Data Management - Signature
Automating the load of millions of images into rapid-search database.
Some of our success stories.
Learn how we met the client challenge - quality solutions, on time, on budget, to specification, and exceeding expectations - with some of our case studies.
- Data Management - Microsoft DTS to Informatica
- Data Management - Signature
- Systems Integration - USA Bank Referrals
- Project Rescue - RM Builder
- Software Development - Vicor Lockbox
- Business Process Analysis
- IT Staff Assessment
- Data Management - ARM Reports
- Quality Assurance - BYO Checking
- Data Management - Microsoft Access Upgrade
Read more
The Challenge
Our client, a large regional bank operating in the Southeast, had acquired another bank and was
in the process of converting that bank's systems into their corporate core systems.
The acquired bank's database of digital images of Signature cards, Corporate Resolutions,
and CD/IRAs needed to be scrubbed and loaded into our client's Signature Card system. The legacy image loader was single-threaded, incapable of supporting new image types, used proprietary image formats, and failed to detect corrupted or incorrectly loaded images. There was concern that the legacy loaded would not be able to load all the required images by Conversion Day.
Our Approach
Our first step was to migrate legacy code to the .NET 3.5 architecture, allowing us to introduce multi-threaded processing to speed up the throughput. We redesigned the loader and database structures to support new image types without additional development. We also revamped the indexing data structures, allowing for easy identification of missing and orphaned images. We added layers of image consistency checking and a client application that enabled a business analyst to view problematic images and recommend action.
Software Technologies
Results
The data that Trident Technologies was provided for Signature cards was loaded successfully in Signature. Throughput with the new multi-threaded architecture was improved, from 5-6 seconds per image to 4-5 images per second. The images were loaded ahead of Conversion Day, and missing/orphaned/corrupted images identified and handled expediently.