I wonder where the calculation of "IP profile = Sync speed * 96.79%" is done?

by the CP / BT Wholesale I think, so that it reflects the amount of user data that can pass downstream at the frame size they use. BRAS units have that task.

"For example, if the reported downstream VDSL2 data rate is 40,000kbit/s and the IP
packet size is 1500 bytes(i.e. Ethernet frame size at End User LAN is 1514 bytes) the
maximum throughput achievable is actually 39,178kbit/s (when measured at the EU
LAN i.e. no VLAN header, but including Ethernet header). Alternatively, if the IP
packet size is 64 bytes the maximum throughput achievable is further reduced to
35,721kbit/s. This overhead is particularly important to consider in respect to the
downstream shaper setting on the CP�s BRAS. CPs are advised to understand and
account for the method that theirBRAS uses to implement traffic shaping."

So, it looks to me like the Openreach DLM sets the sync speed, and the BTW Bras sets the IP profile.

In reply to a post by BatBoy:

So, it looks to me like the Openreach DLM sets the sync speed, and the BTW Bras sets the IP profile.

Yes indeed. Openreach set the DSLAM up to send sync speed / data rate information upstream at connection time, but it's BTw or other CPs that have to decide what if anything to do with it. Openreach advise that traffic shaping is used to limit the downstream data to what the line can handle.

where is the variance ?

The DLM is only concerned with the line speed. It has no knowledge of PPP. I'm happy with that. DLM works on errors, resyncs and the like - data from the DSLAM.

The IA in the DSLAM inserts the speed into the PPP or DHCP session negotiation. The CP can use that to define downstream data rate profiles.

Please explain.

I see, a different page I hadn't read. Thanks.

Perhaps the conflict between http://www.robertos.me.uk/html/bt_wholesale_dlm.html and the Kitz diagram should be resolved one way or the other. Your page says "The two main things it sets are the �target� noise margin and the IP Profile. Please see the links in the menu on the left for details." whereas the Kitz diagram shows the output of DLM going to one or more other systems before arriving at the BRAS where the IP Profile is actually set.

Edited by yarwell (Wed 13-Feb-13 21:04:25)

1. Please list the others.

2. Indeed, Batboy (or someone) raised the ADSL issue presumably to identify "BTw DLM". I agree it is irrelevant to FTTC sync speed, but relevant to IP profiles which is exactly what this thread is about.

3. The setting of the IP profile at the BRAS is not controlled by the BTw DLM. It uses data (the sync speed) from the DSLAM which results from the actions of the DLM.

Where exactly is this second DLM ? For clarity, by DLM I mean something that acts upon the line speed from the CPE to the DSLAM.

I think (hope) we're all clear that DLM relates to LINES and IP Profiles relates to downstream traffic shaping ?

where is the variance ?

Quite. No variance in principle. But there is variance in terminology that is causing the confusion. At least I'm sure the OP has long left the post is isn't getting more confused.

We all agree (I hope) on the basics.

1 - There is a low level DLM at the cabinet that controls the cabinet->modem connection and does almost all the interesting stuff. The immediate effect of this is most easily seen (with unlocked modem) by the sync speed.

2 - There is higher level control (IP profile) at the BRAS. This is set from the sync speed on pppoe connection. It is presumably designed to ensure that the cabinet is not given data faster than it can pass it on to the modem. It is also used to limit speeds based on the maximum the service contract is willing to provide (eg 10/40).

3 - The setting of (2) from (1) is at pppoe connection time; sometimes (but not always) forced when (1) forces a low level resync.

4 - Too much playing with the modem can affect the low level DLM.

5 - Playing with the router can force pppoe resync (which can be beneficial). It will never harm the low level DLM. It will never harm the higher level IP profile. It may lower it, but only where the sync speed is already low.

Terminology variance comes as to whether we use the term DLM for the BRAS system. It clearly is DLM: it's digital and it manages the line (or that depends what you call the line). Nevertheless, I would personally avoid the term DLM for this and reserve it for the low level cab->modem DLM. I don't know if this is common practice. The confusion is made worse as in the ADSL world the two were inextricably intertwined.

~~~

Unknown (to me) is whether the BRAS/IP profile ever gets 'stuck' for reasons that are caused neither by the sync speed nor by the contract terms. I'd still appreciate it if anyone could give more clarity on that. There clearly are cases where people have an IP profile much faster than the speeds they actually see; even after a router reset should have made the IP profile match the sync speed. In some cases these low speeds don't seem to be fully explained by congestion, local wireless setup, etc.

Edited by StephenTodd (Wed 13-Feb-13 21:06:12)

Dynamic Line Management is usually used to describe the parameters that affect the sync from the DSLAM to the CPE. Googling for "Dynamic Line Management" generates a consensus on this.

I think is confusing and unnecessary to consider anything after the DSLAM as a "line" personally. A "line fault" ought to be something involving a bit of wire and not a setting in a database 100 miles away.

I'm not sure the DLM does live at the cabinet, when I looked in one there wasn't a box that would fit the bill unless it's a feature that the DSLAM has internally (not that its location matters much).

Any large database seems prone to errors IME, so a "missing" sync value leaving a profile sat at the wrong setting in the BRAS would not be a shock. In terms of throughput measured vs sync or profile, I'm usually drawn back to the 20M / 30M provisioned capacity and CIR on the cab - exchange fibre, is that coming into play anywhere ?

You can also get horribly low throughput despite decent sync if the error rate is high.

In reply to a post by yarwell:

Dynamic Line Management is usually used to describe the parameters that affect the sync from the DSLAM to the CPE. Googling for "Dynamic Line Management" generates a consensus on this.

The Huawei SmartAX MA5616 MSAN has several dozen(?) different 'port' parameters which affect the sync from DSLAM to CPE.

Just to put the cat among the pigeons, Huawei never explicitly refers to DLM, or at least I can't find a reference.

There is however reference to Dynamic Optimisation Profile/.s It seems that the MA5616 can be configured with up to 256 different optimisation profiles, and each profile can be bound independently to one or more ports of a subscriber line card.

cheers, a

I hope we may get near to a resolution of the disagreement here.

The Kitz links initiated by you re the BTW DLM for ADSLx , (as I get the impression you believe in the gospel truth of Kitz), intimate that the DLM controls the BRAS line Profile and IP Profile settings.

My view for a long time, which I have mentioned elsewhere in these forums in the past but no idea when or where, ever since I started wondering where the IP Profile for FTTC is coming from, is that there is a simple lash-up at the MSAN as it receives the GEA link data. Whereby the line sync value supplied by the OR DLM is fed onwards into the WBC DLM and thence to the BRAS, rather than the sync negotiated by its own line card for ADSLx connections.

At no point have I disputed that the BRAS is where the value is held - though posts in the past by Plusnet staff clearly point to there being more than one place. (Quite apart from the Plusnet copy).

My statement that two DLMs are involved is therefore valid. The OR one controls the line sync. The WBC one controls the throughput using the IP Profile. Both DLMs are involved.

If you wish to argue that the WBC DLM does not control the throughput on either ADSLx or FTTC, then sorry - I'm trying to explain the system in useful terms to people who want something they can understand. You seem to want all explanations to be at the PhD thesis level.