Network Service SoftwareSoftware for Network Services
Network Service Provider Perspective on Network Slicing
With the ability to create fully functioning sliced network infrastructural volumes (either physically, virtually or a mix of both), network service vendors will be able to meet the great challenges of the upcoming 5G services: How can different types of service with very different demands on the same infrastucture be supported and operated?
The blending of network based mixed network based features such as Improved Mobile Broadband (eMBB), Solid Machine-Type Communications (mMTC) and Ultra-Reliable and Low Latency Communications (uRLLC) makes it quite challenging to establish a shared infrastructure that is able to keep the needs of each of them in an orderly and structure. It' much more comfortable to think about the separation on specialised partition types that are conceived and optimised for the kind of service to be provided.
There is a second layer of segmentation that is directly related to the network operators' commercial model, considering how (and how many) these layers are to be provided. From this point of view, network slice can be seen as the next way of delivering wholesaling service. This means that it is linked to the assignment of certain ressources per Verticals Client or Lessee.
Each level of separation goes in the sense of assigning either natural or artificial assets that are tailored to the type of service and appear to be per lessee committed, ensuring separation between the assets allotted. Considering this from a provider's perspective and taking into consideration the similarity with today's wholesalers' offers, different scenarios can be viewed here that lead to slice with different skills, especially with regard to their administration and controlling options, and how much of it the network service providers hand over to the slave tenant:
Built-in slots, which are the volumes used for the provider's own service, while maintaining full controls and administration of these slots. Externally sliced volumes, i.e. volumes that host client service and appear to the client as separate network. Providers administered slots, i.e. the ISP retains full controlling and administration of the slots.
This means that the client can only use the network resource of the provided slip without further administration or monitoring. Lessee-administered windows, which means that the client has full command over the assigned ressources and features. Lessee has full or partial accessibility to a range of surgeries and/or configurations, and Lessee disconnects only the necessary equipment.
What is remarkable here is that we use the word "infrastructure" in the widest possible meaning, i.e. inclusive of all network features that the client can use and use on the disk. It' clear that tenant-guided windows have to be assigned per client, as the client checks them directly.
At the same time, provider-managed slices can accept different clients using the same service in relation to service requests (bandwidth, delay, number of processed meetings, etc.). Fig. 1 shows these different types of slices in a graphic form, showing that although the ability to orchestrate is always within the service providers, either the service providers or the tenants have full mastery over the slip or the service on it, according to the type of slip.
Today, network operators use cloud-like devices (so-called NFV infrastructures: NFVI) that are used to provide virtualised network features (VNFs). NFVIs enable the provision of NFVs that can be dynamically adjusted to the needs of each service called for. UNICA's comprehensive infrastructure is built on clouds that enable large-scale deployments across several locations, but also meets the requirements of vendors such as carriers' degree of availability, scaleability and operability.
In addition to the flexibility of network virtualisation, there is a need for flexibility in controlling the flow of transport according to the service behaviour foreseen. To meet these demands, network programming (enabled by Software Definition Networking, SDN) and its implementation with the virtualisation feature described above are seen as the basis for the convergence with Network Slicing[2,3], which plays a key technology roll in the controlling, administration and operations of telecommunications markets in the near term.
SDN and NFV will be phased into current grids, creating a number of issues that will accompany this technology change. Network-licing will bring an extra level of sophistication. One of the basic aspects of network slice is that each slice behaves like a real, autonomous network. It is this key demand of sliced insulation that every network operator must meet in order to be able to provide such a service to its clients.
Insulation is crucial for the supplier's capability to access the various above mentioned categories of slice and how they can be required and used by its clients. From the supplier's point of view (extension), this results in a provisional challenge schedule (without taking a specific order into account):
slip mode: Every slip must act as a separate network and share the basic hardware and software assets. It is necessary to define control and service mechanism to display and abstracte the right information for each slave-client. In this context, there is a higher level of complexities involved in using outside windows, since the information to be displayed and the restrictions on accessibility must be specified or even arranged between the supplier and tenant.
To scale network assets, it must be clearly defined to what extend a particular slip can or cannot serve a customer. Applying different types of slicing lifecycle assessment (SLAs) to the possibilities of managing, controlling and customizing the slicing will have a direct effect on scale.
One is the size of the mere resources allocated and invoiced (including those required to meet the requirements for security and availability), which is directly related to the type of service required and agreed with the tenant. However, an overly fine-grained supply of sliced products can cause an incalculable number of inspection and administration artefacts to be managed by the supplier, making it impractical.
The scaleability can be strongly influenced by the number of available outside panes administered by the tenants. To solve disputes and secure agreed service level agreements, a number of conciliation procedures are needed to enable effective use of the available natural resource and prevent over-dimensioning. Not only do these mechanism have to exist between the different Slices on the same infra structure, but also within the single layers themselves (like the three major categories described in the above categorization), since the relationship between client and layer is not necessarily 1:1 (as in the case of the externally provider-managed slices).
Not only must the arbitral jurisdiction be used in the process of creating or activating clients, but also (and more importantly) when scale- and/or error incidents occur, so that assets are correctly (reallocated) according to the current SLA. Verticals clients may not need restrictions in relation to cover, serviceability, limitations on ressources, geographic expansion, etc.
This avoids any possible restriction of the network operator with whom he maintains a business relation as the main supplier. As a result, multi-domain slying must be enabled, which means that the general application must normalize functionality and commerce for it. Cutting design and dimensioning: When sliced, transport resources and targets become much less foreseeable, if at all.
In particular in the event that the ultimate decisions on the use of resources and targets (as well as some interim features that could change or amend the transport profile) are in the tenant's responsibility. Appropriate design, sizing and implementation is necessary to make the shift to this new type of service lasting, beginning with adequate information gathering on the use of resources, which includes the most important cyberspace.
The use of slice-based applications requires not only communications between different administration areas but, more difficultly, the incorporation of different areas of engineering. All end-to-end character of slice includes different computer environment and transportation techniques that depend on an open and consequent orchestra methodology. In addition, the lifecycle scenario for third-party slice switching requires switching mechanisms in order to co-ordinate enquiries for the different customer-side ( "tenant-managed") and provider-side ("provider-managed") platform managers.
Interacting with our clients is the cornerstone of our ability to understand the needs of the service we provide. To provide a consequent service portfolios and their integrations into the in-house network administration and orchestrations, suitable abstraction and template have to be developed. Abstraction at all levels is required to provide the various slicing categories described above.
Safety is a crucial component of any common network to ensure correct functioning and, above all, an appropriate proportion of resource for each one. In the event of a malfunction, users of our products must be properly identified, their privileges must be respected through authorisation procedures, and the actions they perform must be taken into account so that further audits can be carried out.
In addition, action must be taken to actively identify and target proactive breaches of public safety to prevent a violation of safety in one area from spreading to the infra-structure and/or other areas. Windscreen aging: The same dynamics and versatility in the assignment of the panes to the tenant also apply to the period during which these panes will be in operation.
It can be assumed that a mixture of long-lived and short-lived discs will coexist on the same structure. As a result, vendors will be affected in several ways, from capacity scheduling (slice-and-drag forecasting inclusive of volume and consumable resources) to safety (data retention per tenant). NETWORKSLING is a highly prospective network service delivery solution that meets the exacting 5G needs and supports new wholesalers.
Whilst using new but well-established technological backbone, such as NFV and SDN, seems an easy way to accomplish network slice, there are some important hurdles that need to be tackled in order to make it viable for network service provision. Whilst a full listing of outstanding questions related to operative network slice is still the object of research, the most important of these are presented here, which relate to the basic isolating characteristic.
4 ] L.M. Contreras, P. Doolan, H. Lønsethagen, D.R.López, "Operation, Organisation and Commercial Challenge for Network Operators under SDN and NFV", Elsevier Computer Network, vol. 92, p. 211-217, 2015. He has been part of the Telefónica I+D / Telefónica Global CTIO Department since August 2011 and deals with SDN, virtualisation, transportation grids and their interactions with clouds and shared network infrastructure as well as interconnect issues.
He is an IETF ( "Author of two RFCs"), ETSI, ITU-T and ONF employee in the field of standardisation, where he organised and co-ordinated the first proof-of-concept for the use of SDN on mobile transportation network in October 2015. Dr. Diego R. Lopez has been Senior Technology Expert for Network Media and Service at Telefonica I+D since 2011.
Prior to Telefónica, he worked for several years in academia, researching network service abstraction and developing related AIs. Currently, Diego focuses on the identification and evaluation of new possibilities in network infrastructure technology and the co-ordination of domestic and multinational co-operation work.
He is currently interested in network virtualisation, network service, network managment, new network architecture and network safety. Mr. diego leads the ETSI ISG on Network Function Virtualisation and the NFVRG within the IRTF. It is driving forward research and innovations in the area of network managment for a uniform cluster and carrier network infra-structure. During his doctoral thesis, Catalin worked as a research fellow at CERN, Geneva, Switzerland, on network technology for the large hadron collider large Hadron experimental system.
In the European research project ESPRIT SWIFT, FP6 NoAH, FP7 4WARD and Unitfy he worked on network measurement, self-testing, self-management and DevOps.