Multisource Analytics for Modern BI
Helping decision-makers navigate a complex, multisource data world
Interacting with billions of rows of data in seconds from a single source is exciting, but big data exploration gets really interesting when you work with data across multiple sources.
- Maybe you have consumer behavior data in Hadoop, but you’d like to enrich that data with some demographic information stored in an Oracle database and partner information stored in a cloud data warehouse.
- Or suppose you have a stream of sensor readings and you want to perform calculations across that real-time stream and compare them with historical metrics.
- Or you might have product reviews indexed in Elasticsearch that you’d like to correlate with product purchase history in an enterprise data warehouse.
All of these scenarios require an agile and flexible approach to multisource analytics to get the best insights.
Multisource Many Ways™ with Zoomdata
Data is everywhere, and it’s your job to make sense of it -- wherever it resides. Zoomdata offers practical approaches to multi-source analysis. They are multisource dashboards, Zoomdata Fusion, cross-source filtering, and keysets.
- Multisource Dashboards simply means that you can bring in as many visualizations as you like. Each one can load data from different back-end data sources.
- Zoomdata Fusion combines multiple data sources so that they appear as a single source to the business user. Fusion is a data source that IT, DBAs, and advanced users can define for themselves or others. When used in dashboards, Fusion issues federated queries and joins data stored in different systems.
- Cross-source filtering allows business users to apply common filters across charts that are populated by different data sources. Cross-source filtering simplifies and accelerates exploration when a dashboard contains data that is logically related but physically in different systems. The business user defines cross-source links as part of their data exploration experience.
- Keysets are an innovative way to perform multi-pass data analysis without requiring IT or developer intervention. Users simply create a “keyset” from one visualization and apply that keyset as a filter on other visualizations. Keysets can be applied to any visualization, and are usable across all data sources.
Few traditional BI vendors offer multi-pass and cross-source set analysis, and those that do require time-consuming involvement from DBAs or IT.
With Zoomdata, users enjoy fast, on-the-fly analysis across a wide variety of data, no matter where it resides.
Zoomdata Fusion
Zoomdata Fusion makes multiple data sources appear as a single source to the business user. Fusion can present data from relational and non-relational sources, including structured and unstructured sources. (No audio.)
Multisource Analysis with Cross-Source Filters
Without requiring any semantic modeling or advanced setup, users can rapidly explore and visualize the impact of applying common filter criteria across all elements on a dashboard, regardless of where the data comes from.
Multisource Analysis with Keysets
Keysets are ideal for multi-pass use cases, and for rapid data exploration and set analysis across a wide variety of data platforms. Keyset Analytics is especially powerful when related data sources are too big to join into a single data source, are streaming live, or are too sensitive to move into curated systems because of security concerns.
Why Multisource Matters
As markets change and technology evolves, so does our data. Our data accumulates faster and grows ever bigger. It will never slow down. We’ll never generate less data.
Putting this data to work requires a variety of data platforms.
Business requirements dictate that some data is so big that it needs to be managed in infinitely scalable systems like Hadoop. Near-real-time analysis is very doable for organizations that land data in modern fast data sinks. Some data isn’t needed for real-time analysis, but needs to be “hot” and return results super quickly. Search engine databases are flexible and allow qualitative analysis through free text search. Some data is stored on-prem, some is in the cloud, and some straddles both through a hybrid-cloud.
And, of course, all these data platforms come with their own tradeoffs, including cost. For all the advances we enjoy, data will persist in a wide variety of systems for a long time to come.
DBAs and IT professionals, want to learn more? Learn about pushdown processing
Traditional BI:
Beware "Single" and "Full Stack" Solutions
Be wary of traditional BI vendors that encourage you to combine and manage all your data into their proprietary “single stack” or “full stack” environment. Look for words like import, ingest, extract, and so on. What might sound like data centralization is far too often data duplication and “black box” data transformations. These are avoidable data governance and lineage problems.
For modern data sources, the problem is even worse.
- Data that’s too big to move simply won’t move.
- Streaming data loses its power when it’s batch imported into systems that are not optimized for running big analytic queries while new data is indexed in near real time.
- Copying large amounts of data to-and-from the cloud is time-consuming, unnecessarily network-intensive, and wastes resources.
Above all -- copying data out of one data source into another is inherently risky, and not always in the best interest of the business analyst.
Maintaining appropriate security privileges across all data extracts is difficult. Removing context such as columns and rows may be the right thing to do for some use cases, but it’s generally better to keep everything in place and hide or filter unauthorized data based on user privileges.
Then there’s the problem of data transformation. When the same data is transformed into multiple secondary databases, which is the source of the truth?
These problems are real-life and non-trivial. There are even products you can buy to track and manage this mess of data duplication. But instead of buying yet another IT product to manage copies of data all over the enterprise, we at Zoomdata encourage you to:
- Leave the data in place unless there's a really good reason to move it
- Lock down the data
- Set authorization privileges so trusted people can get what they need, whenever they need it
- Use Zoomdata Multisource Many Ways to explore relationships, even when related data is managed in different systems
