dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
1816
share rss forum feed


Bill_MI
Bill In Michigan
Premium,MVM
join:2001-01-03
Royal Oak, MI
kudos:2
Reviews:
·WOW Internet and..

WebStar vs. Arris

Click for full size
Click for full size
We added phone on Sat May 12 which changed my trusty old Scientific Atlanta Webstar DPC2100R2 to an Arris TM602G.

You can see the gap while I casually worked out the new data-taking script.

I wasn't keen on Arris but 1 week of data says the two are pretty comparable on this DOCSIS 2 service. The biggest difference is the Arris only reports levels to the nearest whole number which makes the plot look a little different.

"In God We Trust... all others must have hard data".


RootWyrm

join:2011-05-09

Yeah, Bill, that's pretty much why I won't use an Arris 600 series. Their programmers didn't know what the hell they're doing and they were lazy. The tuner config in the TM602g is technically superior to the DPC2100r2, but, well, you see your graphs. I consider the whole number thing a bug; I forget if it's upstream or downstream levels where it always rounds up no matter what. (I have it buried in my notes.)

You might try the Arris TM722g; I believe they're available at BeastBuy, it should be provisioned as though it were a TM602g, and it has a 'Maximum' backup battery for >=8 hours of battery operation.



Bill_MI
Bill In Michigan
Premium,MVM
join:2001-01-03
Royal Oak, MI
kudos:2
Reviews:
·WOW Internet and..

Hi, RootWyrm. Yep, that hits the nail on the head. The word "hokey" comes to mind when I see formatting like this:

Freq/Power: 609.000 MHz4 dBmV

But I know the web-UI and firmware may be totally unrelated in quality. That's sort of what I'm concluding... it's not as bad as I thought.

Recently read how phone service has to be done with WOW equipment, though. It still feels good to be without AT&T for the first time in my life.



Bill_MI
Bill In Michigan
Premium,MVM
join:2001-01-03
Royal Oak, MI
kudos:2

Just wondering if anyone has found a way to clear the Arris event log? It has a table of 20 entries and seems to only update the last 1... or 2... or 3 of them very strangely and unpredictably.

Just another buggy behavior.



RootWyrm

join:2011-05-09

said by Bill_MI:

Just wondering if anyone has found a way to clear the Arris event log? It has a table of 20 entries and seems to only update the last 1... or 2... or 3 of them very strangely and unpredictably.

Just another buggy behavior.

There is no way to fix this, and it is buggy behavior. They also truncate when remote logging (or just break completely.) The log is supposed to continually overwrite, but it does not do so, and occasionally just fills up, causing the modem to crash and requiring a power cycle. Low quality does not begin to describe Arris' codebase. But as I've mentioned before, I suspect these are the same people who produced a firmware (as S-A) that literally bricked the units and took entire headends out.

The only reason WOW requires their gear for phone is because their provisioning and engineering departments are incompetent and/or lazy. The TM722g is 100% compatible when provisioned with a TM602g configuration - it falls back to D1.1 mode as required by DOCSIS spec, and does present valid certificates for EMTA functions including SIP login. The only thing they may need to do is add the TM722g's device certificate as permissible - which is the unwritable certificate stored on the CPE itself.
Edit: Should clarify, Arris hasn't changed their vendor-specific SNMP either. Not in a long time. (Which is still no reason to push it to Moto and Cisco CPEs.)


Inflex

join:2002-09-05

said by RootWyrm:

The only reason WOW requires their gear for phone is because their provisioning and engineering departments are incompetent and/or lazy.

Got anything to back that up, or is that your opinion that you are stating as fact?


RootWyrm

join:2011-05-09

said by Inflex:

said by RootWyrm:

The only reason WOW requires their gear for phone is because their provisioning and engineering departments are incompetent and/or lazy.

Got anything to back that up, or is that your opinion that you are stating as fact?

They've known for over a year that the Cisco DPC2100r2 configuration file attempts to set Arris specific strings which causes errors in the DPC2100r2's log. Takes less than two minutes to change in PacketAce or take your pick of config file editors.
Still hasn't been removed.
There's a bunch of errors in the docsDevFilterIP section which are the result of sloppiness and laziness; filtering 135-139 instead of individual filters for NetBIOS. (i.e. why the hell are you filtering loc-srv?) Broken port 80 redirect which inhibits Akamai GHost operation in certain situations.
Hell, if they want, I'll GIVE them a corrected configuration file sans HMAC-MD5. (Basically the password ensuring you can't run an unapproved config file.)

Motorola SB6180 can run off the SB6120 DOCSIS configuration flawlessly; last I checked, still not permitted. Tuner hardware is just two 6120's linked in-modem. Still a standard D3 modem, still obeys D3 configs. Operationally it disables unused tuners so, yeah, no RF issue.

DPC3k's are not permitted; same deal. They are D3 compliant. The DPC3008 and 3010 will work fine with the same config as an SB6120. Just need to set it up as a catchall on model string "DPC30" and you've got 'em both covered.

There's arguments for and against modem diversity, as it introduces management overhead. However, in the case of Motorola, once you have SB6120 you pretty much have all of them. It's just a matter of changing model strings or appropriately wildcarding.
WOW also argues they don't do it because they don't supply firmware for these modems. Well frankly, given the firmware they supply for the DPC2100r2 - one with known, publicly disclosed security holes - and the glaring security holes in the UltraTV appliance? I view that as a positive.

Oh, and D2 segments are still horrifyingly insecure, because WOW has refused to enable BPI+. There is only one layer of protection, apparently because using even basic CVC and enabling even BPI baseline is too hard.

FYI: I used to work for an MSO many many times the size of WOW.


Bill_MI
Bill In Michigan
Premium,MVM
join:2001-01-03
Royal Oak, MI
kudos:2
Reviews:
·WOW Internet and..

Thanks, RootWyrm See Profile... I'm learnin' sompin.

Notice the channel changes? Are these usually for load balancing? They seem excessive. I think I correlated them to 2-3 second connectivity drops but not conclusive. I'm a EE with no exposure to cable systems except just dinkin' around like this.



RootWyrm

join:2011-05-09

said by Bill_MI:

Thanks, RootWyrm See Profile... I'm learnin' sompin.

Notice the channel changes? Are these usually for load balancing? They seem excessive. I think I correlated them to 2-3 second connectivity drops but not conclusive. I'm a EE with no exposure to cable systems except just dinkin' around like this.

My pleasure!

You are absolutely correct that there are issues with excessively frequent channel changes. SOME channel movement is absolutely normal; as frequencies get more crowded (more modems online) or suffer neighbor bleed (channels nearby) or degrade due to environment (heat, cold, rain, storms, etcetera) it's expected that a modem will automatically switch from for example 603MHz to 615Mhz.

As I have mentioned to WOW in the past, it seems to be an issue of overcrowding and overloading in the existing channels due to a combination of poor planning and poor provisioning practices. Generally the preference should be to order channel advertisements to modems from the center outward, or to alternate top/bottom based on subscriber profile/modem/take your pick, if you cannot randomize. (Arris C4's are incapable of randomized configurations.)
UltraTV does Video Over DOCSIS (VOD) which can really, really hammer a channel hard. Combined with the very low QA/QC performed by Arris and by WOW on these devices, along with trying to shoehorn DOCSIS 2 and 3 into the 603-and-up band? Yeah. I imagine a vampire tap would catch fire from what's going on in there.

The channel hopping is markedly worse over here the past few days, but yes, a channel hop - especially upstream - WILL cause the modem to pause all traffic. Given you've got an EE background, well hell, you know it better than I do. Coaxial is divided into 6MHz bands below a certain frequency (upstream/T-channels) and 6.4MHz bands above. Remember your frequency-division multiplexing for high frequency devices? Yup. Pretty much start there. WOW uses SCDMA to my knowledge - Synchronous Code Division Multiple Access. DOCSIS / North American cable systems operate in a range of 7-49MHz T channel (passive return) and 55.25-1001.75MHz for powered transmit.

I've also notified them twice here about the invalid configuration on upstreams, which they don't seem to get. They have one upstream channel at 16QAM and the other at 32QAM. This is wrong! I don't give a crap what the idiots at Arris said - it's wrong. All upstreams must be the same modulation - either 16 or 32 or possibly 64QAM for some SCDMA/TDMA scenarios. In a mixed modulation environment combined with possible upstream overcrowding - 2 channels is likely insufficient, I believe that gives WOW a total upstream capacity of ~62Mbit for all CPEs in the headend - you end up with excessive channel hop and the modulation change on upstream will almost always trigger the modem to hop downstream channels unnecessarily and also creates those long pauses because it has to resync the upstream due to the modulation change.

Unfortunately, it's also a rock and a hard place situation. To offer more channels, even with virtual channels available, requires more frequency. More frequency for TV means less frequency for modems. Optimally WOW needs to have 4 upstream 6.4MHz 32QAM SCDMA channels split 2 for D2, 2 for D3 phasing to 4 combined. However, they may not HAVE 4 upstream channels available.
For downstream, there should be a similar arrangement except in a 4/4 phasing to 3/5 split and eventually phasing to 6 with D2->D3 replacements. This is what I call a "Pipe Dream" scenario, because we already KNOW that WOW does NOT have the frequency available to do this. Hell, NOBODY has that much frequency available. Unfortunately, rather than acknowledge problems or get proactive about the system, they have been sticking their heads in the sand, looking at a few meaningless stats and going "la la la looks fine to us."


Bill_MI
Bill In Michigan
Premium,MVM
join:2001-01-03
Royal Oak, MI
kudos:2
Reviews:
·WOW Internet and..

1 edit

said by RootWyrm:

I've also notified them twice here about the invalid configuration on upstreams, which they don't seem to get. They have one upstream channel at 16QAM and the other at 32QAM. This is wrong! I don't give a crap what the idiots at Arris said - it's wrong. All upstreams must be the same modulation - either 16 or 32 or possibly 64QAM for some SCDMA/TDMA scenarios. In a mixed modulation environment combined with possible upstream overcrowding - 2 channels is likely insufficient, I believe that gives WOW a total upstream capacity of ~62Mbit for all CPEs in the headend - you end up with excessive channel hop and the modulation change on upstream will almost always trigger the modem to hop downstream channels unnecessarily and also creates those long pauses because it has to resync the upstream due to the modulation change.

Fascinating. Never noticed! I'm seeing Arris upstream:

Channel ID 3/31MHz QAM16
Channel ID 4/37MHz QAM64

Took a look at my old data and the DPC2100R2 was exactly the same. I use the "Channel ID" numbers from the DPC2100R2 as the Arris has only frequency - then I add 10 for sanity in the plot.

I know little about QAM except I assume QAM64->QAM16 should be a faster switch than QAM16->QAM64 since the latter is switching to a much more complex algorithm to get in sync. And yes, I see both up and down switching simultaneously. If one doesn't switch it seems to be a roll-of-the-dice coincidence.

I was in TV and FM broadcasting in the 1970s while in college so I'm familiar with the 6MHz channel space and spectrum of the analog TV signal. You can think of cable as the same RF spectrum on controlled copper - plus the digital steroids of today. That 6MHz standard still survives with something like up to a dozen standard highly-compressed digital streams.

But upstream is totally a cable phenom and quite limited. Upstream (used to be called "reverse") can get to the FM broadcast band? Never knew it could get above 50MHz or so. There was much in the 1970s when all this was being worked out and the worry was the loose-cannon customer equipment leaking into other services. I'm sure today's levels and spectrum has roots from that. FM was particularly loud so I guess they lost out.

Thanks for the detailed info. I'll be looking up some of the acronym soup.


RootWyrm

join:2011-05-09

said by Bill_MI:

Fascinating. Never noticed! I'm seeing Arris upstream:

Channel ID 3/31MHz QAM16
Channel ID 4/37MHz QAM64

Took a look at my old data and the DPC2100R2 was exactly the same. I use the "Channel ID" numbers from the DPC2100R2 as the Arris has only frequency - then I add 10 for sanity in the plot.

Yeah; out here they have it configured incorrectly on Channels 1 and 2. QAM16 and QAM32 respectively, as I recall. (I'd have to check my notes.) The DPC2100r2 can resync faster and has a bit more tolerance, so it can be a bit less noticeable. Well, until it hits the other errors. Then it likes to lock up tighter than Fort Knox on attempted resync, forcing a power cycle.

said by Bill_MI:

I know little about QAM except I assume QAM64->QAM16 should be a faster switch than QAM16->QAM64 since the latter is switching to a much more complex algorithm to get in sync. And yes, I see both up and down switching simultaneously. If one doesn't switch it seems to be a roll-of-the-dice coincidence.

That may be the case, but, beyond my area of knowledge there as far as the modulation. Guaranteed that on the ones where downstream doesn't switch, it's just selecting the same downstream channel. May be semi-randomized in the modem itself. The biggest problem is the frequency syncing, which is the long pause - trying to line things up again. If timing quality is poor at the modem or CMTS - and it's usually the modem - it can take a very long time.

I was in TV and FM broadcasting in the 1970s while in college so I'm familiar with the 6MHz channel space and spectrum of the analog TV signal. You can think of cable as the same RF spectrum on controlled copper - plus the digital steroids of today. That 6MHz standard still survives with something like up to a dozen standard highly-compressed digital streams.

Hooray! Like I said - you probably know that part WAY better than I do. I know the basics of that stuff, and the painful details of PacketCable, but beyond that? I got the basics in RF safety and operation, and a lot more on channel allocations and frequency management. Not so much on the contents or methods.

But upstream is totally a cable phenom and quite limited. Upstream (used to be called "reverse") can get to the FM broadcast band? Never knew it could get above 50MHz or so. There was much in the 1970s when all this was being worked out and the worry was the loose-cannon customer equipment leaking into other services. I'm sure today's levels and spectrum has roots from that. FM was particularly loud so I guess they lost out.

Yeah, in the cable space it'll go all the way to 49MHz typical, but can go higher. I think absolute upper bound is 52MHz using alternate method. (Normally last T-channel is 14 at 43-49MHz with alternate being 14 at 46-52MHz.) Depends on the amplifiers in the headend and some other factors. The passive return channels originally were for public service broadcasting. You know, those TV shows on the public access channels, shot in someone's basement? Yup. Gear transmits it up to the headend over T-channels. Federal regulations require MSOs to allocate a certain number of T-channels to public access/public service, plus local regulations may require additional ones. It's a non-optional thing and they can't be multiplexed.

Thanks for the detailed info. I'll be looking up some of the acronym soup.

My pleasure!


Bill_MI
Bill In Michigan
Premium,MVM
join:2001-01-03
Royal Oak, MI
kudos:2
Reviews:
·WOW Internet and..

It's confirmed with about 5 examples. I watch Leo Laporte's TWiT stream (»twit.tv) and it has these occasional 2-3 second pauses. It's CHANNEL CHANGING data loss. About the same on Arris as WebStar.

I checked my old Comcast data. In 2010 we had *7* channel changes total for THE WHOLE YEAR and some were power outages. Heck, WOW gets that in 12 hours.



mix

join:2002-03-19
Utica, MI

I think this is an issue that WOW_Dan could pass on to an engineer if these frequency changes are so frequent you are noticing the 2-3 second pauses during normal usage (streaming a few hours of video seems like normal usage to me).

Would the 2-3 second pauses disappear if you had a DOCSIS 3.0 modem with bonded channels?



Bill_MI
Bill In Michigan
Premium,MVM
join:2001-01-03
Royal Oak, MI
kudos:2
Reviews:
·WOW Internet and..

said by mix:

Would the 2-3 second pauses disappear if you had a DOCSIS 3.0 modem with bonded channels?

Completely out of my protocol knowledge but would like to hear from someone that knows.

I wouldn't want to put Dan on the spot but certainly welcome his viewpoint. I like to make directly observed facts be known and see what happens. Above is the data.


WOW_Dan
Premium
join:2011-03-24
Naperville, IL
kudos:32

said by Bill_MI:

said by mix:

Would the 2-3 second pauses disappear if you had a DOCSIS 3.0 modem with bonded channels?

Completely out of my protocol knowledge but would like to hear from someone that knows.

I wouldn't want to put Dan on the spot but certainly welcome his viewpoint. I like to make directly observed facts be known and see what happens. Above is the data.

A DOCSIS 3.0 modem would help that, yes. It would be using all the channels at the same time, so instead of locking to one and flipping around, it just balances which frequency it is using, but stays locked onto all of them all of the time.
--
Dan Della Terza
WOW! Internet, Cable & Phone
Network Operations Center


Bill_MI
Bill In Michigan
Premium,MVM
join:2001-01-03
Royal Oak, MI
kudos:2
Reviews:
·WOW Internet and..

Thanks, Dan. Good to know it stays locked on changes and DOCSIS 3 can change any of them with little interruption. This means DOCSIS 3 can change some bondings pretty regular with very minor affect.

Which leads me to the theory a DOCSIS 2 with a single lock may be getting caught up in the DOCSIS 3 system fast balancing activity. So the dual nature of the system penalizes D2. Anyone care to whack at my theory?