Kitz Forum
Broadband Related => Broadband Hardware => Topic started by: Sonac on January 10, 2020, 08:57:29 PM
-
I have decided to ditch my BT Smart Hub so I can run a nice router I have bought; the Ubiquity Edgerouter X, as well as get line stats etc. I am currently using an old Huawei hg612 which I've flashed new firmware onto from this forum. All is working well. I'm also replacing my MK2 master socket with a MK4 in a few days which I've heard good things about.
I've heard from other forums that modems from the likes of Zyxel or Dratek outperform the Huawei due to using different chipsets and there are minor gains to be had. A few extra mbps, nothing major.
I've heard on here that those with the BCM63168 seem to be the best.
Since I have a reasonable slow line of 50/15 (which has slowly gone down from about 65/20 over a few years) every little helps.
Does anyone know what the absolute best performing modem out there is?
Thanks! I'll be making a post elsewhere in the forums soon with some of my line stats etc. I've quite new to all of this and have questions.
-
> Does anyone know what the absolute best performing modem out there is?
I think the ZyXEL VMG 1312-B10A is still unchallenged as a modem - do NOT use these as routers because of the bugs.
Other models in Zyxel’s range are easier to get hold of; those are no longer being sold by ZyXEL but there’s no problem getting them on eBay.
-
Fantastic! Thanks for the suggestion. I will be using whatever modem I get in bridged mode the same as my current.
-
One point to note: the B10D is inferior, it’s the older B10A you need. The suffix-D is the one being sold currently. The A has an analog front end filter which improves performance in some scenarios, the suffix-D lacks this.
-
I tried a few when I was on FTTC from a Huawei cabinet and nothing out-performed a HG612 on my line.
It has to be said, however, that there is not one single device that will be the #1 performer on all lines - so there is no clear answer to your question and you might have to try a few to find the one that is the best one for your line.
-
Okay. That's fair enough. I think i'll change to a B10A then and look for improvements. I've noticed all this disconnecting is having a negative impact on my profile now. I just installed my MK4 master socket (very easy to swap out, took me 2 mins) so i'm gonna leave my line alone for a couple weeks and see if DLS changes my profile.
Thanks for the suggestions!
-
Sonac - like spudgun says the HG612 has a string reputation, it’s just advancing in age now. Our leader herself uses one of the the higher end XyXEL devices as a wireless modem-router (or did, the last time I checked). Be warned though that your line may be too good - in the sense that you may find that changes of modem do not give you any improvement. Also be warned that I’m an ADSL2 user and have an ultra long line 7300m and a mere 3Mbps downstream sync. Other kitizens use the B10A too.
If you wish you can use the enhanced firmware for ZyXEL models built by one of our own members (https://forum.kitz.co.uk/index.php/topic,21545.msg396039.html#msg396039). This is for modem mode only, not for use as a router.
-
Yeah I only put my modem in bridged mode anyway so I'll definitely use that new firmware. I use a Ubiquity EdgeRouter X as my router hooked to a rack mount netgear managed switch which serves my house.
I've put in an offer for a ZyXEL VMG 1312-B10A. But I have found a couple more for a bit more which I can get if needed. But they seem cheap enough.
Fingers crossed
EDIT: Won the B10A for £19!
-
Wo. That’s a good price. Good for you. I’ve only ever done ‘buy it nows’ as I don’t have the patience to cope with auctions these days somehow. I have something like seven B10As, four in service currently and a load of spares in cast of lightning damage which is a constant danger.
-
I'm feeling a bit sad now. Ditched both Zyxel VMG3925-B10B units in favour of the ECI on my Zen line and the HG612 on my Plusnet line.
Plusnet has gone to 56.1Mbit (still slightly under the handback threshold) from 54.8Mbit, Zen is at 66.1Mbit from 62.2Mbit. I don't think its just random resyncs as I power cycled the Zyxels a few times recently and they seemed fairly consistent.
Since I first implemented the Zyxel modems, crosstalk has gone up and perhaps the lack of the filters in these models has become detrimental? That or the ECI cabinets prefer the older modem firmware.
-
Sorry for my ignorance but I've been looking in this firmware talked about from Johnson. I can see a main benefit being these Baby jumbo frames which I also looked up and found a great explanation of how it works from Weaver. However I don't understand what the benefits of this system? Why does it matter if it's 8 bytes more? What exactly will this get me?
Thanks!
-
Hi
Sorry for my ignorance but I've been looking in this firmware talked about from Johnson. I can see a main benefit being these Baby jumbo frames which I also looked up and found a great explanation of how it works from Weaver. However I don't understand what the benefits of this system? Why does it matter if it's 8 bytes more? What exactly will this get me?
Thanks!
It is not going to make any noticeable difference. To explain it using an analogy:
You use a delivery company that has a maximum weight limit for each package of 1500 grams, and all the companies they use also in the process of moving the parcels also have the same weight limit. Now lets assume your labelling and other documentation always weighs 8 grams. When the package gets to the delivery company, it weights in at 1508 grams, over their very strict weight limit. They give you two options now, either send it back to you because it is over their weight limit and you repackage it, or they separate the package into two parcels, causing a delay and potentially breaking the contents! Clearly neither is ideal. So the solution is to make sure when you send packages out, you work on the basis the packet itself must weigh no more than 1492 grams, so that when your labelling is added it never exceeds 1500 grams.
The irony is, the delivery company removes your labelling and replace it with an electronic barcode with no weight, so what leaves them only weighs 1492 grams, i.e. you could be sending an extra 8 grams of goods for the same price.
Baby jumbo frames is the equivalent of organising with the delivery company to change their process, rather than reject the parcel over 1500 grams, they allow you to deliver to them parcels up to 1508 grams. They except these slight heavier parcels, and it causes no problems elsewhere as when they remove your labelling, the weight is 1500 grams, and everyone is happy.
Now the above is only going to make a difference if every time you send a parcel it is always 1500 grams in weight, if most of your parcels are well under the weight limit anyway, that arrangement doesn't give you any benefit.
This is why adding an extra 8 bytes to the packets you send will not see much difference, if any at all. Of course some of us like to optimise everything to the absolute maximum. Networks tend to work on the basis of 1500 bytes being the maximum packet size, the reason we sometimes have a 1492 limit therefore is because of the way we first need to deliver the packets from our modems to the first drop, as there is an additional 8 bytes of address data, which gets removed when those packets go onto the internet. Jumbo packets is a sort of stretching of the specification we do locally, this is why it is often not supported or requires a bit of messing round to do, because it is out of the ordinary.
Regards
Phil
-
That's the best analogy I've ever seen.
-
Hi
It is not going to make any noticeable difference. To explain it using an analogy:
You use a delivery company that has a maximum weight limit for each package of 1500 grams, and all the companies they use also in the process of moving the parcels also have the same weight limit. Now lets assume your labelling and other documentation always weighs 8 grams. When the package gets to the delivery company, it weights in at 1508 grams, over their very strict weight limit. They give you two options now, either send it back to you because it is over their weight limit and you repackage it, or they separate the package into two parcels, causing a delay and potentially breaking the contents! Clearly neither is ideal. So the solution is to make sure when you send packages out, you work on the basis the packet itself must weigh no more than 1492 grams, so that when your labelling is added it never exceeds 1500 grams.
The irony is, the delivery company removes your labelling and replace it with an electronic barcode with no weight, so what leaves them only weighs 1492 grams, i.e. you could be sending an extra 8 grams of goods for the same price.
Baby jumbo frames is the equivalent of organising with the delivery company to change their process, rather than reject the parcel over 1500 grams, they allow you to deliver to them parcels up to 1508 grams. They except these slight heavier parcels, and it causes no problems elsewhere as when they remove your labelling, the weight is 1500 grams, and everyone is happy.
Now the above is only going to make a difference if every time you send a parcel it is always 1500 grams in weight, if most of your parcels are well under the weight limit anyway, that arrangement doesn't give you any benefit.
This is why adding an extra 8 bytes to the packets you send will not see much difference, if any at all. Of course some of us like to optimise everything to the absolute maximum. Networks tend to work on the basis of 1500 bytes being the maximum packet size, the reason we sometimes have a 1492 limit therefore is because of the way we first need to deliver the packets from our modems to the first drop, as there is an additional 8 bytes of address data, which gets removed when those packets go onto the internet. Jumbo packets is a sort of stretching of the specification we do locally, this is why it is often not supported or requires a bit of messing round to do, because it is out of the ordinary.
Regards
Phil
Thanks Phil. Fantastic explanation. It's certainly nice to have everything optimised to perfection, it will be nice to have this feature.
I installed my MK4 Master socket a couple days ago. Unfortunately all the disconnecting has had a negative impact on my DLS profile and I've dropped a bit of speed. Once I connect the B10A i'll leave my line alone for a couple weeks and with any luck i'll be getting the most out of my line as humanly possible.
-
1500 bytes, so 15 more bytes - 1% so 8 bytes = 0.533 % extra payload. So you have to send an IPv4 + TCP or IPv6 + TCP header which is 40 or 60 bytes bloat per packet (per 1460 or 1440) bytes or 52 or 72 bytes depending on whether the particular TCP connection is with/without TCP timestamps. If you can send 8 bytes more payload there is a reduction in the number of wasteful IP+TCP headers sent, that’s why it’s a slight improvement. But half a percent is going to be quite impossible to measure. There are other theoretical reasons why having 1508 byte MTU is helpful which are to do with compatibility with systems at the other end that get a belly ache if you cannot handle 1500 byte IP PDU size. (An ‘xx-PDU’ is a packet including the xx-header and any other reformatting, as received by the xx layer from the layer below, and an ‘xx-SDU’ is the payload as given to the xx layer from the layer above so without the xx header
-
You're forgetting any latency penalty for fragmenting the packet, that potentially magnifies the cost of a smaller MTU.
-
Hi
You're forgetting any latency penalty for fragmenting the packet, that potentially magnifies the cost of a smaller MTU.
Thankfully fragmented packets do not happen often due to Path MTU discovery, and so in many cases MTU may be much lower than our requested supported 1492 or 1500 and we are told to make it smaller, this again means supporting 1500 as opposed to 1492 is unlikely to make much difference. Also because PPPoE is so popular, a lot of networks we connect with over the Internet will limit their MTU to 1492 (or less) anyway in order to avoid any chance of fragmentation or problems.
A good tool for checking a router and firewall is working correctly with Path MTU discovery is below. On my connection MTU discovery is working for IPv4 and IPv6. IPv6 doesn't support fragmenting packets (en-route) so it is important MTU discovery works on that.
http://icmpcheck.popcount.org/
http://icmpcheckv6.popcount.org/
Regards
Phil
-
Don't believe all the hype on the Zyxel modem/routers. I have the fabled 1312-B10A and found it performed worse than my beloved HG612's. If it works out to be faster for you, then that's great - but they don't work for everyone.
-
Don't believe all the hype on the Zyxel modem/routers. I have the fabled 1312-B10A and found it performed worse than my beloved HG612's. If it works out to be faster for you, then that's great - but they don't work for everyone.
It's certainly true that they are not the best for everyone, but they undoubtedly outperform other modems for many people on this forum.
I have the VMG8924-B10A, which has the same Broadcom chipset, and I believe the same front-end filter as the VMG1312-B10A. I have only found one other modem/router (Netgear D6220) that performs as well on my line, and believe me I have tried a good few!
My only real criticism of my 8924 is that the wireless strength isn't the best - it's adequate for my needs but I know that others are better.
As I prefer a single box setup it's the best compromise device that I have been able to find to date.
:)
-
Yeah, I suppose the moral of the story is to try it and see what works best.
I may give it another shot, we'll see.
-
Dee.jay is right, I have my B10A running on slow ADSL so I can’t say anything about any merits in good quality short VDSL2 lines. My previous DLink DSL-320B-Z1 modems were very good - in some respects better, in some worse.