Where Should the Data Reside?

IT organizations evaluating business intelligence solutions have to consider how to provide access to those who need it inside and outside the organization while keeping the data secure. This balancing act can result in two different approaches. Each has advantages and drawbacks that we will explore. Traditionally, an integrated approach that keeps data within the organization has been preferred. However, recently the idea of creating an external extranet hosted at a third-party site as part of a business intelligence solution has been gaining some traction.

In the traditional approach, IT creates an extranet that gives your partners access to data maintained inside your corporate network. Your partners access this information through a secure connection and are granted access only to the data important to their relationship with your business. This access allows your partners to quickly make the necessary decisions to ensure they are meeting the demands of your business. For instance, your supplier of widgets may decide to ship the next batch of widgets to you early because demand has increased and your inventory has reached dangerously low levels. Without your supplier having access to that information, you might run out of widgets, forcing your customers to purchase them from an alternate supplier. This would mean that you've lost sales and disappointed a customer. In contrast, when suppliers have this information, they can manage their own inventory, which frees your staff to focus on other areas of the business.

If IT chooses to move the extranet to a third-party hosting site, one immediate benefit is the separation of your partners' and customers' activities from your corporate infrastructure. This separation relieves your internal network from spikes in activity levels that could stress your network bandwidth and server performance and possibly decrease the productivity of your own employees as they compete for those same resources. With a hosting company managing the infrastructure for your business intelligence solution externally, your IT group doesn't need to build up your internal infrastructure to meet this demand. In addition, your IT group will not have the burden of day-to-day server maintenance associated with the extranet.

What Are the Security Issues?

By externalizing the server, your internal network is completely segregated from potential security breaches that might arise from non-employees entering your network. You may need to externalize your data simply because you don't have the skill set to adequately secure access to the information, nor does your staff have the time to stay abreast of all the new methods for "hacking" through to your information. It becomes the responsibility of the third-party site to keep pace with the ever-evolving hacker community.

Data on the externalized extranet is generally redundant and stagnant, providing some resiliency. If a malicious hacker compromises your data and the solution is rendered unusable on the external extranet, it does not shut down your business. It would limit your partners' ability to get their information but only for as long as it takes to recreate your environment. If your extranet design simply pushes data to an alternate server, your recovery time is limited to recreating the server and pushing the data back out. In contrast, if you are utilizing an integrated approach, your recovery could be considerably longer as other servers and devices exposed could require damage repair as well.

How Much Data Should Reside Externally?

If you already have an existing business intelligence solution and think that an external extranet sounds like a good idea, you need to consider the type of data that will reside externally. You probably don't need (or want) to replicate your entire data warehouse. An external extranet is designed for a different user community. Your customers and partners won't have the same software tools available to them for mining the constructed data. Think about what data customers and partners need to access. You may need to create specific data marts and operational data stores for these two groups. By presenting data that is specialized for their use, you will limit your security risks and enhance the users' experience.

With a different user community comes a varying degree of sophistication of the user. Some partners and customers may fully understand the nomenclature used by your company and industry while others may not. Your business intelligence tool needs to accommodate even the lightest of users in order to keep its advantage. For example, an online analytical processing (OLAP) solution and presentation works well for business users but might not be appropriate for customers.

You also need to consider how these user groups will access the data. You don't want the burden of managing distributed software. You may want to look at a "zero footprint" approach built around Web browser technology. It solves the administrative overhead caused by distributed software, but it could create a new demand for your programming staff to retool your business intelligence solution for display in a browser. A browser-based solution helps meet the demands of a very broad range of users, but this may require a more user-friendly approach to designing a solution. In addition, your IT group will need to investigate the limitations of using a browser when exploring this option. The capabilities of browser technology have come a long way (particularly the advances in Flash technology), but there still are benefits associated with a thicker client.

What Is the Downside of an External Extranet?

The implementation of an external extranet also has its downsides. One of the biggest is that you lose control of your data. When you move your data off your corporate network, someone else is managing the servers for your solution. Questions will arise concerning the security of your data in terms of timely backups, auditing of activity against your data, and potential breaches in security. If you are subject to Sarbanes-Oxley, HIPAA, and other legal requirements, you should be aware that there will be additional security and auditing requirements of the data being exposed. Proper planning for security-sensitive information needs to be part of any successful project. The cost of compromising personal information is more than just litigation. It extends to loss of reputation, bad publicity, and reduced consumer confidence. Many of these same risks reside with a traditional integrated solution. However, in my opinion, the importance of securing your company's valued information is much more fully appreciated by the staff designated to maintain your transactional systems.

Traditional Extranets Offer a "Single Version of the Truth"

With a traditional extranet, customers and suppliers gain access to one business intelligence solution or a "single version of the truth" when it comes to data validity. In the data warehousing community, we all understand the importance of maintaining a "single version of the truth," and we also understand the difficulty in achieving that goal. Any time data is moved from a source system to a target system, there is a risk of losing control of that "single version of the truth." If the external system no longer contains the data equivalent of the truth for your business, your partners and customers could make less-than-optimal decisions. This information gap could create missed opportunities, cause missed delivery schedules, and reduce customer satisfaction. Any one of these outcomes will deteriorate users' confidence in the system, and you'll lose the business benefits of a business intelligence system. The traditional integrated system does not require movement of data beyond your data source, data warehouse, data marts, or operational data stores. Your data remains consistent with your designed systems.

Use an Agile Methodology for Implementation

If you are just beginning your foray into business intelligence, you need to consider what methodology and design will allow you to provide data to your partners most quickly and cost-effectively. Utilizing an agile approach to a business intelligence project provides significant benefits in creating an appropriate solution. An agile approach suggests that small solutions are generated as deliverables every 30 to 45 days. One of the primary problems in running a business intelligence project is insufficient requirements. An agile approach allows you to create smaller deliverables and adjust those to the users' liking. This approach compensates for having unknown requirements at the outset because the solution can be easily modified as forgotten details surface. When using an agile approach, the development team traditionally utilizes reports generated from existing query tools or smaller applications to deliver product to the users. The benefit of this iterative approach is delivering initial results faster, providing the ability to adjust the results quickly, and minimizing costs by using existing tools. This approach also reduces the need to design the data warehouse in its entirety. The initial deliverables can be generated over the existing transaction files without data modeling or creating ETL programs to extract data. Those activities can be revisited after the business' data standards are uncovered through the smaller deliverables.

A major benefit of creating a solution using your existing transaction-based files or an existing data warehouse is that the deployment time for your partner extranet is greatly reduced and even potentially eliminated. There is no need to develop additional tools, and the necessary data does not need to be moved to an alternate location. Simply secure the access and rights to the data, and your solution is available. More importantly, your existing data remains your "single version of the truth." If your solution uses your existing transactional data sources, you can be sure the data is being backed up on a regular basis and is being properly secured by your IT team. Lastly, the solution is simpler. No checks and balances are necessary to validate data.

Considerations for Companies with Integrated Business Intelligence Solutions

For companies that have an existing integrated business intelligence solution and are preparing to provide access to external customers and partners, you should consider the new demands with which your network will be presented. There are ways to segregate traffic and minimize the impact of customer and partner activity from your normal company network traffic. Segregating the network traffic also limits the potential for unwanted access to your network. You will need to put safeguards in place to ensure customers and partners have access only to the data that they need. If your transaction systems are exposed, securing those systems becomes monumental.

What Are the Downsides to the Integrated Approach?

There are some downsides to the integrated approach as well. Any time you open up your existing network, you run the risk of compromising the entire system. Beyond the fears of your data being exposed, the additional workload from numerous partners or customers using your network needs to be considered. Application performance can suffer as data is queried over the entire data set rather than summarized data. Summarizing data in this manner also requires additional server power in order to handle the inbound daily transactions as well as the ad hoc reporting requests. Many performance issues can be eliminated as your business intelligence project matures.

Providing Information to Everyone Who Needs It

Providing partners, customers, vendors, and other entities access to the information they need to speed decision-making is essential to growing the bottom line of your business. Determining the appropriate method for accessing this data will require not only proper planning, but also—and more importantly—a risk versus reward assessment. Remember to consider security, performance, and deployment time when selecting the right solution for your organization. Finally, don't forget to use an agile approach to deployment, delivering results to your users at every step along the way.