For anyone having similar problems, we’ve just discovered a setting that restricted our WAN speeds from 940Mbps down to 250Mbps.

‘URL Content Filter’ under the Firewall ‘Default Rule’ settings.

When a ‘URL Content Filter’ setting is chosen the WAN speeds plummet to 250Mbps max. Selecting ‘None’ will restore full 1Gbps WAN throughput.

This hasn’t been mentioned in any ‘known issues’ release notes that we have seen recently.

This may apply to other settings under the ‘Default Rule’ page like ‘Web Content Filter’ but we have not tested these other settings.


Model: 2927 F/W 4.4.2.3
(May apply to other models too?)

Have you raised a support call with Draytek ? You clearly have a 900Mb+ service. So, you must have HW acceleration turned on. Would be interesting to see what speed you get with your filter off and HW acceleration off. Specs say it should be 800. But, with HW acceleration off i don't get 500 which is the service i have.

Hi Shop,

I ran your requested test several times, over several random days, with only Hardware Acceleration disabled/enabled, the firewall filters are already disabled anyway as I do not require them currently.

I ran the tests from the same node each time (a Windows 10 machine) and ran the tests via each WAN connection (isolated - i.e. no load balancing applied) of a dual-WAN setup on the 2927ax. (Note: we have 2x 1Gbps Virgin Media WAN connections at this location.)

With HA disabled and no firewall default filters applied, I was averaging 350Mbps on pretty much every test.

With HA enabled (TCP & UDP) and no FW default filters applied I was hitting the 940Mbps stated in their specs, although sometimes dropping down to 800 or even 600 at times - but mostly up in the 900s range, possibly a HA max-sessions-limit kicking in, as we have so many nodes running on this environment.

I have seen the specs that state 800Mbps with HA disabled but, I have also seen specs on a DrayTek webinar video that states 600Mbps for non-HA’d WANs and 940Mbps with HA enabled (which included the 2927 models).

…We have many Vigor Router models running across the larger global environment with roughly four 2927s now in-play; 3x 2927ac models and 1x 2927ax, plus some older models at various other sites; all VPN linked.

We have experienced some problems with a couple of the 2927 models and have purposefully switched them around to see if the problems follow that specific router, and it does appear to follow that specific router (all running the same and latest firmware), and yes we have tried rebuilding configs from scratch after initially just backing up and restoring them.

We have raised support tickets with DrayTek a couple of times in the past, even sending our configs to them for analysis, followed by a response of “it’s working fine on our test equipment”, so no, I haven’t raised a ticket for this particular problem and will only do so if I absolutely have to.
Don’t get me wrong, I have had brilliant support from DrayTek in the past, and quite enjoyed discussing matters with them but, the 2927 has been a little different when it comes to support; I won’t go into their External Devices management features on this thread as I could lose focus on this particular titled topic.


I have wondered if there is a hardware gremlin in their production line of 2927s, kind of like how Microsoft had the Xbox red-ring-of-death problem back in the day, but who knows.

I will persevere with DrayTek’s products and hope they iron-out any gremlins in the hardware and/or software areas of their products, at least they are continuously adding to and improving on their products rather than some companies I have dealt with in the past who just give-up on development and leave you wishing for more, more features and more improvements.

What these little boxes can do is pretty amazing at the end of the day, especially if you dedicate the time to work it all out and learn their way of operating. I would rather slim down our carbon-footprint too and run the least amount of power sucking devices, globally, rather than add overkill everywhere with very expensive alternatives offering all the bells and whistles that I will probably never utilise, yet, I do appreciate that they give you some advanced level of stability in their many elaborate features but that’s not my only consideration, comparing the initial cost and continued running cost thereafter is also a factor.

I think I also enjoy the technical challenge of making it work against the odds too.

HodgesanDY wrote:
Hi Shop,

I ran your requested test several times, over several random days, with only Hardware Acceleration disabled/enabled, the firewall filters are already disabled anyway as I do not require them currently.

I ran the tests from the same node each time (a Windows 10 machine) and ran the tests via each WAN connection (isolated - i.e. no load balancing applied) of a dual-WAN setup on the 2927ax. (Note: we have 2x 1Gbps Virgin Media WAN connections at this location.)

With HA disabled and no firewall default filters applied, I was averaging 350Mbps on pretty much every test.

With HA enabled (TCP & UDP) and no FW default filters applied I was hitting the 940Mbps stated in their specs, although sometimes dropping down to 800 or even 600 at times - but mostly up in the 900s range, possibly a HA max-sessions-limit kicking in, as we have so many nodes running on this environment.

I have seen the specs that state 800Mbps with HA disabled but, I have also seen specs on a DrayTek webinar video that states 600Mbps for non-HA’d WANs and 940Mbps with HA enabled (which included the 2927 models).

Hi HodgesanDY
Thanks for checking. Once i get some more time and equipment moved i will re-visit this on my install. The tests i did were pc straight into Draytek lan. At the moment i'm limited to a homeplug connection from the pc to draytek (and i know this will limit the speed test). But i tried someting today which was run the speed test from the pc (over the homeplug) whilst logged into the router gui (HW off) and i looked at the cpu & memory usage. Under these conditions the cpu usage rose from it's nominal 3% to 50%. I then repeated the test with HW on and the cpu usage didn't much change from 3%. In both cases memory usage stayed at it's usual 79%. Is that expected behaviour between those modes ?

Yes, with regards to the CPU usage, the Hardware Acceleration is handling the workload on a separate processing chip, which takes the load off the CPU.

You should really check out this video:

https://draytek.co.uk/our-solutions/videos/draytek-webinars/draytek-router-wan-performance-and-hardware-acceleration

This should help you troubleshoot any issues you might be having with Hardware Acceleration, although your goal seems to be achieving specified speeds without Hardware Acceleration enabled.