menu-close
search-icon
banner

Category Archives: Capacity Management

Actionable predictive analytics: overcoming the analysis paralysis

Why standard forecasting analytics models fail to deliver in today’s world of complex digital networks and why telcos need a domain-specific analytics solution.

“Your analytical dashboards and visualizations look good, but I prefer actionable reports and insights”, said the deputy CEO of a Southeast Asia-based telecom service provider during one of our meetings last year. This was not just one odd instance. We have heard this many times in the past year from other CSP executives. There are many domain-agnostic AI/ML based analytics solution providers in the market, but what telcos really want is an analytical solution which provides end-to-end domain-specific actionable insights. Forecasting traffic or pointing out anomalies is one thing, but how to incorporate those recommendations into capacity planning? What is the root-cause for that anomaly so that it could be prevented in future? Instead of getting lost in analysis paralysis amidst thousands of fancy statistical metrics; a simpler, actionable and reliable predictive analytics solution is the need of the hour.

With the right mix of domain knowledge and analytics advantage, centered around the actual requirements of the network planners; Subex has come up with the concept of actionable predictive analytics. Network planners should be enabled for efficient, reliable and cost-effective capacity planning. Hence, here the focus is more on what matters to the telco network teams, i.e. the business values such as capex optimization, network performance improvement, customer experience enhancement and operational efficiency; rather than on underlying analytical components such as configured models or feature engineering.

Here are two of the most important aspects about Subex’s approach to predictive analytics which are different from the traditional forecasting models –

Multi-variate analysis

Unlike the traditional forecasting systems which predict the future trends for a metric based on the historical pattern of that given metric, in multivariate approach, the system understands the lagging or leading effect on the given KPI from other KPIs. With this, the telco can predict, in near real-time, what is going to happen in the future and adopt appropriate measures to prevent capacity issues. A multi-variate, self-learning forecasting model which runs on the in-house machine learning platform is complemented by domain-specific configurations and expertise which is equally essential for intelligent forecasting.

The figures below compare a multivariate model scenario that considers the lagging effect of KPI1 (e.g. customer complaints) on KPI2 (e.g. capacity utilization) with that of a traditional model that does not give such insights. In the first case, the operator does not get accurate results as yielded in the second case because there is a direct relation between traffic and customer complaints. For example, if there was an aberrant increase in traffic, the operator can take that fact into consideration for accurate prediction of future customer complaints.

One more use-case could be accurately predicting the time to capacity exhaust for a site if one of the neighboring sites is planned for decommissioning soon. In this case, with the help of geo-spatial analytics, the additional load on the given site due to decommissioning of the neighboring site would also be considered for calculating time to capacity exhaust.

capacity exhaust

Domain Specific Insights

Be it wireless or hybrid fiber-coaxial networks, even an accurate capacity forecast is incomplete without the required domain-specific insights. Without a proper root cause analysis for a network element exhausting soon (in terms of capacity), the network planners won’t be able to make the right decision about its proactive mitigation.

These are some questions to consider when developing your network augment action plan:

  • How many customers will be impacted when a given network element hits a capacity exhaustion threshold?
  • Will prioritizing the given candidate for capacity augment above other options result in the best customer experience improvement and maximized ROI?
  • What is the reason for this capacity exhaust? Is it because of seasonality, periodicity or cyclicity? Is it an anomaly due to some one-off event?
  • Will new Capex be required to address the capacity bottleneck, or are there alternatives to new spending?

Some of the insights that could be useful for the planners leveraging predictive analytics for capacity planning and management are shown below –

capacity planning

Predictive Analytics

Apart from the above two key differentiators, some other important aspects for a pragmatic, accurate and reliable predictive analytics solution are scalability and flexibility.

Multi-variate forecast models need to run thousands of simulations across the network to identify the correct correlated metrics for accurate predictions.  Such models need to be configurable, flexible and easy-to-understand for non-data scientists.

anshulbahtt

Anshul is an Associate Product Manager in the Network Analytics team at Subex. An IIT alumnus, Anshul has more than 4 years of work experience which includes working with two of the largest telecom operators around the world – Bharti Airtel Ltd. and Reliance Jio – across various markets within India. His stints at Airtel and Jio included Network Planning & Operations, Customer Experience Enhancement, Algorithms and Analytics for Network Performance Optimization & Automation. Anshul has two patents in the field of coverage & capacity optimization in LTE radio networks on his name.

Smart Performance Indicators — Redefining the KPIs

What is measured, is managed. Are you measuring the right metrics?

While driving, most of us would want to reach our destinations on time. An unexpected congestion on the way can easily turn a nice Friday evening into a frustrating start to the weekend. As the world progressed, it changed a lot the way we planned our journeys. Instead of going with the route that we took yesterday; we listened to the traffic updates on radio, kept ourselves updated with any special events around, based on the evening plans we tried to estimate our start time etc. Sounds better than driving with nothing but rear-view mirror & front view, but still too many things to monitor. Then came the Google Maps — one stop solution to get accurate real-time updates based on multiple data sources, scenario planning capability based on different available options & routes. You are still on the driving seat, but you have all the means to make the right decision. Friday evenings are no more frustrating.

This got me thinking – in our work, aren’t we too trying to reach our destinations — in terms of performance, goals & objectives — on-time every day, every quarter, every year? Having worked at two of the largest Telcos in the world, I thought that the performance management systems, in their current form today, need smarter metrics than just measuring what is easy. On their road to 5G and future technologies, Telcos would need something as smart as Google Maps to guide them to their destination.

With this idea and a small team of talented software engineers & data scientists, we have built — what we like to call it — a machine learning based Smart Performance Indicator framework. Using some of the multi-variate principal component analysis techniques, this framework creates meaningful Smart Performance Indicators out of the available metrics.

Telcos today are sitting on enormous amount of data and measuring thousands of KPIs but –

  • Are they measuring the right metrics?
  • Are they combining several metrics and analyzing their correlations & causations instead of just looking at some simple metrics?
  • They are surely measuring the past performance and comparing it with today’s, but are they looking ahead?
  • Instead of looking at the current values, are they looking at trends and statistics?
  • Are they measuring their KPI responses for an unexpected scenario?
  • Are they able to store enough amount of past historical data to be able to predict the future?
  • And most importantly, are the traditional KPIs learning from the changing trends, anomalies and proactively acting to it?

The traditional KPI dashboards today indicate more about how the network performance has been than about where it is going. The focus is more on short-term operational performance rather than long-term strategic performance. We manage what we measure, hence it has become more important than ever to measure the right metrics, smart metrics. Telcos need to measure what’s important, not just what is easy.

SPI-outcome

Sample dashboards with SPI outcomes

If you are working with a Telco and think that the above problems need to be addressed in your organization, we exactly know what you need. Your Fridays should no more be frustrating, feel free to reach out to me at anshul.bhatt@subex.com

anshulbahtt

Anshul is an Associate Product Manager in the Network Analytics team at Subex. An IIT alumnus, Anshul has more than 4 years of work experience which includes working with two of the largest telecom operators around the world – Bharti Airtel Ltd. and Reliance Jio – across various markets within India. His stints at Airtel and Jio included Network Planning & Operations, Customer Experience Enhancement, Algorithms and Analytics for Network Performance Optimization & Automation. Anshul has two patents in the field of coverage & capacity optimization in LTE radio networks on his name.

Telcos CAPEX is here to stay. Optimizing it is the way to go.

“A lot of people have been talking about how capex is going to come down with SDN and I’ve said, ‘No, it’s going to stay the same for Verizon’  – Fran Shammo, CFO, Verizon. May 2016

This comment right from the head honcho of one of the largest Telcos in US cannot be taken lightly. Despite lot of talk in the industry about SDN / NFV CAPEX reduction benefits, we’re seeing skeptical questions around smooth transition to virtualization. But I will keep SDN /NFV discussion for some other day. Let us focus on the topic – CAPEX spends. Verizon’s CFO has confirmed its CAPEX spend going to stay, despite network virtualization!

The CAPEX focus could be different for different Telcos. For some Telcos like Verizon, their CAPEX spend mainly focused on future technologies, leading the market, greater customer experience etc. For some other Telcos, their budget constraints force them to think hard and do delicate trade-off between strategical “revenue-growth” projects and tactical maintenance projects to keep up with network growth, retain customers, improve quality of experience etc. With this hard balancing at hand, what if Telcos are equipped with smart tools & methodologies that could help in optimizing their on-going CAPEX? But, is such thing exist? I will get there in a moment. Please bear with me.

First, let us go through few industry trends.  In our recent study from Gartner, we got few interesting insights.

Here is short summary on the insights:

  • On an average Telco spends 15% – 20% of annual revenue on yearly Capital Expenditure
  • Increase in Capex spends w.r.t revenue (CAPEX Intensity = Capex / Revenue) is not translating into equivalent increase in revenue growth
  • Correlation between CAPEX Intensity and Revenue growth is a weak factor for Telcos mentioned in the regions. This means revenue growth is not linearly correlated with CAPEX spends
  • 5-year flat growth in revenues across geographies is not encouraging. Max 20% 5-year top-line growth in North America region and deep negative for Europe region (-11%)
  • Cost of capital over last decade is higher than RoI on an average across the industry
  • Notable positive point is the margins are maintained in 25 – 35% range across geographies. And it is imperative to maintain this margins to generate free cash to fund next CAPEX cycle but if not completely.

The above stats where CAPEX spends are not reaping substantial revenue growth indicates two major viewpoints:

  1. Strategic capital investments have a slightly long gestation period but not comparable to capital cycles of traditional industries like manufacturing industry.
  2. Bulk of CAPEX go into maintenance projects. That is, to keep-up with current network demand juggernaut, customer retention, quality of service etc.

For instance, a good chunk of leading Tier 1 North American operator’s CAPEX goes into wireless network for densification and getting future ready for 5G deployments. This could be a case for many big Telcos – investing on future technologies. On the contrary, we have also seen majority of Telcos’ CAPEX going in for second type of investment – meeting current network data growth. This is nothing wrong as such and very much required to keep customers happy.

However, if one looks at this fact in light of recent market research findings from one of the big four audit firms, it gives a different perspective. The research reveals that majority of the Telcos not equipped with enough tools or industry best practices to assess the CAPEX spends on projects, evaluate ROI for each such investment and perform sustainable capital allocation. This is a surprising revelation. It simply means that many Telcos are servicing on-going CAPEX without rigorous assessment on actual RoI vs planned RoI, are not taking forward lessons learned from previous CAPEX cycle. Even the Telcos who do have rigorous processes, right incentive structure, accountability of results etc. actually misses a critical point.

What Telcos underestimate?

The critical point is – generally the assets, especially the network assets are viewed from monetary value perspective only in this whole CAPEX scheme of things. The value that can be derived from un-lit or under-used network asset capacities for the CAPEX planned is not given deserved thought or action. This is because of the fundamental reason that financial and network data of assets are lying in silos. This data is never used together to gather useful insights to put the network assets to sweat to furthest possible aligning with ever growing network demand and broader strategic CAPEX – RoI goals. Telcos can do more with their data. It would require collaborative efforts with right partner to unleash the power residing underneath the siloed systems.

Sai Thilak

Sai Thilak has 11+ years of experience in engineering management, product development, solution architecting and customer deliveries in Telecom OSS/BSS space. Sai currently leads product management for Network Asset Assurance & Data Integrity Management products. He’s passionate about new age products, a voracious reader, biography & history buff, financial market student, cricketer & active blogger.

The legacy network is dead, long live the legacy network!

I was recently asked by a Tier 1 client in North America if Subex can use our network discovery technology to retrieve information from D4 channel banks.  I honestly had not crossed paths with these relics of the voice network since my days doing central office engineering in the late 80’s.  So you may be wondering, in a day-and-age when the buzz is about SDN/NFV, IoT and everything in the “Cloud”, why is someone worried about the humble channel bank?

As it turns out, there’s plenty to worry about.  End-of-life technology can be a significant Opex drain.  Operators incur costs for energy (power, HVAC), maintenance and real estate to keep such equipment in place.  Compounding the problem is that much of this old equipment typically sits racked, stacked and powered… and idle (no traffic).   Channel banks are just one example.  Arguably, the entire fixed-line TDM network is retirement-age (I’m talking about SONET/SDH DACs and ADMs, voice switches, local loop equipment, etc.) and needs to yield to IP/MPLS and VoIP.

In a previous blog post, I wrote about what operators need to consider when planning a transformation from legacy technologies to future state.   For this post, I will stay grounded in the present and focus on this question: What strategies can operators employ to reduce their Opex and Capex burdens when operating a legacy network?   For starters:

  • Use network discovery techniques to determine the operational status of your actively deployed network assets.
  • For all unutilized assets, apply a deliberate strategy to disposition everything.  Too often, because operators don’t have adequate visibility to operational status and utilization of assets in the legacy network, they default to what I call a “rust-in-place” strategy.  Since they lack the visibility, they ignore the problem.  Equipment sits idle or underutilized and costs add up.  My suggestion is to proceed with purpose—if an asset is carrying adequate revenue-producing traffic, fine.  If not, do something about it!

For assets with reuse potential, then the options include:

  • Harvest and reuse elsewhere in the network.  Benefit: Avoid Capex for new purchases.
  • Perform grooms to more densely pack some assets and free up others for reuse or end-of-life monetization.
  • Allocate as spare.  Benefits: Reduced maintenance costs when spares are optimized in terms of count and location.   Customer experience is improved and exposure to SLA penalties is reduced when spares are well managed.

If there is no reuse potential, then consider:

  • Reselling on the secondary market if there is still industry demand for the asset.
  • If not, then recover and sell for salvage value.
  • In both cases, remember that the NPV of averted monthly energy and real estate costs may actually exceed any direct cash received when the asset is sold or salvaged.
  • Don’t overlook other possible financial benefits from disposing unneeded assets such as tax write-offs and reduced insurance premiums.

Most importantly, seek an asset management and logistics partner who can help you squeeze the most value from legacy assets.  Elements of a legacy network cost reduction program include:

  • Automated audits via network discovery
  • Asset evaluation and disposition recommendations
  • Capacity utilization trending and related analytics
  • Asset tracking
  • Turnkey asset recovery services
  • Resale valuation and brokerage services
  • Eco-friendly recycling
  • Analytics for sparing level optimization
  • Spares management
  • Warehousing and related logistics services
  • Warranty and annual maintenance contract management
  • Test, repair and engineering services

Subex provides industry-leading asset management solutions and services.  With our forward and reverse logistics partners around the globe, we can help you to establish a turnkey and highly effective legacy network cost reduction program.

ANDREW-JACOBS

Director of Business Development for Network Analytics
Andy has 20+ years of experience in engineering management, business operations and IT, primarily with Tier 1 operators including Level 3, MCI and GTE. His responsibilities included leading IT development teams that built mission-critical network management, provisioning and inventory systems with thousands of users. Prior to joining Subex, Andy was a Senior Manager overseeing a Data Governance organization at a major Internet Services provider. Andy graduated from the University of Pennsylvania with degrees in Electrical Engineering and Economics (Wharton). He holds an MBA from the University of Colorado.

Making the Connection

risk of congestion

It’s 7AM and I can’t put off getting up any longer, so I look out the window and see there’s a light frost on the grass, which the weather channel warned me about 3 days ago. An hour later I’m at the train station waiting for the 7:42 which is delayed because the frost has caused the points to seize up in a town 50 miles away, so now the entire South East rail system is completely snarled up. They predicted the weather and probably knew there would be a point’s failure, but still the network crashed. So I need to phone work to let them know I’m going to be late, but I can’t connect. Thousands of other commuters around me are also trying to phone ahead, but the network can’t handle it. It seems to happen every day.

We are surrounded by events which are beyond our control, but often they happen in predictable ways. The points failure was perhaps less predictable than my alarm, but we always knew that when the temperature dropped there would be some kind of failure somewhere that would lead to cancellations and a breakdown of the network. We always knew that rush hour would become an agonising crawl into town on overcrowded trains. The congestion could probably have been avoided if they could have predicted which parts of the network were under the most stress and the impact on the network in the event of failure or congestion at those stress points. Additional resources could then be provided at those points, or alternative routes planned to bypass the congestion and limit the ripple out effect, like a fire break.  The problem is only likely to get worse, and the network more unreliable, as the population increases and more people than ever rely on the rail network to get to work.

With the arrival of LTE and rapidly increasing popularity of Video on Demand then telecoms networks are also facing increasing levels of congestion and instability. Global data traffic is predicted to increase by 10 to 20 times by 2019 (Cisco).   In order to meet regulatory obligations and maintain customer experience Capex is set to spiral upwards. MNOs, who are already facing a year on year decrease in ARPU, will struggle to keep pace with demand and the risk of congestion will be ever present.

As with rail networks MNOs need a longer term strategy in place to understand where and when future choke points in the network will occur so that the risk of congestion can be eliminated for the least cost. Subex Capacity Management provides the capability to predict these points of congestion by monitoring and correlating metrics from across the network to provide detailed forecasts of network utilisation. Additional factors can be brought into the forecasts, such as the impact of major events or the rolling out of M2M services and different scenarios played out to understand how the network will respond. By automating the forecasting process network managers can be alerted long before issues become critical and congestion begins to occur. They can evaluate different options for either re-homing traffic or augmenting the network for the least possible cost. Stranded or un-utilised assets can even be recovered and re-located to satisfy demand for very little cost.

CFOs need to find ways to keep increasing revenue while controlling costs, and CTOs need to keep network delivering ever greater speeds as volumes of traffic increase exponentially.  Both need to look into the future to avoid a future of network instability, falling quality, crippling network costs and lost revenue.

Mark Jenkins

Mark Jenkins has worked in the IT industry for over 15 years as a BI and Analytics consultant, and more recently as ROC Product Manager for Subex Ltd. He has designed and deployed solutions for global companies in many sectors including Insurance, utilities and telecommunications. Mark holds a BSc Hons in Computer Science from Manchester University (UK).

A Thorn in the Side? – Handling Over-the-top Content Demands

As I interact with more and more service providers about their network capacity issues, I’ve become sure about one thing – what worked before, isn’t really working anymore.  The CapEx requirement for network equipment just to keep up with the exponential growth in data traffic (i.e., Data Tsunami) is still not getting them ahead of significant congestion issues and customer impacting events.  Why? Traditional capacity management paradigms are not working.

Essentially, feedback from carriers of all sizes and types has exposed one of the most significant shifts in thinking regarding how to go about managing and planning for network capacity.  They know that the rules are all changing and today’s content demands are outpacing the CSPs ability to keep pace.  The first key question is how to get back in front of the capacity demand (we’ll talk about monetization next…stay tuned).  So, why aren’t today’s processes scaling?

  • CSPs use a multitude of human resources and manual processes to manage network capacity.   This may have scaled under slower and more predictable capacity growth curves, but thanks to services like You-Tube & Netflix, entire network capacity is shifting in quantum leaps.
  • Solutions provided by equipment vendors are often platform specific, and reinforce a silo approach to Capacity Management when a holistic view is needed.  Service demand congestion is a network phenomenon which doesn’t care about individual equipment vendors or devices.
  • CSP planning groups leverage data and make decisions based on systems which have 20 – 40% inaccuracy in comparison to the actual capacity availability in the network.
  • Today’s CSP solution approach is often homegrown where 90% of the time is spent on acquiring and understanding raw data.  As a whole, everyone is trying to answer the question of how to proactively eliminate the possibility of congestion, but most are still focused on addressing the symptoms and not preventing the problem

It is surprising to note that even top tier/technology leaders cannot accurately predict where and when capacity issues will impact their networks.  This lack of visibility hurts CSPs considerably because as per our own studies, network events are behind can account for up to 50% of customer churn in high value mobile data services.

And the Capacity Management problem doesn’t really end there; in many ways it’s like a supply chain process. Marketing owns the function of forecasting where service uptake will drive capacity needs across the network. When Marketing underestimates service uptake, there is a real and significant impact to potential revenue: On average, it can take about 3 months from when capacity is fully tapped in a Central Office (CO) to when new capacity can be added to your network.  During that time, customers expecting service availability become hugely frustrated and begin to churn.  Engineering groups are pushed into panic-mode, trying to react as fast as possible – often putting capacity in the wrong places due to inaccurate data – resulting in further congestion, service degradation, an inefficient use of capital.

The message from CXO’s is crystal clear – there is an urgent and dire need to find new ways of monetizing the data crossing their networks. This need is exacerbated with OTT content and net-neutrality. SLA and authentication based revenue models are absolutely dependent on knowing what types of content/services are traversing your network, how much capacity they consume, and how utilization is driven by your consumer’s interests and activities.  This type of analysis requires a critical and intelligent binding of network and services data with business data to truly assess the financial impact to the CSP. Many Business Intelligence (BI) solution leaders will lay claim to abilities here, but actually fall very short of the mark.  Instead, real experience suggests that solutions in the marketplace today either:

  • Can handle the financial aspects of your business but have no understanding of today’s network dynamics in terms of capacity issues and services;
  • Can handle parts of your network very deeply, but do not correlate or provide a holistic view at the service level; or,
  • Can collect some network and service level information, but have no ability to incorporate business data to understand the impact to the business – i.e,. cost, subscriber behavior, propensities

All the above challenges bring us to the inevitable question – what kind of approach does one take in order to tackle capacity management issues? How does one stop chasing traffic and focus on flattening the CapEx curve instead? In order to attain ‘Capacity Management Nirvana‘, a proactive and scalable approach needs to be adopted by CSPs. An approach which not only intelligently binds network and business strategies based on the Data Tsunami realities but also brings proactive and predictive capacity management to the table. At the end of the day, a CSP should have access to all their capacity, the ability to leverage real and immediate feedback on the change in capacity as service uptake increases, and finally, the right tools and intelligence to get in front of what’s coming.

To know more about how a Capacity Management solution can help you address the above issues, download the whitepaper “Energizing Smart Growth with Network Intelligence

Angeline MacIvor started her career at Nortel Networks in the optical domain, and gradually migrating over to the world of network software, with successful tenures at MetaSolv (now Oracle), Syndesis and now Subex Limited. Angeline has been working with key operators in Canada in the US for 17 years, often winning awards for solution innovation in response to customer needs. Angeline is a key driver behind Subex’s capacity management program.

Get Started with Subex