Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[kitz]

Update.

Damnit there isnt a SNMP service - so no MRTG 

Routerstats working, but not via http, can only be done via telnet.

[burakkucat]

It reads as if you have a little job for Asbokid. 

[kitz]

Ive spent a fair time on this, but Ive now come to a dead end,  it would seem that* Thompson in their wisdom decided to take out the SNMP service from the router to make room for some additional security features.

That sucks, because to me that is a really stupid move.
Now maybe Im just being too simplistic in my approach but if because of all the additional code thats been added to routers over the past few years (ie wireless/security) then surely it cant be too hard to increase the memory size.

You know I sometimes think that router manufacturers really dont think things through...  Ive only been saying for about 8 years I could never understand why router manufacturers didnt implement gigabit ethernet in the switches because it would cost them literally pennies to do so at time of manufacture.. but instead it took them about 7yrs to catch on that most networks used gigabit rather than 10/100.

So now they are running short on memory.. hmmm how much would it cost to put a wee bit more on the board?  How cheap is flash memory?
Me thinks that the software developers should perhaps talk with the hardware manufacturers more often.
 



BTW  I believe the reason why routerstats doesnt show the stats page properly now is because of a security feature which changes the page login frame url for each access.

So now because of just one simple thing, Im disappointed with the router.


*Thanks Nicola @ Be* forums

[asbokid]

It reads as if you have a little job for Asbokid. 

The current generation of CPE chipsets don't seem up to the task of demodulating the DSL signal and then routing the traffic at gigabit speeds.  That is presumably why the Openreach eggheads at Martelsham decide to deliver their 17MHz VDSL2 service by splitting the load between a modem and a separate router.

The modem (the HG612/ECI) recovers the ethernet frames and bridges them still at layer 2 (ethernet) to/from the domestic router.  The router is then left to re-write the headers at layer 3 (TCIP/IP, etc).  Still a very processor-intensive task, though.

There are much more pokey* powerful MIPS core chipsets which could be used in the CPE - the SiByte MIPS64 cores, for example, as used in the Huawei DSLAMs, but they gobble power, and can need active cooling.  Or there are ARM11 cores, which power the Huawei HG8240, the Openreach FTTP CPE optical modem-router. [1] That device also has a separate 6 port Gigabit ethernet switch controller, the Atheros AR8316. [2] The same controller is apparently used in the high-end Mikrotik RouterBoards.  So consumer-grade gigabit ethernet equipment is readily available at a price.

This is a mini project of mine.. building a low-cost gbE router using a passively-cooled dual core AMD64 industrial mainboard with a quad port gigabit ethernet controller on the PCIe bus. [3]

As consumers, we are somewhat stuck between a rock and a hard place.  But at least Openreach had the foresight to recognise the problem of today's gutless range of modem-routers. With the earlier 8MHz VDSL2 profile it wasn't so obvious, but as sync speeds have increased, the problem has become more evident.

cheers, a

* EDIT: I always thought that "pokey" meant fast. "Your car is pokey, mate!" "Cheers! 'Tis, innit!"  It seems that pokey means the exact opposite! "pokey: without speed or energy; slow"  Ooops!

[1] http://forum.kitz.co.uk/index.php?topic=10561.0
[2] http://www.chipworks.com/TOC/Atheros-AR8316-AK1E_6Port-Gigabit_E-Switch_FAR-1001-901_TOC.pdf
[3] http://www.fujitsu.com/fts/products/computing/pc/accessories/mainboards/extended/

[kitz]

They do exist now, but they just took such a long time to do so.  Ive been wanting a DGN3500 for a while, but something else always seems to have cropped up first.   Netgear does a couple of models, as do some other manufacturers.

I believe the DGN3500 is quite good according to Azzaka, who got his mits on a couple when they were 1st released as a trial.


>> There are much more pokey MIPS core chipsets which could be used in the CPE

The ironic thing is it WAS there, theres mention of SNMP in one of the manuals..  but they seem to have disabled it, by in effect removing the exe file to make room for other things - bet TR69 is another thing taking up room these days.

Thompson have apparently done exactly the same thing on their latest firmware version of the 585v7 by removing SNMP to make room for something else

---
btw have you seen the latest netgear?  it works on cable/vdls or adsl2+, wondered when the day would come that they become combined units to prevent customer choosing the wrong product.

[kitz]

hmmm... want...  how sexy would this look on top of my new PC (when I eventually get it)


http://www.netgear.co.uk/home/products/wirelessrouters/high-performance/DGND3700.aspx#one

All-in-one—High Speed ADSL2+ Modem (built-in) and WAN Gigabit Ethernet port for cable/fiber
Powerful Dual Core (400 MHz each) processor
High speed access to external USB storage using 2 USB 2.0 ports
Memory: 128 MB Flash and 128 MB RAM
IEEE 802.11 b/g/n 2.4 GHz
IEEE 802.11 a/n 5.0 GHz
Five (5) 10/100/1000 (1 WAN and 4 LAN) Gigabit Ethernet ports
Two (2) USB 2.0 ports
One (1) ADSL2+ port
Management:    
QoS
DHCP
SNMP
Security Features:    
Denial-of-service (DoS)
Intrusion detection and prevention (IDS)
NAT   
Other Connectors:   2 x USB 2.0



See look it can be done...  just a shame its over £100, but so were the 3500s when they 1st came out.
Goes to see if I can find some reviews to drool over.  8)

-----------
Edited to add..

Okay..  thats it..   who wants to buy kitzy a new router 
Seen some pretty good reports and claims of better speeds..  its SNR tweakable, doesnt work with DMT tool.
;drool:

[guest]

The problem isn't that chipsets aren't "powerful" enough, its simply that its not currently possible to comply with the "Code of Conduct on Energy Consumption of Broadband Equipment" in the EU while having gigabit ports, even with low power state links etc. If you turn off wireless then it is, but that's not generally an acceptable solution. That's also one of the reasons that all the ISP-supplied ADSL/VDSL kit is generally 802.11n with only two spatial streams and 20Mhz channels (ie 144Mbps max).

I'm not sure what the implications of non-compliance are, but I can tell you that all the major EU ISPs are insisting their router suppliers meet the code.

The piece of kit I'm beta testing (and at least one other person here is too) doesn't have gigabit ports for that specific reason.

You'll have to wait for the next die shrink before all the shiny stuff gets on - either that or buy your own router

[sevenlayermuddle]

The problem isn't that chipsets aren't "powerful" enough, its simply that its not currently possible to comply with the "Code of Conduct on Energy Consumption of Broadband Equipment" in the EU while having gigabit ports, even with low power state links etc.

Hmm, so that would sound like another bit of half-baked Save The Planet legislation that has backfired then, as many people who currently want gigabit  & wireless adsl will simply run two separate boxes... the sum energy footprint of which will presumably be far greater than a single box that served both functions?

[roseway]

I believe the DGN3500 is quite good according to Azzaka, who got his mits on a couple when they were 1st released as a trial.

Unwilling as I am to disagree with Azzaka, I have to on this occasion. I had one of these, and after the initial euphoria it was a big disappointment, with frequent PPP drops and failure to reconnect. The Netgear forums were burning up with complaints from users about this router, with numerous people returning the router for a refund. The Netgear response was absolutely dire. I've still got mine, and you're very welcome to have it if you'd like to try it.

The DGND3700 has a better reputation, but I still advise a read of the Netgear forums on the subject: http://forum1.netgear.com/forumdisplay.php?f=115

[Edit] There was a short thread about the DGN3500 here, in which you'll see that arobertson545 experienced the same disappointment as I did: http://forum.kitz.co.uk/index.php/topic,11006.msg214947.html#msg214947

[guest]

The problem isn't that chipsets aren't "powerful" enough, its simply that its not currently possible to comply with the "Code of Conduct on Energy Consumption of Broadband Equipment" in the EU while having gigabit ports, even with low power state links etc.

Hmm, so that would sound like another bit of half-baked Save The Planet legislation that has backfired then, as many people who currently want gigabit  & wireless adsl will simply run two separate boxes... the sum energy footprint of which will presumably be far greater than a single box that served both functions?

For now I agree, however its a step in the right direction. 100million households (or whatever in the EU) x 30W = a few new power stations

Big ISPs would simply go with the cheapest option; it appears with this then they have to consider something other than price. We'll see.

I turn off wireless on pretty much every ISP-supplied router as its garbage - I then use another router/AP for wireless only. I don't use the switch function on the router at all, but even if it was gigE it wouldn't save me anything as there's gigE switches around the house anyway. I accept I'm not what you'd call a "normal" user though

The current bit of kit under test will change that as the coverage is exceptional even though its only 144Mbps PHY @ 2.4Ghz on the wireless. Its pretty damn good at dynamically changing channels when it sees interference, need to give it a bit of a kicking to see how good.

[kitz]

Thank you eric for the info..  I had heard that the 3500 had some initial problems, but I thought that it was resolved by the updated firmware.  Quite a disappointment for many there then, because they arent a particularly cheap router. I may be wrong, but I thought Zen had some input on the new FW.

Theres a few more threads in that link for me to explore yet, but I notice they reakon a 3700v2 and 3800 could be on the agenda soon... so we shall see.   Its all pie in the sky for me atm..  it gave me something to drool over whilst recovering from the disappointment of no SNMP in the latest thomson routers.

>> Hmm, so that would sound like another bit of half-baked Save The Planet legislation that has backfired then, as many people who currently want gigabit  & wireless adsl will simply run two separate boxes..

I so agree, I have had to run 2 boxes for about 6 years for that very reason.

[guest]

I so agree, I have had to run 2 boxes for about 6 years for that very reason.

When you get to 4 x 8 port gigabit switches and 2 x FE switches then you can comment on "saving the planet" by ignoring CoE regs

I gave consideration to trying to reduce the number of switches/making a nice neat patch panel setup. The problem with that is a single ethernet connection/room doesn't really work. About the only place it may work is in a master bedroom where you could put 2 x FE in via one cable for video stream and occasional PC use. Everywhere else you will need a switch IME if you're centralising backups*/have kids.

*We have an atom-based 8TB Windows Homeserver box. It is the old version based on Win 2003 Server with Drive Extender (which is SO much better than RAID). That backs up all the machines which are powered on between mid-day and midnight, server level management of backups (day, week, month, year settings); restore is put a USB stick in client machine, reboot and pretty much walk away. It has de-duping capability so we have 449GB in use for backups from 5 Windows boxes (XP and Win7). Trying to do that over FE would be a nightmare, gigE means nobody notices. Saved my sanity with kids a lot of times as I can always rollback the machine to stable state then extract the (school) docs they changed prior to disaster.

[guest]

I so agree, I have had to run 2 boxes for about 6 years for that very reason.

The code of conduct is quite recent.

I'd suggest the boxes you've used have had designers who don't see the point of putting more than 100Mbps (or 802.11n) on a device intended to route 24Mbps d/s and 2.5Mbps u/s

Times are changing with FTTC/FTTP.

[asbokid]

Today's unimpressive range of home routers is due to the low power of embedded CPUs for broadband access equipment.

That is why for its VDSL2 service, Openreach chose to disable the routing functionality of the Huawei HG612, even though it is technically a router and a modem.  The device cannot cope with both tasks at higher VDSL2 line speeds.

Instead, Openreach opted to use an ethernet layer bridge from the Huawei. A separate device performs all the routing (the HomeHub, etc).  And yet the HG612 has the most powerful DSL processor - the 400MHz dual MIPS32 core BCM6368 - available today from Broadcom, the industry leaders.

Similar devices used in North America suffer the same performance issues.   AT&T powers its Pace/2Wire CPE modem-routers with a dual-core 400MHz VLIW Trimedia TM3260. At least on paper, that is a much more powerful CPU than the 6368.  Yet American consumers still find the device lacks the power to route traffic at higher line speeds.

The EC Code of Conduct on Energy Consumption is generous enough and doesn't in itself limit the functionality of home gateways.

The Code provides a specific power limit for each function found in broadband access equipment - the 802.11 transceiver, the DSL front end and line driver, USB controller, the ethernet switch, etc.   Each function is given its own power limit, and those limits are summative.

The power level limit for a 4 port gigabit ethernet switch in the current Code (November 2008) is 3.7 watts. And the limit for an 802.11n dual band transceiver is 5.0 watts, twice the allowance for a single transceiver (2.5 watts).

A well-loaded modem-router could possibly use 30 watts and still be within the Guidelines.

The limitations in today's modem-routers are due to technical issues rather than political do-gooders.

cheers, a


EDIT:  From the Code of Conduct [1]

Choose a WAN side interface (or even two or three interfaces)



Now load up the device...



[1] http://ec.europa.eu/information_society/activities/sustainable_growth/docs/broadband_eq_code-conduct.pdf

[guest]

The 30W figure was an example, not a limit. Reduce it to 20W and there's 1GW of power generation eliminated assuming 100m households.

Add the figures up then add in the (usually crap) wall-wart.

That's why some of the big ISPs in the UK are now going back to building in the PSUs.

We'll see, like I said.