u-ETL
For the many typical data transformations required in the online advertising arena, IPONWEB offers data aggregation as a managed service product called u-ETL.
u-ETL is a managed hadoop cluster used for raw data transformations of our Clients’ data streams from systems such as RightMedia, DoubleClick, Invite Media and our Clients’ own u-Server instances. It is a framework and integration environment that ties all of our products together, essential for processing the TB of data necessary for successful system operations.
The u-ETL service packages:
- Retrieval of source data from feed repositories
- Construction and ongoing execution of required data transformations
- Provides transformation output availability and data rotation
- Monitors and alarms for critical events in data pipeline management
- Issue management for disruptions to the data pipeline
With u-ETL, large volumes of data can be processed in short periods of time through managed parallelism of a large Hadoop cluster. This allows for meeting tight feedback loop requirements, often in the order of 1 hour or less. The u-ETL service provides a number of key data loop outputs essential to other systems in our eco-system.
- A u-Predict loop input
- A budget loop for ad decisioning
- Reporting loops for dashboard roll-ups
- Large data sets for u-Slicer & BI analysis
By using u-ETL, our clients leverage the operational efficiencies and cost savings gained from shared, hosted cloud infrastructure and our experience in monitoring and direct issue management with the data feed suppliers.
An unfortunate fact of life is that data feeds break very frequently. When a data feed provider has an issue it typically affects many of our clients, meaning on behalf of our clients we are able to deal with this much more quickly and efficiently than standalone solutions.
