Tata Teleservices adopts a tiered storage approach to trim infrastructure cost by 40% and leverage data assets for business growth
By Heena Jhingan
Sitting on about one billion call data records (CDRs) per day—amounting to about 8TB of storage per month—Tata Teleservices (TTSL), the parent company of Tata Docomo, was grappling with controlling cost and labor involved in managing CDR data.
The telecom service provider, which offers voice and data services in retail and enterprise markets, was storing and maintaining multiple copies of data in the past as mandated by the Telecom Regulatory Authority of India (TRAI) to retain CDRs for at least 12 months. The data retention was not meant just for TRAI compliance, but also to cater to the needs of its various business teams.
Stacking and maintaining multiple copies of this data resulted in a sprawling storage footprint, that would in turn consume a lot of power. This approach was not only expensive, but also made the data vulnerable to security threats.
CIO Ashish Pachory looked at this challenge as an opportunity to consolidate data storage, to cut down infrastructure cost and for more effective data lifecycle management. He decided to consolidate data from all locations at their Hyderabad data center and added a low-cost yet efficient storage layer at the middle level.
Classify and store
“Earlier we were using extensive enterprise class storage solution across, so each time the storage needs spiked it would mean expenditure, Pachory explains.
He adds, “Even data extraction from storage tapes was a tedious task. A single request of CDR retrieval would be a time consuming process, despite having outsourced the process which was done using robotic arms. We realized that the time taken, effort, labor and cost involved to do this every month was immense.”
Pachory thinks they had a couple of challenges before them. First, was to define what was important information and the second was to create an architecture to store information in an optimal way.
“We decided to consolidate storage, and add a middle tier to the storage system that would store of six months data to high enterprise storage solution and the rest on the tape,” he says.
Pachory elaborates that they consolidated the CDR data with Solix Information Lifecycle Management framework and HP storage hardware. The tiered storage approach along with storage virtualization and a BI layer ensured that data was stored, archived, and retrieved efficiently to help teams extract relevant information.
The service provider needed to define policies for retention, archival and movement of data across tiers. The data needed to be classified based on the usage pattern and retention policy and for this they used Oracle’s database solution.
The project was undertaken in partnership with Tata Consultancy Services. It took them six months to complete the implementation right from concept, selection of vendors, to hardware acquisitions and roll out.
“The approach reduced our storage size and till now since the deployment, the cost reduction over the period has been to the tune of Rs 2.5 crore, which in itself is an achievement,” says Pachory.
Challenges
The project did see some initial challenges, but everything fell in place with time. Pachory believes that people-perceived performance challenges always erupt from changes in technology and processes.
“Besides, there are several databases across locations; the challenge for us was defining data integrity and establishing relation between the data,” he says.
Leveraging data
Even though the need for storage optimization was primarily driven by the TRAI mandate and cost control, the CIO has a road map in mind to take this implementation to the next level by monetizing the data assets that they are maintaining.
Pachory has been experimenting with tools like IBM Cognos that will be instrumental in last mile or location level analysis. “We are already doing a PoC with Hadoop and have plans to build this in a big data framework,” he informs.
Performing analysis on the data repository will help TTSL plan its strategies better. For example, access to data for churn and customer analysis will allow the marketing teams to have a better view of the customers and launch targeted campaigns accordingly.
To maintain user transparency between online, near real-time and historical data, Pachory says, storage level virtualization has been built. This allows business teams to extract relevant data without worrying on where and how data is stored, something like this was not possible earlier.
The implementation has enabled access of processed call data records to legal and regulatory teams to meet their requirements. Going forward, Pachory feels by providing very granular analysis and actionable information at a location, sales teams of the company will able to drive up sales. “We anticipate a 20% increase in revenue,” he concludes.
