Since the inception of the Internet, there has yet to be a topic as controversial as "Push". In fact, it's hard to pick up a trade journal that doesn't talk about it. So, without writing a dissertation on the value of "Push", let's explore an extension of this idea. "Managed Push" is a methodology that lets everyone in the information chain have the tools, access and security to obtain enterprise-wide data when, where and how they need it.

For example, you're the CFO. You walk into the office 20 minutes before a board meeting. You flip open your laptop, click on your Web browser and, voil‡: The report you need on operating expenses by branch is sitting there waiting for you. No scrambling to pull spreadsheets together, no scratch pads with columns of arithmetic, no last minute conference calls to finalize the numbers. You're done, on your way, ready to go. At the same time, throughout the corporation, other decision-makers, knowledge workers or analysts, call them what you will, have exactly the information they need in time for their meetings. This scenario is only a recent possibility, a result of combining "Push", the warehouse and the Web.

How Many Ways To "Push"?
"Push" can be defined in three broad categories: indiscriminate, selective and publish and subscribe. Indiscriminate "Push" is an ongoing trickle of updated information, used to distribute data from inside, or outside, the organization on a real-time basis. Examples include news releases, stock quotes and sports scores; distributed to the desktop, pager or fax. For single or small groups of subscribers, this is a very effective method for up-to-the-minute information. The downside occurs when network bandwidth is clogged with information that may not be critical to a company's success.

Selective "Push" gives the user, and the company, more control. Instead of a constant channel of information, subscriptions can be defined at the server level. A filter sorts and prioritizes data to be delivered to the desktop, funneling external and internal information to subscribers. This is effective for pushing pre-defined sets of information to classes of employees, say by job title. These smaller packets of information place less stress on network traffic. The drawback is profile administration. Management tools are not yet available for large companies and the server administration alone often requires several people.

Publish and subscribe is a balance of subscriber control and administrator process management. Instead of pushing massive amounts of information to the desktop, data is published one time to the server by the administrator. Subscribers then determine when they want the information, and under what criteria. When the scheduled time arrives, the information is pushed to the users. Subscription is usually by time or event, requiring scheduling and monitoring software. Users have the ability to discontinue a subscription or change the delivery schedule. Administrators need not worry that resources are being used to push information to a user who has changed their requirements. To achieve this flexibility, subscription interfaces and management tools must all be in place.

Separating Informational from Operational Systems
Today, there are two sides to information systems: operational, or transaction, and informational. The operational side includes the systems familiar to most people. These are the HR, payroll, finance or manufacturing systems that run the daily operations. Operational systems require data to be stored in a specific manner to achieve maximum performance from those systems processing the day-to-day tasks of running the business.

The informational side was created to pull corporate information together. Informational systems support analysis and decision-making and require the data to be stored in an entirely different manner for correctness and clarity. Data warehousing lies within the informational, or decision support, side of the organization's information systems and has its own unique requirements for information flow.

Data from multiple operational systems is collected, transformed and integrated onto a single platform for decision support. These warehouses are today referred to as "data marts", "operational data stores", "reporting databases" or "query databases". All have the same function - to provide a universal resource for knowledge workers to locate information upon which to make their daily decisions.

Pushing Data to the Warehouse
Pushing data requires that there be several elements in place. First, there must be a process to gather data from legacy, operational and external databases and push it to the warehouse. Often, data is in different media, formats or types and can have multiple definitions. Moving data to the warehouse is not simply a matter of writing a program, but of intelligently delivering the desired data to the appropriate targets.

A second consideration is the sheer volume of data being moved. Operational databases may contain gigabytes of data from systems having continuous update cycles. Specifically, updating the warehouse with only the desired data, which may be just the changes since the last update (known as Changed Data Capture), is an excellent opportunity to use indiscriminate "Push". Timing and data selection are paramount to "pushing" data from the operational side to the warehouse.

Pushing data from operational systems through the warehouse and out to the desktop is a process management opportunity. To effectively manage and distribute corporate data, a set of tools must be integrated into the data movement/push scenario. These tools include built-in scheduling mechanisms and recovery and restart, to ensure job completion and data integrity.

Central to process management is a transformation engine, or concurrent manager. This engine can avoid bottlenecks and single points of failure while executing processes in a distributed manner. In addition, by load balancing work across multiple servers the concurrent manager creates a scalable solution that IT administrators can use to leverage data across multiple environments. Finally, system tuning and process refinement is simply achieved with statistics on rows moved, transformed and loaded; in effect, creating a continuous information loop.

The result is a tightly integrated "managed push" solution which gives corporations the tools and processes to actively leverage the wealth of information available from enterprise data. The data warehouse becomes a reservoir of mission-critical information to be shared across the organization. By combining scheduling, monitoring, movement and recovery facilities, organizations can automate the information environment and directly contribute to the success of any warehouse project.

Managed Push Opens the Warehouse on the Web
OK, so now you have a data warehouse with data you've "pushed" from the operational systems. Nevertheless, there's one essential issue: a data warehouse is a technical solution to a business problem. Decision makers, who would most benefit from the warehouse, are the least likely to use a query tool or run an SQL-based report writer to get information. The IT teams that built the warehouse have a wealth of integrated, corporate-wide information but do not have a way to deliver it to information consumers. Fortunately, the Web has reversed that scenario.

A recent study published by Robertson, Stephens & Company noted that they expect the Internet to replace client-oriented tools due to the low cost of deployment, easier administration and distribution enabled by the Internet. (Robertson, Stephens & Company, August, 1997, "Turning Data Into Decisions and Information Into Insights," p.13.) It is not a stretch of the imagination to see how the Web will open a whole new venue for accessing, reviewing and sharing organizational data. The challenge is defining the process and having the tools to enable such an application.

With an informational system infrastructure in place, the most effective method to push information to the knowledge workers is "publish and subscribe." This "Push" method extends "managed push" to the desktop by giving knowledge workers the power to schedule only the information they need, when they want it. These knowledge workers may require a stream of defined reports and information in a repeatable and consistent manner. "Publish and subscribe" is a perfect application to deliver integrated, enterprise-wide information to the desktop, without burdening the recipient with additional training or systems knowledge.

Additionally, "publish and subscribe" can streamline the information delivery process. It is much more efficient to publish views of the data and assign security to those views once, in the warehouse. Users may subscribe to views or subsets of views, and determine the delivery format and schedule. By publishing data once, to the server, the data is contained in a central location, reducing data redundancy. All users are now able to obtain and reference the same information, providing a common basis for decision-making across the organization.

Pushing From Beginning to End
"Managed Push" moves the crucial business information out of operational systems to the desktop, with maximum control and flexibility. The universal interface, cost-effective deployment and accessibility of the Web browser have transformed the information in the warehouse into a resource available to any knowledge worker within the corporation.

Only when these leading technologies, combined with the push methodology, are used together,can organizations bridge the gap between knowledge workers and operational systems. The incorporation of a "managed push" ensures that companies can leverage their data in a productive, measurable and secure manner.