Databricks’ new data lakehouse aims at media, entertainment sector

After launching industry-specific data lakehouses for the retail, financial services and healthcare sectors over the past three months, Databricks is releasing a solution targeting the media and entertainment (M&E) sector.

The M&E Data Lakehouse is now generally available and has industry-specific features the company calls accelerators, including real-time personalization, Steve Sobel, the company’s global communications director, said in a blog post.

“The idea behind these so-called accelerators is to provide pre-packaged analytics and use-case capabilities to ultimately accelerate delivery and time-to-value for customers,” said Doug Henschen, senior analyst at Constellation Research.

“One can imagine that the general purpose version of Databricks Lakehouse gives the company 80% of what it needs to put its data to productive use to drive business insights and enterprise data science. The idea of ​​the industry-specific version of Lakehouse is to get customers in certain industries say 90% of the way to being productive with their data,” said Henschen.

The other 10% represents the effort involved in initial deployment, data loading, configuration, and setting up management tasks and custom analytics, Henschen said.

The Data Lakehouse is a relatively new data architecture concept, first championed by Cloudera, which, unlike the concepts for, offers both storage and analytics capabilities as part of the same solution data lake and data warehouse store data in native format or structured data, often in SQL format.

Some of the focused solutions that are part of Databricks’ new M&E Lakehouse include recommendation engines, a CLV (customer lifetime value) module, a streaming quality of service module, and game toxicity detection.

While recommendation engines help create more personalized experiences for consumers with AI-powered content recommendations that drive engagement and monetization opportunities, the CLV module identifies valuable customers with models focused on spending patterns to help businesses retain users and get better ones Marketing investments to businesses said. The recommendations also include suggestions for product development decisions.

“The most effective recommendation engines are very specific to industries and use cases. They require specific data inputs, models, algorithms and provide very specific recommendations. Delivering accurate, reliable recommendations is no easy task, so accelerators can be helpful starting points for businesses,” said Henschen.

The new Data Lakehouse’s streaming quality of service and gaming toxicity detection capabilities are very case-specific services. While the Streaming Quality of Service, as the name suggests, analyzes both streaming and batch data to ensure users are served optimal, tailored content, the gaming-specific service uses natural language processing for real-time detection of toxic Language to ensure optimal quality gaming experience for users.

Partner solutions to increase functionality and acceptance

As with other data lake and data warehouse providers – such as Snowflake, which was also on an industry-focused solution release tour – Databricks also wants to offer its customers more functionality by collaborating with other companies, which in turn should promote the acceptance of its new Lakehouse solution.

“Partnerships can be time-saving for customers as long as they adopt time-saving, pre-built integrations between partner platforms and solutions. Typically, such partnerships start with the most popular solutions in a specific industry or with deeper integrations with partners already established in a specific industry. The more partnerships there are, the better for the solution provider,” said Henschen.
Some of the partnerships within M&E’s Lakehouse solution include the company’s strategic relationships with AWS, Cognizant, Lovelytics, Labelbox and Fivetran.

While the partnership with AWS focuses on bringing more data and analytics capabilities to the M&E sector, the partnership with Cognizant aims to maintain video quality for customers.

Cognizant’s solution combines telemetry data with artificial intelligence and machine learning to quickly identify and fix video quality issues in real time to resolve issues such as playback errors, delayed time to first frame, or a rejection issue, the company said.

The company’s collaboration with Lovelytics focuses on baseball. As part of the solution, baseball team managers can optimize strategy for a game by using predictive analytics via artificial intelligence to forecast performance.

The solution also uses biomechanical indicators to signal and prevent potential player injuries, the company said.

The joint solution with Labelbox is aimed at media companies and is intended to help companies derive more value from unstructured data.

Databricks has partnered with Fivtran to offer a data integration service that it claims can ingest data from over 180 sources, including operational, advertising and marketing technology solutions.