For those many of you who are experiencing issues with TCL cable-modem internet performance, can I suggest you hang in there for a couple of weeks? 

I know most of you will anyway because face it, unless you live within a couple of km of a Telecom exchange then cable internet absolutely hoses all over anybody’s ADSL – Unless the cable operator tries really hard to screw up the HFC cable-modem system, and this accurately summarizes what TCL has done for the last 6 or 7 years.

However, TCL have within the last 2 weeks completed a modernization upgrade to HFC cable-modem and CATV transmission equipment in the Wellington HFC network – along with a CMTS augment – that will already be making a difference in some areas of Wellington and Kapiti. This augment itself doesn’t solve too many problems but it does provide the platform for cost effective upgrades that are easy to implement and which will improve cable-modem performance in the Wellington area.

The major issues currently affecting cable-modem performance are: The ARP issue, the size of the nodes (directly affecting in the “contention ratio”), and CMTS through-put. ARP Issue: Most of you are aware that TCL runs the cable-modem system as one huge flat VLAN, hence the ARP issue. The fix for this issue is to segment the VLAN into several smaller VLANS. TCL have trialed a fix (called VLAN segmentation) for this issue, and the result of the trial was better than hoped for.

The reason TCL have not fixed the ARP issue in recent years was not a technical issue, nor even a financial issue, it was a contractual issue with a number of business customers who had blocks of IP addresses from TCL. These contractual issues have been worked through, and the VLAN segmentation project is being implemented this week and next week. Yes, I’m saying the ARP issue should be resolved by end of next week.

Give it another week after that to allow for the usual delays and issues. Node Size Issue: TCL HFC network in Wellington has node sizes of up to 2000 homes/node. This is huge by international standards and quite an issue. TCL have planned and recently committed to a project to reduce node size (node segmentation). Each node will effectively get split into either two or four nodes depending on current size.

This split is made possible by the equipment recently installed in the HFC network and with a modest capital outlay. TCL absolutely do have plans to start segmenting nodes this financial year, and some should be completed prior to Christmas. I cannot understate the difference this will make. All your peak time woes should disappear. CMTS Through-put: Current CMTS units used in the TCL network have a 100Mb/sec Ethernet connection to the IP cloud.

It would be fairly accurate to say that currently, on average, there is one CMTS per node. One CMTS with its one Fast-Ethernet connection has a max through-put of possibly 60-70Mb/sec on a good day, so it doesn’t take a genius to work out that two or three 25Mb/sec (Warp Speed) customers on any one node will probably stuff the node. However – DOCSIS 3.0 is coming.

This is definite and it is coming soon-ish (I have no direct knowledge of a timeframe on this but when I find out I’ll tell you). Whatever other goodies DOCSIS 3.0 brings, it will also force TCL to upgrade CMTS units and the new units have Gigabit Ethernet connections - so end of CMTS through-put issues (for now). More when there is more to say.

[Moderator edit (MF): segmented, split this very long single paragraph post!]