Database Reference
In-Depth Information
The IBM Accelerator for
Telecommunications Event Data Analytics
The telecommunications (telco) industry has been an early and active adopter
of the IBM Big Data platform, especially in dealing with data in motion. This is
no surprise, given the data management challenges that telco companies face,
especially in the area of high-velocity data. Based on experiences in providing
solutions for telco businesses, IBM has built the Telco Data Accelerator (TEDA
for short, but officially known as the IBM Accelerator for Telecommunications
Event Data Analytics), which focuses on in-motion processing of call detail
records (CDR). The TEDA shipped with Streams, which we talked about in
Chapter 6.
CDR processing is an acute pain point for telco providers today, mainly
because of high volumes and high velocities. Every time someone places a
call, telco switches create a CDR, which includes information such as the call-
ing number, called number, time stamp for the start of the call, call type
(voice or SMS), duration of the call, quality information, and more. With the
advent of cell phones, the number of CDR logs has greatly increased
because additional CDRs are created for the same mobile call
every
time the
call is transferred to a different cell tower. For telco providers with many
customers, it's very difficult to process these CDR logs proactively, especially
during peak usage periods. To give an example of the scale of these volumes,
one of our large Asian telco clients has to deal with about six billion CDRs
per day, where the daily peak rate is over 200,000 CDRs per second!
Most telco providers load batches of CDR data into a relational database,
and then perform CDR processing, including transformation and analytics,
used for applications like revenue assurance or fraud detection in real-time.
There are also post-processing activities, such as removing duplicate entries.
This means there are delays of many hours before the telco business can track
billable activity or other issues.
The TEDA is designed to push much of the CDR analytics and post-
processing activity out of the at-rest persistence layer and get the work done
in-motion, using Streams, just after the CDR is generated. This approach not
only provides the telco provider more immediate access to billing information,
but a wealth of additional information can be harvested by performing
in-motion analysis of the CDR data.
