Recent research conducted by Quocirca and published in a new report, ‘Masters of Machines: Business insight from IT operational intelligence’ (commissioned by Splunk to find out how businesses in various sectors are harnessing the power of operational intelligence) shows that the average volume of daily commercial transactions driven by a European enterprise’s IT systems is over 40,000. For a telco this figure almost triples to 110,000. Such transactions may be automated or customer initiated and can include account creations, device activations, call routing, records and so on. This is a big number to be impacted if the supporting IT systems fail or underperform in some way. Not surprisingly, the more transactive a business is, the more it is reliant on IT with telcos being some of the most IT-reliant.
In many cases, supporting such high transaction rates requires flexible IT infrastructure. The more transactive a business is, the more likely it is to have changed the way that infrastructure is provisioned to serve the requirement. Highly transactive organisations are more likely to turn to virtualisation to enable the flexible provisioning of the customer workloads; they are also more likely to use cloud-based services as a primary or secondary resource to support such applications.
However provisioned, maintaining the performance and availability of the IT systems requires insight in to what is and has been happening. Historic insight allows systems to be retuned in anticipation of future requirements, whilst real time insight can address issues as they occur. Such insight relies on access to operational intelligence, which, in itself, relies on data collected across an organisation’s IT infrastructure. This is so-called machine data and its use, alongside data collected about customer activity, is the key to linking the performance of IT systems to commercial activity and success.
The ability to gather and act upon operational intelligence varies widely, but the bottom line is that more transactive organisations recognise the need more than their less transactive counterparts. Telcos, more than any other sector, recognise this with 90% saying quality access to machine data is essential to deliver operational intelligence. This does not mean other organisations cannot benefit; in fact their inability to do so may be one of the reasons they are less transactive. Furthermore, most organisations are likely to become more reliant on IT to drive more transactions over time and they need to be prepared for that and realise the long term benefits to be gained.
Of course, one of the most obvious benefits is be able to better provision and tune IT systems to support customers. However, the most advanced users of operational intelligence go well beyond this and use it to provide real business insight beyond the IT department to other parts of the business and, in some cases, on to business partners. For many, the intelligence is provided direct to executive management; over 60% already do this to some extent, although many of these would like to improve their capability to do so and 22% would like the ability to do so in the first place.
Examples of the sort of insight that can be provided include:
- A call centre can monitor actual call volumes and/or waiting times and see if these correlate with other data, such as customer type or geographic location. Early warning of arising issues is thus possible.
- Recognition of username/password attempts on a system that do not match normal usage patterns may identify a potential breach.
- Attempted systems updates can be checked against configuration change requests; if there is no match a change may be unauthorised and can be blocked.
- Key performance indicators (KPIs) can be monitored; for example, are response times for customers transacting on web sites acceptable? In what circumstances are targets not being met and at what times of day or in which regions?
The challenges should not be underestimated. Machine data needs to be gathered from a wide range of devices and on-demand services that support any given organisation’s IT infrastructure. Processing all this data to provide real time insight is a great example of a big data problem. Big data has been an overused term in the IT industry of recent, so let’s put this in context by looking at the five Vs often used to characterise it:
- Volume – each commercial transaction driven by an organisation’s IT infrastructure will touch multiple parts of the supporting infrastructure generating log records at every stage. If each transaction generates tens of such records, based on the average of 40,000 transactions a day, which likely lead to an average 10 million plus items of machine data a day. For telcos you may as well triple this; the annual figure will be in hundreds of billions. This is truly big data volumes!
- Variety – the data comes from a range of different systems; servers, storage systems, routers, on-demand services, end user devices, application and database logs etc. To turn this in to true operational intelligence requires correlating all this with data about the users themselves, their device, browser activity, experience, and their transaction records.
- Velocity – to respond to issues as they occur means processing this data in real time, often providing historic context to current events though cross correlation.
- Veracity – in-depth insight into customer activity will give true and timely information about their actual experience and how IT issues are impacting business performance. Only with such an accurate view can an IT department be responsive to business requirements and deliver expected service levels.
- Value – hopefully, the value of being able to do this goes without saying; the evidence can be seen through Quocirca’s research as the remainder of this article will show. However, that value is only delivered if the supporting technology and tools for processing huge volumes of machine data and turn in to true operational intelligence are available.
The top challenge identified in the research was the ability to collect, search and analyse the volume of data involved (i.e. volume), followed by the integrating data from multiple sources (i.e. variety). To gauge how well organisations were disposed to gather and use operational intelligence, Quocirca developed an operational intelligence index (OI Index). This was based on an organisation’s ability to achieve the following:
- Search and investigate (machine data)
- Proactive monitoring (of service levels)
- Operational visibility (of customer experience)
- Real-time business insights
The report lists more details, but each of these had three sub-categories (so 12 overall) and, depending on an organisation’s ability in each area, an OI index score of between 0 and 3 was assigned to each organisation, where a score of 3 would rate as excellent and 0 would be non-existent. Telcos led the way with an average score of 2.23, compared to an overall average of 1.92.
The correlation of OI index with aspiration in other areas was very strong. These organisations were gathering machine data from 65% of their infrastructure compared to 45% of those with a median OI score and 15% of those with the lowest score. Over 80% of those with a maximum OI Index of 3 were able to provide a comprehensive insight of business performance at a board level, for those with a median score it was 48% and just 3% for those with a low score.
For Splunk, who commissioned the research, one of the most interesting findings was the sort of tools organisations had in place for gathering and analysing machine data to provide operational intelligence. 64% were using traditional business intelligence tools, whilst 57% were using spreadsheets (hardly tools for processing billions of machine data record of varying formats). Only 27% were using custom designed tools, such as those provided by Splunk, which provides a software platform for real-time operational Intelligence.
There could be two reasons for this; first it could be that such tools do not deliver, but this is not what the research showed. Those using purpose built tools were gathering machine data from more of their infrastructure (52%) then those using other methods, for example just 44% with business tools and 43% with spreadsheets. For many, these will just reflect their early achievements with purpose built operational intelligence tools as the market is still maturing (compared to the long history of business intelligence and spreadsheet use).
So, the message from Quocirca’s new research is clear—machine data is a key resource for providing operational intelligence and understanding business performance. However, achieving the goal of providing such intelligence across the business in real time requires tools that are up to the job and only those that with the highest operational intelligence capability are achieving this—and it is telcos that are leading the way.
This article was first published in Global Telecoms Business