Data virtualization has been around for decades and has always been controversial. In the 1990s, it was called virtual data warehousing or VDW-- or as some skeptics liked to say, "voodoo and witchcraft”. It’s also been known as query federation and more recently, data services. The idea is that business users don't need to know the location of the data; they merely need to log into the data service and all data appears as if it’s local to their server, modeled in a fashion that makes sense to them. 

 Andrew Sohn is the Global Head of  Data and Analytics at Crawford & Company, a $1B+ service provider to the insurance and risk management industry, where he designed and leads its data and digital transformation strategy and program. With more than 25 years in the industry, Andrew has managed a broad range of infrastructure and application technologies. He’s a strong advocate of data virtualization technology and believes it is an integral part of a modern, agile data ecosystem. 

Key Takeaways:

• Organizations are still skeptical to use data virtualization in their analytics architecture.
• Over the years, the capability of data virtualization has got better as a package and it has become much more useful due to the advancement in technology.
• Data virtualization is useful in cases where you don't need to move the data and replicate it. 
• It is possible to introduce data virtualization in your existing analytics ecosystem and reap its benefits.
• Andrew recommends data virtualization as a component for modern data architectures.
• Having an API strategy is key to get real-time data to your users. 
• Just like an electrical switch, APIs can be turned on and can get you data wherever it lies. 
• Data virtualization has its roots in the data warehousing but it also has a hook in the modern world where it can provision more transactional systems.

Would you like to speak to an Eckerson Group expert about data and analytics?

Below is one question and answer from the podcast

Wayne Eckerson: In your experience, what components should be used to build a modern data architecture?

Andrew Sohn: Data virtualization would definitely be a component of a modern data architecture. Other than that, it should have an abstraction layer, an API strategy with a very robust API management platform, a data lake strategy and a data repository strategy or a data warehouse/data mart that has been processed for various purposes. There may be some others to execute specific use cases.