During 2020, Chorus and service providers worked with MBIE and SamKnows to investigate and improve the End Customer experience of Fibre Max services. One outcome from this investigation was the publication of the Broadband Performance appendix to Bitstream service descriptions, that provides key learnings and guidance to Service Providers on how to get the best performance from Chorus Bitstream solutions.  

The initial investigation primarily focused on downstream traffic. However, as part of the introduction of new 300/300 and 500/500 services and our Big Fibre Boost initiative, Chorus has undertaken further investigations on upstream traffic and these learnings, as well as feedback from Service Providers, were added to the Broadband Performance Appendix in December 2021.  

What are the changes?  

The joint performance investigation identified that a key element in maximising the customer experience is aligning the border policies, i.e., ensuring that the Customer Premises or Service Provider Network egress policies are aligned with the Chorus Bitstream ingress Traffic Contracts. In particular, these egress policies should shape traffic to align it with the Chorus ingress policers, as documented in the Bitstream Performance Specification.  

This alignment is more challenging upstream, however, as while active queue management on the CPE will improve performance, most low-cost End Customer CPE have limited upstream shaping capability. This can result in occasional bursty traffic that overloads the Chorus ingress traffic contracts, resulting in frame loss. Testing showed that this can, under some circumstances, reduce practical throughput to ~ 80% headline speed.  

Joint testing during the Big Fibre Boost demonstrated that reducing the GPON Dynamic Bandwidth Allocation (DBA) bandwidth closer to service information rate can compensate for End Customer CPE having limited upstream shaping capability.  

At the time, these modifications were limited to the 300Mbps upstream services as further investigation was required to determine the optimum settings for 500Mbps, and updating live services was more complicated. This investigation has now been completed, and we are looking to implement these changes in our network as soon as possible.  

How does reducing GPON DBA bandwidth improve performance?  

As all ONTs on the same PON share the point-to-multipoint fibre access, GPON uses a DBA mechanism to allocate upstream timeslots to each ONT, according to their subscribed bandwidth. Frames are stored in the ONT buffer, and transmitted upstream according to their subscribed bandwidth, resulting in similar behaviour characteristics to a shaper.  

When UFB was originally implemented, Chorus found that this mechanism applies a serialisation delay effect on upstream frames and we deliberately ‘overclocked’ the DBA to compensate. However, serialisation delay is much less of an issue for higher speed services. 

 By reducing the GPON DBA bandwidth to close to the service information rate on these higher speed services, the GPON DBA acts as a pseudo upstream shaper, smoothing bursty traffic that might otherwise exceed the per-service ingress policer at the I-NNI.  

What impact will these changes have on your customers?  

This change will be implemented by:  

• Updating the Chorus 500Mbps GPON DBA policy;  
• Using a script to apply this to all Chorus OLTs.  
• The policy will dynamically update all Customers using 500Mbps upstream, including Small Business Fibre and Max offers.  

There will be no customer outage and there is no requirement for you to make changes to either your CPE or network.  

Testing has shown that this change will not impact other services. Customers’ throughput will be improved under bursty conditions, and will not be impacted under non-bursty conditions.  

Contact 

If you have any questions, please reach out to your account team.