New upload speeds and Network Management

[Ignitionnet]

Originally Posted by Welshchris

Virgin must not be updating all the areas on the fasterupload page then, as there are atleast two mentioned on here thats been done that doesnt appear on the faster upload page.

The entire area doesn't get upgraded at once it's on a per-port basis. People here may have just been fortunate and been in the early upgrades. No point in putting sites on that list when just a few cards have been upgraded.

---------- Post added at 11:51 ---------- Previous post was at 11:49 ----------

Originally Posted by theoldbill

To VM,

Great news about the increased upload speeds and 100mbit on the way, but please can you deliver the 50meg/1.75 first, many thanks

Sincerely, another customer on a 100% utilised upstream cable (Telford cmts 11 cable 020).

These upgrades will provide the additional capacity you need. Indeed I suspect quite strongly that this is why there has been a spike of upstream congestion issues - they are pressing on with these more comprehensive upgrades which take a bit longer than simple business as usual capacity relief.

[Chrysalis]

Originally Posted by Ignitionnet

Indeed I suspect quite strongly that this is why there has been a spike of upstream congestion issues

can give more insight please?

[Ignitionnet]

Originally Posted by Chrysalis

can give more insight please?

Just seem to be more reports here and on official forum, same as there was an uptick of downstream issues pre-overlay network release with its' big downstream capacity increase.

No insider knowledge here I'm afraid just a possibly not very educated bit of speculation.

[Chrysalis]

I think I understand you now, you mean normal capacity relief work is on hold so the speed upgrade work can get done?

When a port is upgraded what happens to the users on it whilst its been upgraded?

Ignitionnet, when they say my specific cable is fully utilised, how many customers would this be serving typically and what would be the fix - resegmentation or light another cable altogether etc??

[pip08456]

Originally Posted by theoldbill

light another cable altogether etc??

Light another candle!

[Ignitionnet]

Originally Posted by Chrysalis

I think I understand you now, you mean normal capacity relief work is on hold so the speed upgrade work can get done?

When a port is upgraded what happens to the users on it whilst its been upgraded?

You mean while the resegmentation / physical work is being done? Same as always, they go offline.

The upgrade comes in 3 parts.

1) Physical network work - resegmentation, noise busting, in some cases rebuilding.

Rebuilds and resegmentation are service affecting, modems are physically disconnected from their CMTS / uBR, noise busting can be service affecting as groups of modems may need disconnection to find the sources of noise.

2) Change of configuration on CMTS / uBR to 16QAM DOCSIS 2 upstreams.

Virtually non-service affecting. When the configuration change is made the CMTS / uBR will let all modems know the new parameters and they will obey them from that point on.

3) Change of tier to new upstream levels.

Initially not service affecting but will only take effect when modems are rebooted. Eventually a brief service impact - modems will be remotely disconnected to apply the new tiers when they reconnect.

Step 1 is prone to issues - resegmentation can cause issues due to new network characteristics.

Step 2 is the key - when this configuration change is made some modems may fall offline as their return path isn't good enough for these upstreams. The error rates on the upstream must be monitored, either automatically or manually, and the changes rolled back or upstreams changed to QPSK until issues are isolated and resolved.

Step 3's only risk is that the upgrades are not sufficient for capacity demands, however this is likely to be apparent after Step 2 so this risk is minimal.

You may all rep me now (Yes I'm joking).

[Hugh]

You mean the network fairies don't just sprinkle magical optical fibre dust, and everything works?????

[pip08456]

Originally Posted by Hugh

You mean the network fairies don't just sprinkle magical optical fibre dust, and everything works?????

VM have them on contract in case step 3 doesn't work!

[Chrysalis]

ok so on step1 they just stay offline during the work, dont get temporary moved to another port? thats the bit I was interested in thanks.

[Ignitionnet]

Originally Posted by Chrysalis

ok so on step1 they just stay offline during the work, dont get temporary moved to another port? thats the bit I was interested in thanks.

Resegmentation means that a part of the customer base will be moved to a new port. While they are being moved they are offline.

Those not being moved will usually stay online.

You're talking a matter of minutes of downtime, certainly no more than an hour.

EDIT: I had a thought. To move them to another port temporarily would be no different from a resegmentation onto a new port. Pointless and unnecessarily complicated as you'd need to put new IP scopes onto the receiving port just for the sake of saving customers less than an hour of downtime. RIPE would slap VM with the clue stick if they thought they were doing this and it would make address management a total gimboid.

---------- Post added at 13:56 ---------- Previous post was at 13:51 ----------

Originally Posted by theoldbill

Ignitionnet, when they say my specific cable is fully utilised, how many customers would this be serving typically and what would be the fix - resegmentation or light another cable altogether etc??

Typically, could be anything up to say 100.

With these upgrades the fix is a bit of both along with some more. A second upstream is brought online to split the load between your existing one and that one, these upstreams are upgraded to give twice the capacity of the existing ones, and depending on needs, mostly downstream for 100Mbit, there may also be some resegmentation, splitting some of the existing customer base onto shiny new line cards.

[pip08456]

Originally Posted by Ignitionnet

With these upgrades the fix is a bit of both along with some more. A second upstream is brought online to split the load between your existing one and that one, these upstreams are upgraded to give twice the capacity of the existing ones, and depending on needs, mostly downstream for 100Mbit, there may also be some resegmentation, splitting some of the existing customer base onto shiny new line cards.

I presume that it would also relieve a big part of upstream congestion, at least in the beginning depending on the uptake of 100Mb.

Does the 50Mb upload use a second upstream on upgrade?

[Ignitionnet]

Originally Posted by pip08456

I presume that it would also relieve a big part of upstream congestion, at least in the beginning depending on the uptake of 100Mb.

Does the 50Mb upload use a second upstream on upgrade?

The upstreams aren't bonded, it's two upstreams with customers split between them.

[Chrysalis]

ok I had hoped the reason why my port had sudden congestion (Wasnt gradual, just suddenly appeared) was because users were temporarily put on for these upgrades. Based on what you have said this is unlikely?

[Ignitionnet]

Could have been a migration from the DOCSIS 1.1 network, or just the student effect.