Analytics: open middleware for multi-source and multi-target

 

Analytics:  open middleware for multi-source and multi-target Let's imagine that data visualization tools no longer have to manage the complexity of sources? Let's imagine, conversely, that sources no longer have to worry about the diversity of data visualization tools?   With this in mind, we imagined oa-lake (and we implemented it in major accounts ):  an orchestration and unification layer between data systems and business tools,  based on a distributed columnar & in-memory SQL engine.   {oa-lake} : upstream interoperability  : combining sources in a single format  No refactoring required: oa-lake   understands the sources as they are. This is particularly useful in a context of partial migration, where the same reporting environment still relies on legacy sources (Oracle, DB2, etc.) while integrating modern Cloud databases (BigQuery, Azure SQL, Redshift, etc.). In a migration context, we thus avoid "big bangs", often synonymous with... major catastrophes. {oa-lake}  : downstream interoperability: a single data model for dataviz oa-lake centralizes business logic and business rules in a single layer, independent of data visualization tools. Visualization solutions (Looker, Power BI, Strategy, etc.) thus become simple query and display engines , relieved of the burden of modeling and business rules. oa-lake is also able to reproduce the principle of semantic layer from one tool to another in a migration context — for example, from Business Objects to Power BI — by restoring business objects in a usable form and without loss of functional consistency. At the heart of the platform, oa-lake is based on a distributed, columnar, in-memory SQL engine. Transformations are performed in flat SQL , without dependence on proprietary technology, with high scalability and incredible performance! Thanks to Parquet storage, oa-lake acts as an intelligent cache : loads are optimized and the load on sources is reduced. {oa-lake} : A lightweight architecture serving a major evolution of the IS {oa.lake} : A transparent layer: data lineage to govern the system from end to end oa-lake  is not a black box: it is anopen architecture, based on Parquet storageand adistributed SQL engine. Every data, every transformation is traceable. And this is what we are proposing by associating it with end-to-end data lineage , from operational sources to the dashboard cell, to build simple and powerful governance.  oa-lake puts control back where it belongs: in a single layer, which unifies  sources, relieves targets of their complexity, and creates simple and factual data governance. Let's talk about it?