Finding the needle in a big data haystack

Predicting and preventing network pain points – Part I

In a recent blog post we highlighted the five ways in which network downtime can be significantly reduced in order to improve customer service and brand loyalty. In this post we aim to address the first point, ensuring visibility, and provide some guidance on how this can be achieved.

As many companies in the telco sector will know, big data is nothing new; the volumes of data available have always been extensive, and for regulatory reasons some classes of data have to be stored for many years. However, frustratingly for many in the industry, the big data hype is drowning out the potential deliverables.

A common misconception is that ‘big data’ equals ‘unstructured data’ or ‘useless data’, which unfortunately, for some is true due to a lack of strategy or even a business case for using the data. Purely storing data will not make it useful, but that in itself is no justification for discarding it.

To make data useful there must be realisable benefits to the business and an effective business strategy to take steps to ensure visibility of that data to the relevant business groups. However this is where there is a cost implication and problems can start; if the wrong tools are used for the job the costs can be considerable and the task can very quickly become just like looking for a very expensive ‘needle in a haystack’!

A real ‘needle in a haystack’ case 

The pain: A mobile operator suffered a significant increase in the number of dropped calls on their network, equal to 2% of total calls. The operations department was using legacy fault management reporting and didn’t know what the problem was until the following day, when a retrospective report was pulled.

The diagnosis: A fault with the equipment at one of their major cell sites, which could have been picked up in minutes instead of hours or days. Unfortunately, the customer care department’s data wasn’t integrated with the legacy fault management software, meaning that the cause and effect couldn’t be identified and resolved in real-time.

The remedy: Ensure your operations provide a vertical (top down) view, so you can simultaneously view network and customer problems, and prioritise accordingly.

For a preventive cure that can collect and analyse raw terabytes of network and service platform performance data; in real-time, contact us.

5 ways to predict network pain points

Predicting and preventing network nightmares

Despite a solid performance from network operators over the London 2012 Olympics, last year saw a few incidents of network outages, which caused considerable chaos and fall out among the mobile consumers. In some instances, it took up to 24 hours to identify the problem and rectify the situation.

If you consider that some network outages can affect all customers in a particular geographic region and in some cases the entire country, this is no insignificant amount of damage to brand, not accounting for loss of trust and revenue, and brand damage among both consumer and corporate clients.

In a competitive market, a real-time and top down view of customer and network operations is absolutely critical to avoiding these hiccups. There should be no reason for a multiple hour delay to identify problems, if the following five areas are covered.

1. Ensure visibility – can your operations provide a vertical (top down) view, so you can simultaneously view network and customer problems, and prioritise accordingly? In many operations, the customer complaint arises, with still no way of linking this back to network performance, creating an all too common ‘needle in a haystack’ situation.

2. Report as one – the horizontal (multiple vendor and domain) view, your management reports need to be able to track events and data across a number of systems, companies, affiliations and centres, efficiently. You cannot afford to extract data from multiple sources and present this as a disparate view to management. You need some way to automate a single, horizontal view of reporting.

3. Deploy predictive analytics – any analytical system can benefit from the application of streaming analytics. Communications companies can analyse call data records in real time, among a multitude of other information, to offer better service and rates.

4. Optimise the infrastructure – you should be able to exchange and interface between network operators, receiving accurate and dependable performance reports of shared infrastructures. Third party systems, which provide a consistent and honest view of multiple operators, are imperative.

5. Respond rapidly to change – are you currently restricted in your ability to quickly and cost effectively enhance operational systems to manage new service offerings to meet the very demanding needs of your customers? The need to implement a solution that evolves with your business, and to engage with a flexible and reliable partner with a good track record of success, is essential when working with the complex process of unification and migration to the next generation of systems and services.

If you can tick the box with regard to the above five areas, your customer service and network operations should run seamlessly, hand-in-hand, and you should be sleeping well at night. However, as incidents have shown, this clearly isn’t always the case.

We will drill down further into each of these areas over the forthcoming weeks. In the meantime, if you have any questions you want answered, or would like to share your network nightmares, please comment below.

Is a Streaming Analytics Correlation Engine Powering Your Big Data?

Hype or no hype, big data is not a new phenomenon and can provide businesses incredible insight. That is, if businesses know what to look for, how to use the data and have the right tools to hand. In this … Continue reading →

Big Data Never Sleeps

– Takeaways from the TM Forum Big Data Analytics Summit SysMech Sales Director, Terry Harding, reflects on his attendance at the recent TM Forum Big Data Analytics Summit in Amsterdam and the key points emerging. Big data is in itself … Continue reading →