6 min read

Tanna Solberg
October 26, 2020 – 9:31pm

October 29, 2020

Cloud technologies make it easier, faster, and more reliable to ingest, store, analyze, and share data sets that range in type and size. The cloud also provides strong tools for governance and security which enable organizations to move faster on analytics initiatives. Like many of our customers, we at Tableau wanted to realize these benefits. Having done the heavy lifting to move our data into the cloud, we now have the opportunity to reflect and share our migration story.

As we embarked on the journey of selecting and moving to a cloud-driven data platform from a conventional on-premises solution, we were in a unique position. With our mission to help people see and understand data, we’ve always encouraged employees to use Tableau to make faster, better decisions. Between our culture of democratizing data and rapid, significant growth, we consequently had servers running under people’s desks, powering data sources that were often in conflict. It also created a messy server environment where we struggled to maintain proper documentation, apply standard governance practices, and manage downstream data to avoid duplication. When it came time to migrate, this put pressure on analysts and strained resources.

Despite some of our unique circumstances, we know Tableau isn’t alone in facing some of these challenges—from deciding what and when to migrate to the cloud, to how to better govern self-service analytics and arrive at a single source of truth. We’re pleased to share our lessons learned so customers can make informed decisions along their own cloud journeys.

Our cloud evaluation measures

Because the cloud is now the preferred place for businesses to run their IT infrastructure, choosing to shift our enterprise analytics to a SaaS environment (Tableau Online) was a key first step. After that, we needed to carefully evaluate cloud platforms and choose the best solution for hosting our data. The top criteria we focused on during the evaluation phase were: 

  • Performance: The platform had to be highly performant to support ad-hoc analysis to high-volume, regular reporting across diverse use cases. We wanted fewer “knobs” to turn and an infrastructure that adapted to usage patterns, responded dynamically, and included automatic encryption.
  • Scale: We wanted scalable compute and storage that would adjust to changes in demand—whether we were in a busy time of closing financial books for the quarter or faced with quickly and unpredictably shifting needs—like an unexpected pandemic. Whatever we chose needed compute power that scaled to match our data workloads. 
  • Governance and security: We’re a data-driven organization, but because much of that  data wasn’t always effectively governed, we knew we were missing out on value that the data held.. Thus, we required technology that supported enterprise governance as well as the increased security that our growing business demands. 
  • Flexibility: We needed the ability to scale infrastructure up or down to meet performance and cost needs. We also wanted a cloud platform that matched Tableau’s handling of structured, unstructured, or semi-structured data types to increase performance across our variety of analytics use cases. 
  • Simplicity: Tableau sought a solution that was easy to use and manage across skill levels, including teams with seasoned engineers or teams without them that managed their data pipelines through Tableau Prep. If they quickly saw the benefit of the cloud architecture to streamline workflows and reduce their time to insight, it would help them focus on creating data context and support governance that enabled self-service—a win-win for all.
  • Cost-efficiency: A fixed database infrastructure can create large overhead costs. Knowing many companies purchase their data warehouse to meet the highest demand timeframes, we needed high performance and capacity, but not 24/7. That could cost Tableau millions of dollars of unused capacity.

Measurement and testing considerations

We needed to deploy at scale and account for diverse use cases as well as quickly get our people answers from their data to make important, in-the-moment decisions. After narrowing our choices, we followed that with testing to ensure the cloud solution performed as efficiently as we needed it to. We tested:

  • Dashboard load times; we tested more than 20,000 Tableau vizzes 
  • Data import speeds
  • Compute power
  • Extract refreshes
  • How fast the solution allows our London and Singapore data centers to access data that we have stored in our US-West-2a regional data center 

We advise similar testing for organizations like us, but we also suggest asking some other questions to guarantee the solution aligns with your top priorities and concerns:

  • What could the migration path look like from your current solution to the cloud? (For us, SQL Server to Snowflake)
  • What’s the learning curve like for data engineers—both for migration and afterward?
  • Is the cost structure of the cloud solution transparent, so you can somewhat accurately forecast/estimate your costs?
  • Will the solution lower administration and maintenance? 
  • How does the solution fit with your current development practices and methods, and what is the impact for processes that may have to change?
  • How will you handle authentication?
  • How will this solution fit with our larger vendor and partner ecosystem?

Tabeau’s choice: Snowflake

There isn’t a one-size-fits-all approach, and it’s worth exploring various cloud data platforms. We found that in prioritizing requirements and making careful, conscious choices of where we wouldn’t make any sacrifices, a few vendors rose to the top as our shortlist for evaluation. 

In our data-heavy, dynamic environment where needs and situations change on a dime, we found Snowflake met our needs and then some. It is feature-rich with a dynamic, collaborative environment that brings Tableau together—sales, marketing, finance, product development, and executives who must quickly make decisions for the health, safety, progress of the business. 

“This process had a transformational effect on my team, who spent years saying ‘no’ when we couldn’t meet analytics demands across Tableau,” explained Phillip Cheung, a product manager who helped drive the evaluation and testing process. “Now we can easily respond to any request for data in a way that fully supports self-service analytics with Tableau.” 

Cloud adoption, accelerated

With disruption on a global scale, the business landscape is changing like we’ve never experienced. Every organization, government agency, and individual has been impacted by COVID-19. We’re all leaning into data for answers and clarity to move ahead. And through these times of rapid change, the cloud has proven even more important than we thought.

As a result of the pandemic, organizations are accelerating and prioritizing cloud adoption and migration efforts. According to a recent IDC survey, almost 50 percent of technology decision makers expect to moderately or significantly increase demand for cloud computing as a result of the pandemic. Meredith Whalen, chief research officer, said, “A number of CIOs tell us their cloud migration investments paid off during the pandemic as they were able to easily scale up or down.” (Source: IDC. COVID-19 Brings New C-Suite Priorities, May 2020.)

We know that many of our customers are considering or already increasing their cloud investments. And we hope our lessons learned will help others gain useful perspective in moving to the cloud, and to ultimately grow more adaptive, resilient, and successful as they plan for the future. So stay tuned—as part of this continued series, we’ll also be sharing takeaways and experiences from IT and end users during key milestones as we moved our data and analytics to the cloud.