When people think of UPS, they generally don't think of the technology required to deliver over 3 billion packages per year. With 327,000 employees in more than 200 countries, there's a lot going on behind the scenes to get deliveries to their doorsteps.
In 1998, when we began assessing our growing technological needs, we determined we were facing several problems. First, our users had no single view of business and customer data as it was distributed across multiple platforms. Second, our data storage requirements were growing at a phenomenal rate. Third, the availability of the data had to be timely if we were to serve our customers better. To solve these problems, we turned to Hewlett-Packard (HP) and its proven data warehouse solutions.
Working with HP, our assessment began with our Information Library, housed on an aging legacy system. This library was initially created to perform efficiency monitoring and performance analyses. However, trend analysis on users and CPU MIPS showed that it had outgrown its original intent and was being used for operational reporting, including financial, human resources and security. To address these new users, we realized that the new system would have to meet several stringent requirements. The system would have to be highly scalable because of the exorbitant amount of ever-increasing data required by these users. The system would also have to be a performance powerhouse because of the number of concurrent users who would be querying the database at any given time.
The HP 9000 V-Class server line met all of these requirements; and in August of 1998, UPS migrated its legacy Information Library to an HPV2250/12 running Oracle7.34 against an EMC data store. This migration and the data conversion were not small undertakings considering how data storage had changed over the years, but the migration was successful and the V2250 has easily met our performance expectations.
Since its inception, our new UPS Operations Information Library has grown to 500GB of raw data and is expected to grow to 1TB. It supports over 4,500 users, 350 concurrently, with no signs of performance degradation. We feel that we are now positioned for growth with the V2250; and, should the need arise, there is always the option of clustering to our HP 9000 V2500/16 for application failover between the Information Library and our customer data warehouse. To date, we have not needed to do so because of the high reliability and availability of the HP V-Class servers.
Our efforts included our customer databases, a total of three that were ported from non-HP servers to one HP9000 T500 and two HP 9000 K-Class servers in 1996. Unlike the Information Library, these databases are closely aligned with data warehousing concepts, negating the conversion need. The data still needed to be scrubbed and merged into one cohesive warehouse before we could get to a single integrated customer view a necessity for campaign management, product monitoring and sustained customer relationships.
Not surprisingly, the technological needs of this warehouse were the same as those for the Information Library. Unlike the Information Library, however, this warehouse was expected to grow exponentially by comparison, so a more robust platform was needed. Once again, our sights fell on the HP 9000 V-Class line, and a V2500/16 running Oracle8.5 was selected.
In 1998, when the warehouse was first built, it was loaded with 700GB of data that quickly grew to 1.2TB within the first year. Implemented in a star schema, which closely matched our users' perception of data, we added meta data to control access by the 1,000 users (200 running concurrently). This warehouse is expected to grow to 3TB within the next year; and it has become our most important business intelligence application. Used to make business critical decisions on a daily basis, it has never let us down.
Initially, access to both warehouses was through a Microsoft Access front end, which was familiar to the Information Library users. This front end, however, often required a knowledgeable user, familiar with the logical database model.
For the varying needs of the different UPS groups, we assembled a toolkit that included Allaire's ColdFusion to allow the groups to develop their own Web-based applets for pulling database information.
In a little less than a year, data throughput more than quadrupled in response to user demands a great indication of our success.
|The following technologies were used in the UPS solution: HP 9000 V2500/16 Enterprise Server for the Customer Data Warehouse, HP 9000 V2250/12 Enterprise Server for the Operations Information Library Data Warehouse, HP-UX 11.0, Oracle7.34 and Oracle8, EMC Disk Arrays, StorageTek silo for tape backup, and Microsoft Access and basic Web development tool (Allaire ColdFusion) for client access.|
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access