Bagel Brands is the parent company for Einstein Bros. Bagels, Bruegger’s, Noah’s New York Bagels, and Manhattan Bagel. Headquartered in Denver, Colorado, the company serves 1,100+ domestic franchise and license locations supported by 16,000+ employees.

To drive its data analytics and data science initiatives, Bagel Brands chose Snowflake on Azure as its primary modern data platform, and has AWS instances as well to take advantage of Snowflake’s multi-cloud capabilities and utilize the best-of-breed tools and data sets. It retained a Resident Solutions Architect (RSA) through Snowflake’s Professional Services team to ensure a successful production implementation and increase time to value. 

Challenge: A Complicated Data Architecture

To align with internal and industry regulations, it is common practice to keep separate data environments for operational reporting and data science. The challenge with Bagel Brands’ previous data architecture was that data was physically separated. Data moved through many SQL Server systems, transforming each time, before getting to the analytics layer. This created data silos across the organization that led to limited visibility and lack of data governance, and strained its ability to make business decisions quickly and efficiently. 

This not only impacted the business’s goal to be a data-driven organization, but also more resources were needed to pay for storage costs and build more data pipelines to shuffle data across silos. For example, Databricks was used to process ETL workloads, moving data from SQL Server into its Snowflake environment, adding additional monthly cost to its operations. And because SQL Server cannot be turned off, Bagel Brands was spending tens of thousands of dollars paying for 24/7 compute every month, even when idle.

Solution: Pulling Stakeholders Together to Maximize Snowflake with RSA

Jessica Lee, Bagel Brands’ Director of Data Science and Analytics, and Director of Data Architecture Engineering, Anu Vadrevu, decided to pursue a new, simplified data architecture with Snowflake as Bagel Brands’ modern data platform. It became a collaboration between data science and analytics teams that were looking to derive business value, supported by the IT teams that were making the technical decisions. 

“With this new direction, I wanted a Snowflake RSA because it needed to be done right from the beginning. We have a lean team and I’d much rather co-design with the RSA to make sure Snowflake is set up quickly and efficiently for our organization. That means optimizing everything from cost, compute, modeling, architecture, and more,” Lee said.

To avoid creating extra data silos from intermediate SQL servers, the raw data is pulled directly from data sources into the Azure data lake storage landing zone. Using Snowpipe, data is then loaded into Bagel Brands Insights’ Snowflake Data Cloud and external data from Snowflake Data Marketplace can be directly brought in here. With a single source of truth to manage and govern, users can consume trusted data through tools such as Power BI and Thoughtspot. 

Results: Simplified Architecture with Performance and Cost Savings

Snowflake’s Data Cloud Deployment Framework is founded on the key principles of: 

  • Mitigating data silos and securely governing enterprise and local data assets
  • Enabling federated development of data assets across business entities
  • Supporting corporate separateness to satisfy internal and regulatory requirements
  • Providing repeatable patterns for agility, extensibility, and ease of administration

With strategic advising from Scott Redding, its Snowflake RSA, Bagel Brands implemented Snowflake in a systematic way. Firstly, by removing SQL Servers and simplifying the data architecture, they didn’t need to physically keep data environments separate anymore. Snowflake can create a near-unlimited number of databases and support everyone with scalable compute. Data science and operational reporting teams can have their own databases that are logically separated and can be easily used across databases without moving or copying the data. With Snowflake, there are no data silos; everything is in one place with the ability to securely govern the data.

“In addition to the Snowflake compression, the cost savings we’ve had from removing SQL Server allows us the storage to start pulling in data from throughout Bagel Brands. We can pull everything in and dive deeper, allowing data scientists to review answers to insights the business may not have considered yet.”

 —Jessica Lee, Director of Data Science and Analytics, Bagel Brands

Secondly was replacing its Databricks deployment for ETL workloads with the existing ADF and ADLS setup, because Snowpipe data loads are a fraction of the cost. From a performance perspective, the introduction of Snowpipe and its instant triggers has reduced the lag caused by time-based triggers.

Conclusion

Beyond the project-based advising, Bagel Brands has collected the learnings from its RSA and shared them across the organization. “Scott has relevant industry knowledge of the retail sector, and that’s really helped us to be able to learn from his experiences and avoid pitfalls early on,” Lee said.

“It’s incredible to be able to say, ‘This is our vision and what we’re trying to accomplish,’ and then our RSA works with us to help realize it.”

 —Jessica Lee, Director of Data Science and Analytics, Bagel Brands