HomeData LakeData Virtualization vs Data Warehouse

Comments

Data Virtualization vs Data Warehouse — 3 Comments

  1. Pingback:Data Mesh defined | James Serra's Blog

  2. Hi James,

    Thanks for the great posts about “Data Virtualization”:
    1) https://www.jamesserra.com/archive/2018/02/data-virtualization-vs-data-movement/
    2) https://www.jamesserra.com/archive/2017/08/data-virtualization-vs-data-warehouse/

    One area of strong recent interest is about leveraging the rich data-semantic of large-scale high-performant central tabular model (AAS or PBI large premium dataset) as quick-to-market zero-data-movement” Data Virtualization layer. With higher appetite for good central semantic layer (leveraging PBI large datasets), more experts are avoiding duplicated data logics in other classic Data Virtualizations when richer data + semantic already exists in their large PBI datasets. (Data Virtualization principle of zero-movement)

    They see additional cost & technology efficiencies (SOLID principle) on leveraging the rich central data semantic layer beyond the traditional BI “pattern” (reporting tools querying the tabular model). Now it’s possible a true zero-data-movement “Data Virtualization” layer for:
    a) by upstream ETLs systems consuming aggregated/augmented information (using PBI REST API for DAX Query)
    b) by the brand-new “Data Query for PBI Datasets” (up to 3-levels of composite Direct Query Datasets).

    Pattern vs Anti-patterns

    What makes this topic interesting is not that many champions are already doing it (considering the performance and volume limits of course), but that many classic “architects” are feeling out of their comfort zone and are rushing to label this as an “antipattern” (in the name of “performance concerns”, etc.).

    So your post “Data Movement vs Data Virtualization” helps a lot to review the key principles that should drive the conversation of “Data Virtualization” (pattern vs. anti-patterns).

    The answer to this is important, as more and more organizations are seeing huge value on leveraging the rich data augmented data + semantic content there as good foundation for Data Virtualization (not only for Visualization but for upstream transformations at the next level).

    What do you think James?
    Do you see any “philosophical” antipattern in using a well-designed large-scale high-performant Analytic Service (based on AAS or on a PBI Large Premium) as a valid architectural solution for a fast-to-market zero-data-movement Data Virtualization layer? Or should organizations duplicate the same logic on other Dataflows, Dataware houses, etc.

    Thank you, James!

Leave a Reply

Your email address will not be published. Required fields are marked *

HTML tags allowed in your comment: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>