Agenda and Minutes

Purpose of Meeting: HSCN Technical & Security Workstream – Peering discussion

Date: 19/05/2016Time: 10.00-12:00

Location: Thursday 19th May, Ambassador’s Hotel, Bloomsbury, London & Webex

Agenda / Owner
Review of HSCN connectivity options / KG
Chair / Organisation / Department
Kate Gill (KG) / Chair, HSCN Solution Architect
Michael Bowyer (MB) / Chair, Innopsis
Terry Brown / HSCN Technical Architect
Various industry representatives
Simon Foster / GDS PSN

Minutes

Obligations Framework scene setting

New version of Obligations Framework released yesterday. Michael Bowyer (MB) urged attendees to digest it properly. New comments and any revisions will be made to this new version.

MB urged industry to proceed at real pace in regards to the review of the Obligations Framework, and the suggestion of new ideas to be incorporated into it. Need to fit in with the HSCN timescales.

MB encouraged open and free participation from all session attendees. Everyone in attendance was a technical expert, and could therefore enjoy this forum of engagement on that level.

Action: MB to send out the new version of the Obligations Framework to all registered attendees today, and to put the current version of the Framework on the Innopsis website.

MB explained that the Authority is working to very tight deadlines, and therefore, the deliverables had to be realistic and achievable.

Kate Gill (KG) explained that the objective of the Technical and Security working group was to get to a mutually agreed technical solution within the tight timescales.

Industry warned that the Obligations Framework needed to stay in synch with the design documentation. KG confirmed that both would be issued simultaneously during the next iteration.

KG said today’s workshop would concentrate on the technical topics of the working group and that another workshop scheduled on 2 June 2016 would focus on the security elements.

Peering / network interconnects discussion

KG explained that the HSCN requirements had been established in order to reach a disaggregated network outcome and that no major infrastructure changes were wanted. An easy path to transition was required and all of this was needed by the end of this calendar year.

Industry asked what was meant by “easy?”

KG explained that “easy” meant hitting the target time and being technically viable.

MB stated that the aim was to have an open market for all vendors.

Industry asked if this was a push and pull scenario; the push is this engagement process but the pull will be driven by NHS funding. The new funding model will not change until April next year. There is therefore a gap between this point and when the solution is required (end of calendar year). Commercial working group discussion.

KG explained that the funding model was being worked on in terms of when funding would change to new model; but there would be a possibility of doing some business before April 2017 as early adopter projects and that HSCN would work to enable these including the funding aspects.

KG explained there was a need to deal with the network traffic and that there was also a need to remove a large backbone, as was currently the case. Too much ‘local’ traffic was currently going through the core (e.g. GP surgery talking to a hospital down the road). Also, a lot of existing traffic went externally to BBC, YouTube, etc. This was all a part of the solution requirement.

Terry Brown (TB) said there was now a shift in focus from the backbone traffic to more regional network.

Industry asked if there would be a change to the off-shoring policy currently in place.

KG said the current Department of Health (DH) position was one of devolved responsibility, i.e. Trusts, GPs, etc. decide what they want to do, with guidance. At the top level of government, they do make arrangements with big telecommunications organisations. Under NPfIT, off-shoring policy was mandated, as the contract was held centrally. This was a DH interpretation of the law (Data Protection Act). This question was about providing services with the maximum capability. KG noted that there was nothing to stop industry from selling services to customers outside of England. NHS Scotland did not use as in England, although they did have access to these services, but do use their own different services. They are different networks from a security viewpoint and so still need to be connected via a gateway.

Industry asked if there was a lot of regional traffic then why was HSCN not allowing access to the network access layer.

MB encouraged the group to return to this question later on during the workshop.

Industry asked if any other approach for interconnections had been considered, for example, cloud connecting.

KG answered that that this is what the programme was working towards; HSCN would provide options but not force people down a single route. HSCN did have a preferred approach, but it was not prescriptive.

An additional complexity identified was IP addressing – TB confirmed that there were about 350,000 NHS IP addresses, which may or may not be sufficient for this initiative. Another was CoINs on long contracts that wanted to move as a group. This is to be considered in the DLN vs AC discussion. Both issues were parked for another day.

Industry asked if HSCN would fund own peering exchanges to make this work.

TB responded that the existing telecommunication facilities would be ideal; much more convenient but that it was being considered if it was required to hit timescales.

MB asked industry to consider whether TB proposal would be good or bad for them, and to also consider if there is an alternative.

KG explained that the peering route was the more favourable route in order to avoid creating a new centralised core network. Peering is easier than having a core. It provides a rapid path to connectivity. It provides a level playing field for all suppliers. It simplifies internet connectivity for everyone, and makes cloud interface easier too. It also makes it easier to connect to other external networks. This is why HSCN favours peering, from a technical perspective. It is up to industry to debate and decide if they agree. However, there were several challenges to peering: we do not know the regional distribution of traffic, and there is lots of it – estimated 500GB of traffic by TB.

Industry asked if this figure would scare people off, and whether it was wholly accurate – much of this total traffic amount will stay local and will not actually traverse the full network – this should be reflected in the figures quoted. Better regionalisation is also a challenge. Therefore, it was better not to design a peering node to carry over 500GB of traffic, as this would not be required. What is required is more likely to be around 30GB or less.

KG reiterated that a network to with the correct capacity was to be designed and built.

Industry stated that there was both a technical side and an operational side to the peering option. HSCN should engage with an internet exchange provider, to understand how to govern a peering network. Peering with new DLNs is also a consideration, as this would be more complex if it was centrally managed. Commercial working group discussion.

KG confirmed that it was not yet clear whether the HSCN programme would pay in the event of one-point integration.

Industry commented that this would speed things up, but that end-user transparency must also be considered. HSCN should drive uptake by being a central supplier, so that customers have one point of contact. It was important to consider the end-user in all decisions and for HSCN to drive these discussions.

MB responded that this approach would actually create a barrier to market for vendors. It was important to set aside capacity concerns and to firstly focus on establishing the service/solution. There was a separate Service Management workstream set up to discuss Service Management obligations.

Industry commented that getting rid of DLNs and going with peering was a good idea because there would be no need for different types of providers.

MB stated that industry must record all concerns and ideas relating to this process in written format, so they can all be considered.

Industry asked why separate DLNs were needed.

KG responded that it was important to discuss and agree DLNs and APs and to consider the technical, security and service management implications.

KG welcomed feedback from all on the current version of the Obligations Framework in this respect. HSCN are happy to consider options for network providers being of one type and all connected directly to the proposed interconnection mechanism. SME providers needed to consider of this would be commercially viable when reviewing options.

KG briefly went through the PowerPoint presentation slides and explained that the connection architecture would determine the ease with which the chosen approach could be used, as well as learning of its effectiveness.

Industry commented that peering would happen as a matter of course, as it is the current evolutionary trajectory, allowing DN-SP to peer with each other. There was a need to differentiate between the access providers and others. It was also asked what bandwidth would be coming in.

TB responded that in healthcare there was no mandate to have any specific security assurances.

KG commented that HSCN was happy to use the shared infrastructure but some aspects that were to be considered were commercial and some were technical. Use and enhance existing network to do what is needed but what must not happen is that it ends up just being the current GCN. Need to establish and ascertain if PSN is suitable in the first place. PSN could provide a cost saving too, by only routing local traffic onto the national network when required, and otherwise keeping it local. PSN ‘Lite’ could be provided by any existing PSN providers or by none of them.

HSCN is not a data controller, as per NPfIT, they just assure the solution. SME present in the room liked the idea of using an existing network however are not currently PSN suppliers and so would need a mechanism for

MB reiterated that the PSN model was being proposed and asked the group for further alternative options.

Industry said that one option was to turn the NHS transition network in to a peering network. Gateways are already in place and this de-risks delivery; it can be stood up quickly. Can utilize BT exchanges . This approach would also deal with any management infrastructure challenges but concerned whether this would deal with the ‘level playing field’ issue.

Industry commented on the national peering network exchange. The Authority stands up infrastructure and everyone hangs off that. Why is it attractive to a vendor? Cheap interconnect allows peering to be set up easily.

KG responded that it was important to decide if it did matter if you could not peer to everyone. This approach may not work for smaller vendors – how easily can they get to peering points? Also the question of resilience, but this should be an issue between the vendor and customer only; the cost of service will be ranged accordingly. This would work for regional CoINs that do not need to talk to London, for example.

Industry asked how quickly could PSN be stood up and that this could then become another option.

MB said everybody had to be realistic about estimated timescales for each of these options when submitting the written proposals.

Actions:

  • PSN Lite paper to be written by Keith Smith
  • Transition Peering Network paper to be written by Peter Negus
  • Third party HSCIC owned paper to be written by Marco Dellarole

KG said that all papers needed to satisfy the requirements slide presented during the workshop and that the papers should be between two and three pages in length. Should debate/discuss Obligations Framework requirements. Each approach should be designed to make HSCN attractive to as wide a community as possible – accessibility and scalability are big concerns – must keep the costs down. Proposals can include commercial questions and comments. Must consider broader customer base when writing proposals. All three documents must be completed and submitted by COP Wednesday 25th May 2016.

Action: KG to clearly frame the key criteria, in order to avoid any perceived ambiguity from industry.

Action: KG confirmed a separate face-to-face session was required to properly debate the submitted proposals and that this would be in a further arranged workshop session. KG to setup

Action: TB/KG to prepare two-pager document on IP addressing and to share this with working group members.

ACTION SUMMARY

Item / Description / Owner / Updates / Due / Status
1 / MB to send out the new version of the Obligations Framework to all registered attendees today, and to put the current version of the Framework on the Innopsis website / MB / Done
2 / MB agreed to manage the production of papers for the optins for network interconnects between suppliers.
MB to forward contact details for submission of proposals.
Three papers to be written by COP Wed 25/05 as follows:
PSN Lite paper to be written by Keith Smith
Transition Peering Network paper to be written by Peter Negus
Third party peering paper to be written byMarcoDellarole / MB
3 / KG to clearly frame the key criteria, in order to avoid any perceived ambiguity from industry. / KG / Done
3 / KG confirmed a separate face-to-face session was required to properly debate the submitted proposals and that this would be in a further arranged workshop session. KG to setup / KG / Being arranged for 8th June
4 / TB/KG to prepare two-pager document on IP addressing and to share this with working group members. / TB

1