With the addition of IBM Data Gate for watsonx.data, the strategy for managing the data that will power your enterprise's artificial intelligence is becoming clearer than ever.
Everyone is moving as quickly as possible to take advantage of AI to modernize their workflows, to alter the way they interact with their customers and constituents, and to extract value from long-held data repositories. They are also dealing with a set of widely varying imperatives:
1. Maximize Data Safety
2. Ensure Data Accuracy and relevance
3. Return value from data as soon as possible
4. Accept data from new sources quickly
5. Control the cost of data
A few of these have traditionally been in direct conflict; data safety and accuracy typically meant adding data to well-modeled schemas in relational stores, often following elaborate processes for approval that took weeks or even months. And every byte of data was processed as if it was the most valuable data in the enterprise, whether it was queried once every ten years or once every ten seconds. Opening mainframe data to our distributed routines has had its challenges as well, with cost-conscious organizations reluctant to open mainframe systems to the additional workload.
IBM Data Gate for watsonx.data addresses that need directly, by providing efficient replication of Db2 for z/OS and even VSAM and IMS data directly into iceberg tables in watsonx.data. That capability moves IBM's lakehouse even closer to the center of the maelstrom of data curation needs required by artificial intelligence. The growing support for open file and table formats in watsonx.data means your data administrators can respond much more quickly to the needs of users and developers, without sacrificing things like data quality and understanding.
Think of it this way: AI initiatives are not diesel trains, slowly dragging massive payloads up to whatever speed their aging rail systems can support. Instead, they are massive rockets, blasting off with huge payloads in a sustained explosion that will keep them moving at a constant velocity for the entire trip. And any rocket scientist will tell you, early course mistakes can be much harder to correct than later ones. If your AI is built on bad data, then all your effort will only get you the wrong answer faster.
So, does object storage matter? Absolutely; it's essential to cost controls and availability.
Do multiple engines matter? Of course! Fit-for-purpose is vital to responding to the real needs of your users.
But for my money, the most important part of watsonx.data is that it makes open formats governable. It allows for centrally managed repositories of open data that can be rapidly examined for quality and enriched with metadata. And it offers those capabilities in a manner that eliminates shadow IT.
So the strategy is crystallizing-watsonx leaves no data behind, and provides a firm foundation for AI.
#watsonx.data------------------------------
Bradley Rowen
Principal WW Data and AI Tech Seller
IBM
------------------------------