You have read all the reports, evaluated the vendors, and made your decision: Data virtualization is the right investment for you.
Just one catch. Like all enterprise-grade data management middleware, before you can gain data virtualization’s benefits, you first have to implement it. Successfully, that is.
For many, that will seem a significant challenge, but it doesn’t have to be. If you spend the time and effort to understand best practices and lessons learned from others who have succeeded, it will help you achieve data virtualization’s many benefits—and a lot sooner and with less effort than otherwise.
The purpose of this paper is to help ensure your data virtualization implementation success. Based on hard earned knowledge gained by adopters and system implementers around the world, consider this your introduction to the 12 most essential data virtualization implementation best practices.
While every journey starts with a few steps, it helps to know where you are going.
Data virtualization can support many data-driven business objectives, multiple lines of business, teams, and roles, with data that spans diverse sources and types. Given this breadth of opportunity, you must focus your efforts to achieve the biggest business impact. It is essential that you align your data virtualization implementation strategy with your digital transformation strategy.
Will your data- driven success be realized by better customer engagement, faster innovation, operational excellence, orsome combination?
The answer to this question informs your data virtualization implementation strategy.
Will your data-driven competitive success be realized by better customer engagement, faster innovation, operational excellence, or some combination? The answer to this question informs your data virtualization implementation strategy.
For example, if your digital transformation strategy is to increase cross-sell revenues within your installed base, then focus your data virtualization efforts on building 360 degree views of your customer data. Knowing your customers’ preferences, prior purchases, current activity, and more provides the insight you need to intelligently offer just the right cross-sell product at just the right time, and do so in a way that engages and delight them.
Over time, your digital transformation strategy and data-driven response will evolve and grow, and, so, too, will your data virtualization implementation. The good news along the way
is that data virtualization’s ability to do many things well and quickly will accelerate your transformations.
Successful data virtualization implementations can be complex. It helps to have a comprehensive plan that covers your journey from project kickoff to steady-state operations. Be sure your implementation plan includes these elements:
To improve your odds of success, engage system integrators (SIs) with strong data virtualization practices and positive references. These specialists have many assets that can help you in several significant ways.
Rationalizing today’s increasingly distributed data landscape is a big reason why you decided you needed data virtualization in the first place. Logical data warehouse, data fabric, and data- as-a-service are typical labels for these new types of modern, distributed data architectures, and data virtualization is a common element in all of them.
Data virtualization’s consistent, secure, and governed common data layer makes it easy for your business users to find and use all your most important data regardless of location. And it provides the decoupling opportunity that lets your IT team store and manage your data, independently from how it is consumed. So
it stands to reason that your data virtualization implementation strategy aligns to your data architecture strategy.
But just as Rome was not built in a day, neither is a distributed data architecture. As you transition from your current to your future state of data architecture, you will need to execute some difficult migrations. Using data virtualization as a migration tool lets you insulate your consumers as you move their data sources, insulate data sources as you migrate your consumers, or do both depending on how your data architecture evolves. For more detailed tips and tricks on how you can use data virtualization so accelerate these architectural driven migrations, read this blog.
Data virtualization is not the best solution to every data integration challenge. Sometimes it is, but sometimes it’s better to pre-consolidate your data in a warehouse or mart—and sometimes a combination of integration methods is your best choice. Getting this decision right for each data integration project is not only important to the success of that project, those successes also influence your organization’s perception of data virtualization overall. And a positive reputation based on a positive start will help you garner the executive support and additional resources you need for long-run success. So it is important to select your use cases wisely.
How do you know which use cases are the best fit for data virtualization? There are several best practices you can use to answer this question.
Because many factors can influence whether or not to use data virtualization, you need to intelligently consider all these factors in a consistent and systematic way. TIBCO provides
a Data Integration Strategy Recommendation Tool that lets your architects, designers, developers, and/or project leaders make more informed decisions on when to virtualize, when to consolidate physically, and when to apply a hybrid combination.
Synthesized from conversations with industry analysts and hands-on project teams—and now adopted by TDWI for their data virtualization courses—this tool applies 13 decision factors as follows:
Pattern-match common data virtualization use cases In some cases the best way to decide on whether to use data virtualization is to compare and contrast your current use
case against proven data virtualization use cases. When you
identify a match, you can proceed confidently. Applying Data Virtualization: 13 Use Cases that Matter is a useful resource to help you compare and contrast your use cases with the most frequently adopted data virtualization use cases. For each case, the paper:
If you are like other data virtualization adopters, the longer you use data virtualization, the less you will rely on the strategy recommendation tool and pattern matching approaches above. You will simplify your decision process by applying
data virtualization as your default method of data integration for every new use case request. And you do so even when you know that physical data consolidation may be the ultimate implementation method.
Why virtualize first? Your benefits from a virtualize-first approach include:
As noted above, your data virtualization implementation can support many diverse use cases from simple projects to
complex enterprise-wide data architectures. When faced with so many opportunities, what is the best way for you to get started and then expand to broader use?
Based on knowledge gained from hundreds of data virtualization implementations, staged deployment and adoption has proven the best practice. The key to success is start small, build your skills, and then grow based on proven results.
Implementing a series of small projects is your best starting point. This lets you provide the virtualized datasets you need to quickly deliver business benefits, while avoiding the risk inherent in large projects. Because you are new to data
virtualization, keep your implementations simple. Do not over engineer the views and services you build. At this stage be “good enough for now,” knowing you can improve these assets in the next stage when you have expanded your team, refined your skills, and developed your best practices.
With the winds of success at your back, you can expand your data virtualization scope. Instead of focusing only on the datasets you need for presenting use cases, you can think more broadly. This is the time to develop your standards and engineering best practices, perhaps in the context of a Center of Excellence. And this is when you gather reusable virtualized datasets you can use to support larger domains, for example, the dataset you need for:
Eventually you can pull together multiple domain-scale implementations into a broader enterprise-wide data virtualization implementation. This is the stage where you:
But recognize that your definition of enterprise scale can be different than other organizations. For example, enterprise data virtualization at one of the largest banks in the world spans seven business domains including high net worth customers, credit card customers, back office operations such as security services, fixed income risk management, and more. Yet significant lines of business are still in earlier stages.
Data virtualization centers of excellence (CoE) are a proven people, process, and technology approach you can use to build competency, drive consistency, and accelerate value realization. When implementing a data virtualization CoE, you have two options you can choose based on your organization’s unique situation and needs.
Either way, your goal remains the same with multiple benefits and results including:
and resources.
So you can fully focus your early CoE efforts, start small by limiting the scope of data virtualization CoE functions, projects, or scope in some way. As you achieve results, communicate them to reinforce the CoE team’s successes. Then use the momentum you gain to accelerate further implementation success.
Data Model.
One of data virtualization’s greatest business benefits is its ability to shield your data consumers from your data
complexity. Delivering business-friendly data that is easier to understand and use, lets you gain the data-driven insights you need to compete and win.
But you cannot achieve this with data virtualization alone. You need to combine your data virtualization implementation efforts with your enterprise data model management efforts.
The Adopt in Stages and Leverage a CoE Approach best practices above describe key process elements you can use to align the timing and scope of your common data virtualization and data modeling efforts.
The data virtualization abstraction best practices summarized below describe the data layer architecture you can use to instantiate your enterprise data model. Its three primary layers include:
of “logical” or “canonical” views that represent your most important and commonly used business entities.
Across your data virtualization enterprise data model implementation, governance is critical. Ensure your data model governance roles and responsibilities extend into your data virtualization CoE. Mandate a data model stewardship process that maintains consistency using shared workflows. Implement a common change process that ensures implementation success.
For more detail on these modeling best practices, see this whitepaper.
When you implement data virtualization for the first time, you may be concerned about query performance, possible production system impacts, and how it aligns with your existing data integration and storage strategies.
Questions that typically arise include:
This whitepaper, 10 Best Practices for Using TIBCO Data Virtualization to Achieve Performance SLAs, summarizes the essential design for performance and workload management best practices you can use to design for optimized data virtualization performance. Here is the list:
One of the biggest benefits of data virtualization is how it enables you to consistently implement data security across your data sources and consumers. With it, you can ensure that your authenticated users can see only appropriate, authorized data and nothing more. And when even higher levels of security are needed, data virtualization can mask and encrypt your data. This whitepaper describes these capabilities in more detail.
Because security impacts every view and data service you build, every data source you query, and every user who consumes these datasets, you must consider data security holistically across every stage of your data virtualization implementation. For best success, apply these best practices:
Your data virtualization implementation will force you to drill down into your data. As a result, you will uncover many data quality challenges. But the good news is you can use data virtualization several ways to directly improve your data quality.
Data virtualization’s centralized metadata-driven syntactic and semantic transformations and enrichments let you standardize your datasets and encourage reuse. This means everyone is on the same page, improving not only the quality of data itself, but every application where it is used.
Data virtualization also affords you the opportunity to make fewer copies of your data. Fewer copies improves your data quality by substantially reducing the transformation errors and entropy inherent in typical multiple-copy data warehouse and data lake deployments. For more on how data virtualization lets you create fewer copies, read To Improve Data Quality, Stop Playing the Data Telephone Game.
And when things change, as they inevitably do, it’s a lot easier for you to modify data virtualization’s centrally managed metadata definitions than it is to modify the multiple distributed ETLs and database schemas you need when using physical data integration. So your data quality doesn’t degrade with every business change.
If yours is like most organizations, your data management tools include a broadly diverse set of sometimes adjacent, sometimes overlapping, vendor applications that span metadata management, master data management (MDM), data governance, data catalog, data modeling, data quality, and data security — as well as the various data integration types including replication, bulk/batch ETL, and data virtualization.
Often purchased at different times for distinct purposes, these siloed data management capabilities increase complexity, costs, and organizational friction.
While data virtualization provides many of the core metadata management, modeling, quality, and governance capabilities you require, it does not cover everything you need. But it can be a great foundation for better synergies with other data management solutions. Doing this right will let you achieve better data management success with less effort.
The first step is to embrace today’s data management functionality convergence. When selecting your data virtualization vendor, look for those who also provide advanced data management capabilities in converged platforms you can use to unify your data management capabilities. These include metadata management, MDM, data governance, data quality, data catalog, data modeling, data security, and other data integration methods. This will simplify your data management silo problem, while saving time, cutting costs, and reducing friction.
The second step is to simplify your approach to data management. Have you ever wondered why you have treated your metadata, master data, reference data, transaction data, and streaming data assets as unique, unrelated data types and manage them in unique, unrelated ways? You no longer have to. Data virtualization lets you unify what is common across your data assets including how they are accessed, modeled, secured, governed, and more. This holistic approach makes it easier
to manage your data assets, especially your most important, frequently used ones.
Once you complete your initial implementation, what happens next? Do you rest on your laurels or push ahead pressing your data virtualization fueled advantage? With so many new use cases and valuable benefits now accessible, and new data virtualization capabilities coming your way with each new release, pressing onward is the best answer.
Pressing onward requires stamina. A stamina fueled by achievement as success begets success. But first you must communicate your successes.
And keep pace with the ever changing world of data virtualization, so you can take advantage of new opportunities as they arise. To stay informed:
Implementing data virtualization can be a significant challenge. But one made easier if you master the best practices and lessons learned from others who have succeeded. This paper provides 12 implementation best practices you can apply to help ensure your data virtualization implementation success.
Take advantage, so you can make data virtualization your organization’s cutting-edge advantage.
Cheers to your success.
Global Headquarters 3307 Hillview Avenue Palo Alto, CA 94304
+1 650-846-1000 TEL
+1 800-420-8450
+1 650-846-1005 FAX
TIBCO Software Inc. unlocks the potential of real-time data for making faster, smarter decisions. Our Connected Intelligence platform seamlessly connects any application or data source; intelligently unifies data for greater access, trust, and control; and confidently predicts outcomes in real time and at scale. Learn how solutions to our customers’ most critical business challenges are made possible by TIBCO at www.tibco.com.
©2021, TIBCO Software Inc. All rights reserved. TIBCO and the TIBCO logo are trademarks or registered trademarks of TIBCO Software Inc. or its subsidiaries in the United States and/or other countries. All other product and company names and marks in this document are the property of their respective owners and mentioned for identification purposes only.
