On 2nd November 2009 we advised that, starting 4th November, Entanet was monitoring new traffic prioritisation measures for approximately three weeks. To remind yourself of the detail please see the NOC notice of 2nd November.

The feedback we have received strongly supports Entanet’s decision to continue to apply the measures as a means of improving the user experience of the majority of customers. As a result our prioritisation measures will continue to be applied. Where customers report access speed issues, we have encouraged Partners to share the experiences with us directly and with specific details so that we can investigate accordingly.

In addition to the Questions and Answers listed in our communication on 2nd November, please make note of the following:

Q: How is traffic prioritised?
A: In order to deliver a better user experience of interactive services, less priority across the DSL network will be given to non-interactive bulk services at times of congestion. However, there are no specific or hard set trigger points at which prioritisation is applied and released on the protocols. Compared to other communications providers that set specific and permanent maximum limits on speed of access to non-interactive bulk services irrespective of bandwidth availability, Entanet’s measures are applied dynamically and traffic types are given the highest possible speed according to their priority versus bandwidth availability in real time.

Q: What traffic types are considered non-interactive bulk services?
A: Peer-to-peer and newsgroups protocols are given lower priority as non-interactive bulk services.

Q: When is traffic prioritised?
A: It is important to note that the prioritisation measures are applied dynamically. This means that, whilst the prioritisation rules run 24/7, varying levels of speed restriction are applied to non-interactive bulk services as the demand for interactive services bandwidth increases. Consequently there may be times when no traffic prioritisation is taking place and all services enjoy a higher speed.

Q: Is the Anti-Loss Tool (ALT) being used?
A: No, the ALT, including the high bandwidth user marking system, is not in operation because we are applying the traffic prioritisation measures.

Please note these measures continue to be monitored and are subject to change.

We have been monitoring traffic across the Entanet DSL network following increased feedback about poor connection speeds on our standard broadband packages, both in the charged and uncharged periods. We have identified that approximately 90% of customers are seeing their user experience adversely affected by approximately 10% of the customer base who are using their connection for non-interactive bulk services, for example news groups and peer-to-peer. Our aim is to deliver a positive user experience primarily for interactive services, for example web browsing, video streaming, VoIP and hosted application access.

In order to achieve this we will be applying techniques to give interactive services a higher priority across the DSL network. This action is in line with clause 2.5 of our Ts & Cs which refers to carrying out improvements, preventing overload of the network and preserving the integrity of the services. It is also being taken following consultation with several of our larger Partners. For the avoidance of doubt, we have so far not applied any protocol shaping or protocol prioritisation techniques.

Q: When will we start this?
A: We will begin prioritising traffic from 00:01hrs, Wednesday 4th November 2009.

Q: Is this action permanent?
A: We will monitor the impact of the traffic prioritisation for a period of approximately 3 weeks. If it is determined that this approach is successful then it is likely we will continue with this activity.

Q: Will we advise Partners of the impact of prioritising traffic in this way and whether it will become permanent?
A: Yes, aside from the overall improvement to service that we expect Partners and EUs to see as a result of our traffic prioritisation, we will seek feedback from Partners and confirm our decision on whether the measures will become permanent.

Q: How will it work?
A: Interactive services will be given higher priority across the DSL network. We do not intend or expect to impose any hard capping of specific protocols. This is unlike certain other communications providers who set permanent limits on specific protocols. Rather, we intend to dynamically manage traffic such that all services are available according to their priority. Therefore, varying levels of speed restriction will be applied to non-interactive bulk services where they adversely impact interactive services. It is not possible at this stage to identify specific parameters.

Q: Will Partners and EUs be able to see how traffic is being prioritised?
A: Not at the outset, although we will be considering if specific information can be provided in the future. In the meantime we will continue to provide details of our node utilisation on noc.enta.net.

Q: Will the existing Business and Family product prioritisation remain?
A: Yes, the current product attributes are unchanged.

Q: Will we advise Entanet-billed EUs directly?
A: No, as we are not introducing a product change we do not need to directly advise Entanet-billed EUs. However, this information is provided formally both on noc.enta.net and the e-billing site.

Q: What will happen to the Anti Loss Tool (ALT)?
A: During the period that we prioritise traffic as described above, the ALT (including the high bandwidth user marking system) will not be in operation.

Q: Will this affect EWCS customers?
A: No, our wholesale customers will not be impacted by this measure.

BT Openreach have declared a force majeure event in several areas. The text of the declaration is below:

We have today introduced a number of changes to these pages to enhance the functionality and ease of use of our noc pages.

These changes include:

• corrected RSS feed problems
• human friendly permalinks
• inline comments visible on front page
• style improvements

Whilst network changes are taking place as previously advised on this site, it has become necessary to temporarily remove access to the IPStream Connect status tool. This is because the tool could present incorrect information at times when necessary changes to bandwidth availability are being made on an IPSC interconnect by interconnect basis. We aim to restore access to the tool as soon as possible and apologise for any inconvenience that this restriction may cause you.

Further to our communication on this site (http://noc.enta.net/?p=381) we write to explain the reasons for reduced performance over the past few days and to apologise for any inconvenience caused to customers. Please be assured that we are working to ensure normal service levels are resumed as soon as possible.

In our previous communication we advised that, as users re-authenticate in order to connect to the IPSC interconnect serving them, we expect user volumes and bandwidth demand to vary among the interconnects and to be unrepresentative of performance expected when all moved users are connected to IPSC. It is therefore possible that variable performance may be experienced, especially during off-peak hours, until such time that normal volumes are established.

The process of moving connections from IPStream Centrals to IPSC is being executed without issue. However, at the point of executing the move, it is not possible for Entanet to know which IPSC interconnect individual users will be connected to, as this information is not available from BT Wholesale. Furthermore, as complete execution of the move relies on users re-authenticating to establish their connection on the new interconnect, it has not been possible for us to estimate the total bandwidth required to serve them. It is not feasible for us to set an artificially high bandwidth level, as doing so commits Entanet to that level with BTW for a minimum 3 month period. Bandwidth usage is therefore being continuously monitored and adjusted as users re-authenticate.

Whilst managing an increase in bandwidth availability per interconnect it has also been necessary for Entanet to operate the Anti-Loss Tool (ALT) on IPSC as it does on IPStream Centrals, in order that all users receive service. However, in this unsettled period, at times the application of the ALT has uncharacteristically affected performance to a greater degree. Further incremental increases in the bandwidth level proved to successfully suspend the ALT, albeit only temporarily.

These affects are currently being experienced mainly in off-peak hours only. During peak daytime hours the available bandwidth results in some minor activation of the ALT. Nonetheless we regret that off-peak users have experienced differing performance levels than they are accustomed to and will continue to actively manage bandwidth to a successful conclusion.

Further to the previous post, we have now moved the bulk of the remaining realms to IPStream Connect. Please be aware that end users need to re-authenticate in order to connect to the IPSC interconnect serving them (note that dropping the session for more than 30 seconds will be required to ensure the session is re-established on the new interconnects).  As users do so over the coming days, we expect user volumes and bandwidth demand to vary among the interconnects and to be unrepresentative of performance expected when all moved users are connected to IPSC. It is therefore possible that variable availability may be experienced, especially during off-peak hours, until such time that normal volumes are established. The ALT is operational on IPSC and will take effect whenever necessary as normal.

The interconnect graph will be adjusted to reflect the available bandwidth at each handoff (instead of a percentage of the maximum capable on the interconnect). Please note that, in the short term, the illustrated results are unlikely to be indicative of available capacity.

In order to address some of the issues seen recently with an imbalance on our Central Pipes, we are making improvements to the method of distributing users between LNS equipment. These changes will enable us to better balance the pipes through the day to day re-connections that take place, which in turn will lead to an improved customer experience for all users. One of the disadvantages of such as system however, is that it removes the ability for end users to be able to see which Central Pipe they are currently connected to, as specific LNS equipment cannot be tied to an individual central.

Whilst it has been a difficult decision to make, we have decided to make these changes as the loss of information would have happened on the switch to the upcoming IPStream Connect product, which is expected for Q1 of 2009. These changes have been introduced to a small number of devices and will continue to be rolled out in the coming days.

We have requested that a further 155 segment of DSL capacity be lit on Central 6, we expect this to be made live within the week. We are also happy to inform you that we will be introducing two new 622 Central Pipes ( 7 and 8 ) with an expected live date of the 01st October 2008.