be nice if MT could consider implementing EPON support directly in the products. Symmetric 10G/10G-EPON would look like the best choice for the future at this point. No special SFP with its own CPU running its own OS inside, just an optical transceiver for the right PON Tx/Rx wavelengths over a single fiber, and do all the PtMP Ethernet TDMA magic in the device running RouterOS. Implement both the PON client (ONT) and ISP (OLT) side, to replace the proprietary devices (but also inter-operate with them if needed). I'd suggest EPON rather than GPON because it's a simpler Ethernet based open standard.

Do you know if XGS-PON (ie Symmetric 10G/10G-EPON) passive splitters currently exist ?

Hello to All !

hAP ax2 - use only 20/40/80 ? Without 160 ?

One has to start from somewhere I guess. Dual-chain hardware is probably the most affordable and available to get for the first devices...

New performance test result is so much better now for AX2, thank you Mikrotik!

New performance test result is so much better now for AX2, thank you Mikrotik!

It gets better, but it's not much faster than IPQ-4018/4019.