«Delivered to | Northern Communications Information Systems Working Group c/o Government of Yukon Delivered by | Nordicity Date | Foreword The Project ...»
Reliability is a measure of how often the network fails while service availability is a measure of the total time the services delivered over the network are available. Reliability is impacted by factors that are beyond the direct control of the service providers (e.g., a satellite failure), and factors that can theoretically be controlled by the service providers. Service availability is directly related to network reliability. The time it takes to restore service, either on the main connection or on a backup connection greatly impacts availability. Availability of the backbone network is severely impacted by an absence of adequate redundancy.
No specifc standard for reliability is set in this study, as this service standard is more properly set in light of operational experience – post-engineering and post- build/implementation of improved connectivity. Reliability is also a function of the technologies utilized in the network, which in the case of the three Territories means that the reliability of one part of the network or community may vary a great deal. As such, no single standard would be appropriate to set, nor could it be measured in this study.
5.3 Redundancy Redundancy provides greater reliability of network operation and access given risks of network disruption and outages. There are three major means to achieve redundancy in a backbone
Redundant hardware/software for the electronics;
Diverse paths; and, Physical diversity.
The highest level of service availability and reliability can only be achieved if all of the above are implemented following sound network engineering practices. Achieving this level of redundancy throughout a network with the geography and construction challenges of the Territories would be a costly endeavour, but would provide the highest levels of service and future capabilities to the citizens of the Territories.
In the South, network providers achieve redundancy on backbone networks via redundant ring architecture. These networks are designed with the objective of customer transparency to network outages. Customers do not expect to be forced to curtail any normal operations as a result of a backbone network failure (electronic equipment or network path). Well-designed backbone networks are able to reroute all trafc carried on the failed component of the backbone network. This level of redundancy is a laudable goal in the long-term planning horizon of the Territories.
The target standard for redundancy within the Territories is that every point-of-presence on the backhaul network should have a redundant connection to the backbone network. Ideally, the redundant connection should be capable of supporting all services carried over the primary backhaul connection. The associated costs of such redundancy varies greatly on the technologies being employed at each individual point-of-presence, with some low-population sites appearing to be cost-prohibitive in the near term to provide full redundancy.
The analysis of the current shortfall in bandwidth for both the main and the redundant backhaul links shows a major shortfall in the bandwidth capacity of the redundant backhaul links. The prime reason for this is the diferent delivery platforms used for the redundant link, - which in many cases in the Northwest Territories and Nunavut is satellite. Nunavut in particular, is exposed in many ways to the lack of redundancy, as all of its communities are served only by satellite, with no redundancy provided at all. In that way, not only are communities isolated and lacking a robust road infrastructure, so too are the telecommunications networks in Nunavut. As a result, this lack of redundancy becomes even more obvious and pressing in certain communities.
In addition, the fact that there is a separate backhaul link at a POP does not necessarily imply that this link can be used as a redundant link for the services carried over the prime link. The bandwidth available on the redundant link is used for existing services and cannot be used to carry trafc assigned to the prime link without turning down or throttling the trafc for existing services on the redundant link.
Until such time as sufcient spare capacity is available on the alternate backhaul connection, careful planning and confguration of trafc routing will be needed to ensure that high priority services are given a priority status for transit over the available bandwidth. This also applies to services carried over the “redundant” connection, as some of these may be of sufcient priority to remain on the connection leaving less bandwidth for services on the failed primary connection.
Detailed trafc engineering would be required to identify the trafc fows to assign priority and to program data switches to route the appropriate trafc types to the redundant route(s) automatically. In reality, class of service assignment and routing will most likely be applied, and only the highest class of service trafc associated with the most critical applications will be routed automatically, assuming that the available bandwidth on the redundant backhaul connection is less than the bandwidth of the applications on the prime connection.
5.4 Minimum Target for Redundant Bandwidth 50% of the projected bandwidth is generally considered to be the minimum redundant bandwidth required for the redundant connection, based on considerations of network topologies, redundancies, trafc growth, etc. As a general planning guideline, if the technology selected for a redundant connection is the same technology as the prime connection, the bandwidth capacity of the redundant connection should at least match that of the prime connection. In the case of selection of fbre optics for a redundant link, it should be provisioned with a minimum bandwidth of 10 Gbps.
For the purposes of this study, where no redundant link existed for a given community, it was assumed that in the event of failure of the prime link, only the high priority trafc (peak-period trafc generated by the Health, Justice/Public safety and Social Services categories) would be rerouted over the redundant links. This was in recognition that the predominant redundancy technology was satellite.
In the optimization model, the following defnitions and calculations were used to examine the
existing bandwidth in the three Territories:
Main Link: the current highest backbone link available in each community (ACIA report data as baseline);
Redundant Link: the current second highest backbone link available in each community (ACIA report data as baseline);
Average trafc: the average trafc estimated and modeled in this study per community;
High priority trafc: the trafc generated by the sector of health and Public safety/ Justice at peak period (as modeled in this study); and,
For the main link, it is the diference between the average trafc and the main link capacity.
For the redundant link, it is the diference between the high priority trafc and the redundant link capacity.
Figure depicts the overall percentage of communities that are currently below the recommended standard for connectivity based on the above-outlined assumptions. For full details of these shortages, please refer to Appendix 2.
Figure : Percentage of communities below the recommended bandwidth service standards
60% 45% 38% 40% 15% 20% 6% 0% Percentage of communities with bandwidth shortage on the main Link Percentage of communities with bandwidth shortage on the redundant Link In summary, redundancy calculations were based on the assumption that 100% of the projected bandwidth used for critical applications: health, Justice/ Public safety and security would be covered.
5.5 Service Quality Quality of Service (‘QoS’) is determined by a number of parameters that can afect the capability of delivering data in the manner required to provide an acceptable level of utility and user experience for the applications transported over the IP connection. In practice, the applications are categorized into groups of services/applications that have common characteristic
requirements for network transmission. Major groupings are:
Streaming media (IP TV, video – one-way, audio over IP);
IP telephony (VoIP);
Videoconference (2-way, ‘real time’);
Circuit Emulation Services (TDM over IP/Ethernet);
Industrial control systems; and, Online gaming.
Each of the above applications is afected to one degree or another by available bandwidth, latency (delay), jitter, and lost or dropped packets.
5.6 Suggested minimum standards
SUGGESTED MINIMUM SERVICE STANDARDS
THIS OVERALL STARTING POINT SERVICE STANDARD WAS 9 MBPS DOWNLOAD AND 1.5 MBPS
UPLOADBased on the analysis above, the following suggested minimum standards for bandwidth and
service quality were devised:
Bandwidth – diferentiated according to the characteristics (population, demand of user categories, simultaneous usage, platform(s), etc.) of the sector entity. The overall starting point service standard was 9 Mbps download and 1.5 Mbps upload. To be clear, this is the service package that would be subscribed to, NOT a guaranteed 24hr/day throughput guarantee. However, the physical infrastructure would generally support those speeds as needed. The derivation of these service standards is explored further in this chapter, as well as in Appendix 2.
Latency (two-way) – As diferent technologies exhibit great variation in latency, no specifed latency standard. While fbre optic sites exhibit latency of 10ms, satellite sites exhibit latencies of several hundred milliseconds.
Jitter (Packet Delay Variation) – 0.5ms average, not to exceed 10ms maximum jitter more than 0.1% of time. This is an industry-accepted standard taking into consideration the applications in use.
Lost or dropped packets - 0.1%. This is also an accepted industry standard.
6. DYNAMIC OPTIMIZATION MODEL AND
FINANCIAL COSTING ANALYSISA dynamic optimization model was developed based on a series of relevant inputs and parameters, each set out in a series of inter-linked spreadsheets, in order to provide the fexibility to model various scenarios and to update data as it may change in the future.
The model uses the following inputs:
$2200/Mbps/month for satellite usage (10-year lease);
Costing of the Microwave and Fibre optics networks (details in Appendix 2);
Population per community;
Population/Category/recommended minimum bandwidth for sector-based applications; and, Available current networks requiring upgrade to meet the current and future bandwidth requirements.
6.1 Dynamic Optimization Model The Dynamic Optimization Model integrates the diferent data modules– described below into a rigorous framework for the assessment of alternative solutions for the delivery of the proposed service standards.
The modules are the following:
Backbone identifcation module: This module identifes the backhaul/backbone technologies connecting each community to deliver the bandwidth by considering
the following parameters:
The distances between connected communities;
Line of site distance for the microwave links; and, The route distance for the fbre links.
Population distribution module:This module identifes the diferent categories of users in each community, and provides the necessary inputs to the Trafc Estimation Module to generate the forecast trafc for each community per sector of activity.
Current backbone bandwidth module: This module identifes the current available backhaul and backbone solutions and their trafc-carrying capacity. This will also provide the necessary information to characterize the main link and the secondary link that could be used to connect to the Internet. This module also provides the possible redundancy path that could be used in the case of network failure. This module also provides the inputs to the Backbone Dimensioning Module and the Trafc Estimation Module to assess on any shortage of bandwidth that may occurs on each link in carrying the trafc estimation fgures proposed. Since each community is diferent, this module compares the diferent options available (Fibre, Microwave or Satellite) and chooses the best one for each backhaul/backbone component.
Trafc estimation module: This module provides the trafc predictions for each community based on the previous inputs provided by the Population Distribution Module, using a series of technical parameters. Some of these parameters can be changed in order to refne the results based on the initial assumptions or statistical trafc behaviour.
Benchmarking module:This module is developed to benchmark costs in other jurisdictions. The three available technologies (Microwave, Fibre and Satellite) were considered in order to provide the most cost-efective price to deliver the bandwidth needed according to the current network infrastructure. The benchmarked jurisdictions were chosen to match the community characteristics in the Territories.
Alaska is considered a good match for the fnancial inputs (the other jurisdictions do not provide sufciently-detailed construction costs). The Alaska benchmarking research provided a cost estimate per End User/Mile with a 50% contingency for the Microwave and Fibre Optic construction costs. For the Satellite costs, the model uses the $2200/Mbps/Month as an estimate 5 As more up-to-data and/or specifc industry data becomes available, the Dynamic Optimization Model will enable dynamic modeling of cost impacts using diferent assumptions from other jurisdictions.
Costing estimation module: This module uses the following modules as inputs: (i) the Benchmarking Module; (ii) the Trafc Estimation Module; and (iii) the Backbone Identifcation Module. For each community, this module calculates the costs of upgrading the current technology in order to handle the estimated trafc generated by the services speed identifed as a minimum service standard.