Menu
Reply
On our wavelength
  • 31
  • 1
  • 8
Registered: ‎01-12-2014
Message 1 of 6 (230 Views)

Superhub performance and comparisons (SH2ac vs 3)

Here's my SH3 running things;

medium.png

As you can see, it's saturating and very laggy.

Luckily I was in a  position to re-instate my SH2ac. This is what a 2ac looks like in comparison;

sh2ac.png

 And further improved by then putting the 2ac into modem and using a crappy £20 TP-link wifi router to do the actual routing in a much more efficient manner than the 2ac can do.

2ac modem.png

All spikes in graphs are SpeedTests being performed which typically max out the connection. Looks at how clean and low even the maximum latancy has become. Nothing really ever peaks over 40ms and gaming lag is now non-existant.

Why are these "superhubs" out-performed by the cheapest £20 routers that the likes of Tesco sell?

My connection is now great - thanks to Tesco's Smiley Happy

Reply
0 Kudos
Fibre optic
  • 329
  • 13
  • 118
Registered: ‎31-07-2015
Message 2 of 6 (211 Views)

Re: Superhub performance and comparisons (SH2ac vs 3)

[ Edited ]

You have to remember that BCM graph has little to do with actual router performance. All it shows is ICMP response. It can be used as rough connection benchmark - but not to compare completely different network configs, like modem and route mode.

You should know that most often you get "more latency" when you ping the router itself than when you ping another device *through* the router.

This is simply because router does not place answering pings to high priority. It prioritises routing packets *through* it.

That's why it might appear that you get more latency in router mode (when BCM pings SH itself) than when you use cheapo third-party router in modem mode (when BCM pings that third-party router through SH).

 

Reply
0 Kudos
Fibre optic
  • 495
  • 9
  • 126
Registered: ‎22-06-2012
Message 3 of 6 (172 Views)

Re: Superhub performance and comparisons (SH2ac vs 3)

Looking at the BQM is appears in changing from your Hub 3, to 2ac that you actually changed your managed CMTS. The first BQM looks a lot like a utilised area (although I've seen a lot worse). So in this instance, the puma 6 issue wasn't the worst thing on your list!
--------------------------------------------------------
Look behind you, a three-headed monkey
Reply
0 Kudos
Superuser
  • 10.87K
  • 237
  • 1.98K
Registered: ‎15-09-2009
Message 4 of 6 (148 Views)

Re: Superhub performance and comparisons (SH2ac vs 3)


Guybrush85 wrote:
Looking at the BQM is appears in changing from your Hub 3, to 2ac that you actually changed your managed CMTS. ...

Changing Hubs on same cable segment won't typically change CMTS provisioning as that requires physical re-segmentation.  However that top SH3 BQM does look like it only has max latency more compatible with 8/10 bonded D/S and some evidence of SNR/FEC noise ripple(probably U/S) so I suspect there is a significant time/date difference of far more than the 4days between the SH2AC Router v Modem BQM's?  

Saturating D/S on 8 channels on 70mbps Tier is a lot more easier(noticeable) than on 16 or more - but unless the OP has the DBG (Downstream Bonding Group) from the SH3 stats we probably won't know for sure.

Here's an illustration from SH3 TBB  that is provisioned from brig19 ArrisE6k CCAP/CMTS with 20 D/S bonded (8MHz) channels (with centre frequencies from 170.75MHz to 322.75MHz inclusive):

 

My Broadband Ping - brig19-220/20-bn37n_SH3RM

Here's SH2AC TBB on same cable segment (and same CCAP/CMTS) that is currently locked on first 8 D/S channels of the node/cable segment channel plan (viz 170.75MHz to 226.75MHz inc):

 

My Broadband Ping - brig19-V200/BN37N-SH2ACRM

NB These are live graphs so may disappear at some point in the future due to mac domain,resegmentation or line card changes.  The Average(BLUE) spikes are regular hourly SamKnows monitoring samples and one can see the much publicised Puma6 max latency effect onICMP on SH3 but UDP latency and RTT down/up et al on SK Reporting shows approx a 4mS increase in latency on SH3 which still has 104% Average Daily Downstream rate on advertised V200 Tier compared to only  101% on SH2AC.  

The comparisons between PeakTime are even more noticeable when we compare "hour-of-day" SH3 then SH2AC over say August2017:

Screenshot 2017-09-11 21.30.54.png

 

 

Screenshot 2017-09-11 21.31.35.png

 

 

Regards Tony
"Life is a Binary Inspired Turing Computed Hologram"(don't PM or @Mention me - in case ignoring you offends)
Reply
0 Kudos
Fibre optic
  • 495
  • 9
  • 126
Registered: ‎22-06-2012
Message 5 of 6 (118 Views)

Re: Superhub performance and comparisons (SH2ac vs 3)

Interesting. I had in the past moved from one hub to another and when I checked which CMTS I was on (can't say how long after I did this mind you), I had moved CMTS so just tied the 2 together. Sounds like work had been done in the area at the time though, but all I know is I wasn't complaining, that's for sure!

The max latency outside of peak hours on the hub 3 BQM certainly would imply 8-12 DS channels bonded, then again someone on the puma 6 thread with 24 DS channels (if we take it at face value) has a BQM as clean as any on a 2ac.

--------------------------------------------------------
Look behind you, a three-headed monkey
Reply
0 Kudos
Superuser
  • 10.87K
  • 237
  • 1.98K
Registered: ‎15-09-2009
Message 6 of 6 (72 Views)

Re: Superhub performance and comparisons (SH2ac vs 3)


Guybrush85 wrote:

Interesting. I had in the past moved from one hub to another and when I checked which CMTS I was on (can't say how long after I did this mind you), I had moved CMTS so just tied the 2 together. 

Understandable conclusion based on empirical observations probably many of us "CATV wrinkly's" have previously encountered over two decades of DOCSIS in it's various releases. 

FYI you only have to look at some of my archived threads to see how CMTS can change eg here's one from 7yrs ago: 

BRIG9 > BRIG16 > BRIG15 Resegmentation issues

and another from nearly 3 yrs ago to date:       BRIG15 > BRIG19 . Utilisation progression

 We can see how SNR/FEC issues also can be co-lateral effects of some of these re-segmentations/upgrades and how those noise issues can last from only a month to sometimes 5 months in extreme cases!

Regards Tony
"Life is a Binary Inspired Turing Computed Hologram"(don't PM or @Mention me - in case ignoring you offends)
Reply
0 Kudos