Ten Mistakes to Avoid When Virtualizing Data – Part 2
September 6, 2011 1 Comment
In my prior blog post, I revisited mistakes one through five from my November 2008 Virtualization Journal cover article entitled Ten Mistakes to Avoid When Virtualizing Data. In this post I will address mistakes six through ten and summarize what has changed since 2008.
Mistake #6 – Failing to Simplify the Problem
While the enterprise data environment is understandably complex, it is unnecessary to develop complex data virtualization solutions. The most successful data virtualization projects are broken into smaller components, each addressing pieces of the overall need. This simplification can occur in two ways: by leveraging tools and by right-sizing integration components.
Roles and Reference Architecture for Data Abstraction Success is an article that directly addresses this recurring mistake with common sense advice about using a well-organized team and data virtualization reference architecture to rationalize complex data landscapes into a set of reusable data objects.
Mistake #7 – Treating SQL/Relational and XML/Hierarchical as Separate Silos
Historically, data integration has focused on supporting business intelligence needs, whereas process integration focused on optimizing business processes. These two divergent approaches led to different architectures, tools, middleware, methods, teams and more. However, because today’s data virtualization middleware is equally adept at relational and hierarchical data, it is a mistake to silo work on these key data forms.
Over the past three years, these technology silos have broken down to support business requirements that cross them. How Data Virtualization Increases Business Intelligence Agility identifies a number of ways that data virtualization can federate relational and hierarchical data sources.
Mistake #8 – Implementing Data Virtualization Using the Wrong Infrastructure
The loose coupling of data services in a services oriented architecture (SOA) environment is an excellent use for data virtualization and a frequent use case. However, there is sometimes confusion about when to deploy enterprise service bus (ESB) middleware and when to use data virtualization platforms to design and run the data services typically required.
There is greater clarity here today than three years ago. As such fewer organizations now make this mistake. SOA + Data Virtualization = Enterprise Data Sharing and Data Services Platforms–Bringing Order to Chaos provide advice on how best-in-class data virtualization implementations leverage SOA principles and technologies.
Mistake #9 – Segregating Data Virtualization People and Processes
As physical data consolidation technology and approaches have matured, so too did supporting Integration Competency Centers (ICC). It was a mistake to assume that these ICCs, cannot or should not also be leveraged in support of data virtualization.
This mistake has been recognized. Today, data virtualization strategy, design, development, and deployment is often delivered side-by-side with other integration techniques within a larger ICC. If I Could Only Recommend One Data Virtualization Best Practice, What Would It Be? highlights the importance of this integrated approach to the data virtualization competency center.
Mistake #10 – Failing to Identify and Communicate Benefits
While data virtualization can accelerate new development, perform iterative changes quicker, and reduce both development and operating costs, it is a mistake to assume these benefits sell themselves.
This remains true today. What’s So Great About Data Virtualization? provides an excellent summary of the challenges data virtualization addresses as well as the benefits delivered. In addition, other’s data virtualization successes can provide a lens through which you can view your own. Use these as guides when communicating your data virtualization successes.
Netting it Out 2011 vs. 2008
Data virtualization’s early adopters gained critical knowledge when implementing their data virtualization solutions. Mistakes were made. But lessons were learned.
Many of the mistakes experienced back then are no longer valid today. And those that remain have been mitigated with improved technology and implementation best practices.
To err is human. But if you are willing to learn from your peers, to err using data virtualization will be less frequent.