The growth of the Internet of Things (IoT) coupled with operators' move to SDN and NFV is driving the need for advanced analytics that can make sense of a tsunami of big data.
Unstructured big data coming from video, sensors, machine-to-machine (M2M) interactions, text, social media and more is a relatively new phenomenon for network operators to tackle. According to Heavy Reading analyst Ari Banerjee, who follows the telco analytics market closely, this influx of data requires new techniques like MapReduce, Hadoop and real-time streaming technology to begin to make sense of it all. (See Defining Big Data & Telco Analytics for Heavy Reading's definition of big data.)
"You can see the real-time use cases of those when you look at the financial industry," he explains. "NASDAQ uses these technologies because one nanosecond here and there makes a difference between a stock buy and hold. Latency is critical. We are bringing those technologies into the telecom domain."
The key players
The market for data analytics is clearly still in the early stages, but key players are already emerging -- and they look a lot like the key players in other areas of telecom. Banerjee classifies the leaders as those that really understand the role of analytics in telecom, plus a few others. That means your traditional strongholds like IBM Corp. (NYSE: IBM), Teradata (NYSE: TDC) and SAS Institute Inc. , fast-emerging companies like Oracle Corp. (Nasdaq: ORCL) and Amdocs Ltd. (NYSE: DOX), as well as equipment vendors like Ericsson AB (Nasdaq: ERIC), Cisco Systems Inc. (Nasdaq: CSCO) and even Huawei Technologies Co. Ltd. (See Amdocs Wants to Be Big in Carrier Big Data and Allot Goes Deep in Big-Data Analytics Game.)
There is a reason for the traditional telecom vendors' interest in analytics: They already own most of the data the telcos need. These guys provide the revenue management systems or have equipment entrenched in the network. The source of the data is their systems, so why not do the processing and provide more business value? It's as lucrative an opportunity for them as it is for their telco customers. Banerjee says this is an opportunity they've tackled with partners in the past, but they are now investing to tackle it on their own. (See Telco Big Data Market to Thrive.)
At the same time, because of the aforementioned lucrative opportunity, the space is also attracting a number of startups, such as Splunk Inc. , Guavus Inc. and Vitria Technology Inc. (Nasdaq: VITR). The majority of these startups are basing their wares on Google (Nasdaq: GOOG)'s Hadoop, which is open source and much cheaper than running a data warehouse. Hadoop is also changing how operators are approaching the market. (See Big Data Attracts Big Dollars, New Faces, Guavus Colors in Its Big Data Picture , Zoomdata Raises $17M to Beautify Big Data and Apigee Acquisition Brings Analytics to APIs.)
"Operators used to be selective in data they are storing," Banerjee says. "Now with the hardware becoming cheaper, they are trying to incorporate a lot of that data volume, which has increased."
Analyzing IoT, SDN & NFV
The progress of the Internet of Things is a big reason why data has gone from terabytes to zettabytes in recent years, a growth trajectory that will likely continue as operators deploy SDN and NFV as well. Banerjee tells us that there are a lot more measures that need to be taken in a virtual network functions environment, especially when it comes to the OSS layer for service management, quality of service and assurance functions. Virtual network functions require a platform to measure them, track their performance and manage them. (See Data Analytics in a Virtual World and NFV Just Made OSS Hot.)
IoT requires all that as well; however, with M2M connections, the type of data being pushed out is still a bit of an unknown because it's coming from all kinds of different devices in many different formats that all has to be put together to be analyzed. That's why a number of companies are promising to make sense of IoT with analytics platforms built especially for it. (See Cisco Paints IoT Into the Big Data Picture and Cisco Puts a Fog Over IoT.)
"There is so much information that needs to be understood and analyzed and triggers need to happen based on certain actions," Banerjee says. Take, for example, the freight shipping industry. If fish is being shipped across the country and it goes bad, data analytics can determine who pays based on whether it was kept at the temperature indicated or other issues that might have arisen. Or, an increasingly common example, insurance companies have to measure many different elements that all must be kept straight to determine a driver's costs. (See Could Data Be the New 'Currency'?)
Operators are looking to provide the connectivity here, but also the value-added services based on what they can do with that data.
"The whole IoT space and analytics goes together and it goes into how do you use data for different reasons from creating applications to more value added services," Banerjee says. "You cannot do that in the blind without specific actionable data."
Operators are trying to do this and take their involvement in IoT to the next level, but it's still early days. Banerjee says that right now M2M is only 2% to 3% of operators' overall revenues. They want to get that number to 5% to 10% over the next three to five years, which will be necessary for them to justify investing more in analytics for the space. (See Big Data Saves T-Mobile Big Bucks and Telefónica Battles Big Data Hype.)
It's certainly the direction operators are moving -- towards IoT, SDN and NFV with analytics laced throughout -- but challenges do, of course, remain. Banerjee points out that while it's hardly news, it remains true that any advanced analytics strategy requires three to four months of data cleaning first, throwing out any garbage data to get to the quality stuff. He said that the sheer volume of data is challenging as is getting the right level of information. Privacy issues and regulation remain tricky as well. (See That Big Data Sinking Feeling and The Big Data Challenge: 10 Tips for Telcos.)
"Next year, I bet the challenges will be the same," the analyst says. "The same thing comes up over and over again -- the data quality problem and data integration problem can never be solved. We have so much data coming in in different, unstructured formats that putting it all together to make it actionable is a big effort. We have to find a common format and take out redundancies. That's a big mess."
— Sarah Reedy, Senior Editor, Light Reading