The journey to agile analytics

  • 28 March 2022
  • 0 replies

Userlevel 5

This content, written by Nouras Haddad, was initially posted in Looker Blog on Mar 21, 2017. The content is subject to limited support.

The traditional approach to analytics is far from agile. Before you can even ask a question you need to first extract the necessary data and store it, which involves at least three different additional parties: The ETL Team, DBAs, and the BI team.

And should you need to ask a slightly different question, it’s back to the drawing board.

This process and its inherent latency causes two related problems: First there’s the obvious bottleneck around how quickly you can iterate and ask new questions. Second, in order to get around this bottleneck, you and your colleagues may be extracting small subsets of the data in order to perform quick analyses, rather than waiting for lengthy processes to conclude. This results in data chaos, since multiple stakeholders are looking at slightly different pieces of the puzzle and applying different definitions. Unfortunately, it’s too common for this to produce conflicting results for the same KPI.

A crystal ball for your data

At Looker, we deliver a completely different scenario. One in which the business user simply has a governed window into the data, and can ask virtually any question of the data and get an immediate response they can trust. Finally, truly agile analytics.

How it’s possible

Looker doesn’t require you to move your data before you can analyze it. Instead, Looker’s data modeling language, LookML is built on top of SQL, enabling you to extract business logic from your analysts’ heads and into software. Because all users access the same data through a common governed lens, all users will get the same answer.

Also, Looker is partnered with Denodo, a company with whom we share a similar credo. The Denodo Platform integrates data sources, and like Looker, Denodo doesn’t require you to move your data. The Denodo Platform uses data virtualization to create secure views of the data, as needed, in real time, across multiple different sources. This is an alternative to ETL that allows you to interact with your data sources without physically centralizing it. This is very useful for quickly accessing new data sources for ad hoc exploration.

Think of the Denodo Platform as a virtual data warehouse. Architecturally, the data comes up into the Denodo Platform from multiple sources, including SaaS applications, traditional databases, business planning tools, or Web analytics.

Next, the data gets passed to the Looker data modeling layer, where it is standardized, before being delivered to Web interfaces, embedded in various scripts, packaged for scheduled delivery, or served via REST APIs, ultimately in the language understood by business users.

This is agile analytics at work, and as far as the business users know, they’re simply asking questions and receiving answers.

If you’d like to learn more about using Looker + Denodo, check out my presentation at the , Wednesday March 29th in the Americas, and Thursday March 30th in EMEA and APAC.

0 replies

Be the first to reply!