Tail: Good practices from the startup that migrated to Google Cloud in three days without shutting down services

About Tail

Tail is a data intelligence SaaS platform that brings the power of big data analytics and machine learning to every business. This startup helps brands in every industry optimize their marketing strategy by giving instant valuable insights through analysis of customer and prospect data as well as exclusive web-behavior and geo-behavior data. Tail is based in Sao Paulo, Brazil, and operates in Latin America and the U.S.

Industries: Technology
Location: Brazil

Tell us your challenge. We're here to help.

Contact us

After migrating in just three days, Tail was able to improve its infrastructure, streamline its IT professionals’ routine, and cut costs.

Results

  • Migrated its entire infrastructure (more than 300 servers) to Google Cloud in three days without shutting down services
  • Enabled to stop, review, and optimize the legacy infrastructure
  • Cut search times in the database from minutes to milliseconds
  • Made data processing and management easier by migrating the company’s HDFS to Cloud Storage
  • Increased agility lets the IT team unlock features and take more calls
  • Reduced latency for delivering scripts from Tail’s platform from 400 ms to 180 ms

Without shutting down its services, Tail migrated its infrastructure to Google Cloud in three days.

There are countless cases of businesses that are migrating their infrastructure to the cloud, but some companies also owe their very existence to it. Such is the case of Brazilian startup Tail, which helps with business marketing strategies through its own data intelligence SaaS platform since 2012. Since their business model involves scaling infrastructure during the day and the unpredictability of traffic, Tail would be practically unfeasible if the company depended on hiring servers and renting data centers.

The startup’s structure allows for a much faster delivery capability than other large players in the market. A typical project is usually implemented in up to 60 days, with no need for external consulting, providing customers with real-time behavioral data analytics of their consumers, smarter activations, instant database enriching, and GDPR compliance.

However, over the years, this structure’s evolution began to create bottlenecks for Tail’s team, in particular because the cloud solution the company had hired in 2012 did not have so many cloud automation services, which prompted the startup to begin developing some of them in house. Also, since the service’s Hadoop distributions did not support Tail’s software’s version, the team ended up creating its own Hadoop distributions, along with its own HDFS installation distributed among several HDs and servers. In other words, the company was handling even storage.

For seven years, Tail kept building its own services, but reached the point when its 10 IT professionals no longer had the time to review and refine processes. “We wanted to keep evolving, but couldn’t stop to check what had already been done, how to improve our legacy infrastructure, what to do with all the things we already had,” explains Fernando Babadopulos, Tail’s co-founder and CTO. Then, in 2019, they had an idea: what if the company migrated its infrastructure?

Train, learn, and then migrate

To verify that their idea could be put into practice, they needed to see if there was a solution that made sense both from a technological and a financial standpoint. First and foremost, the solution had to allow Tail to carry out the migration without having to shut down services provided to customers. After a series of analyses and talks between the technical team and the C-Suite, Tail found the ideal solution in Google Cloud.

Before starting, they had to handle an important issue: the IT team had very little knowledge of Google Cloud’s solutions. That is why they decided to become certified in Google Cloud. “The first certification I obtained was Data Engineer, and I can assure you it was essential for us to get the migration done,” says Babadopulos. Since the team would carry out the migration on its own, studying the platform’s tools, concepts, and naming conventions was decisive to make the process work.

In 20 days, the team was able to obtain seven Google Cloud certifications: Professional Cloud Architect, Professional Cloud Security Engineer, Professional Cloud Network Engineer and four Professional Data Engineer certifications. “It was nice because we forced ourselves to study, managed to get certified, and used all that knowledge to plan the migration,” says Tail’s CTO. That was the next step for the team.

As the migration process had to be “hot,” without shutting down the company’s services, planning was fundamental. “We planned for about a month, performing tests on particular pieces of the system. We have four different databases and each of them needed a different migration plan,” explains Babadopulos. With migration tests and pilots, the team managed to understand every step of the process and estimated that migrating the entire infrastructure would take two to three days.

They had to run a tight ship to avoid issues that would force them to pause the system, especially after reaching the point of no return, when nothing is synchronized with the previous service anymore. Once migrated, the infrastructure already had to be operational.

All that planning paid off. In three consecutive days working around the clock, they migrated over 300 servers. The team faced few setbacks during the process, none of which affected offering continued service to customers. The migration took place in December 2019, and its results were already being felt shortly afterwards.

A new routine for the IT team

The impacts of migration were felt in various ways by Tail’s IT team. First of all, there was the opportunity of taking a minute to review and optimize the entire legacy infrastructure. An example of this was the creation of cloud services. The startup managed to retire part of its software and adopted Google Cloud tools such as Cloud Load Balancing, Compute Engine for machine scaling, and BigQuery for searching and analytics in the database.

“In the past, we had a few issues with scalability and response times, and BigQuery took care of that very well. We ended up going from minutes to milliseconds for some searches,” says Babadopulos.

Tail also solved its storage issue. Whereas the team had to store large data volumes outside the VM for seven years, now HDs and servers are no longer necessary. The HDFS were migrated to Cloud Storage, where the startup can process and manage data more easily, without having to copy it from an external system.

As for access to infrastructure, the previous solution granted permissions to only two professionals. With Google Cloud, three team members now have access. This means not only an easier, more secure way to engage and work in this environment, but also an impact on the team’s daily routine, since now it can unlock features more quickly and attend to calls.

Other benefits mentioned by Tail’s CTO are the continued default data encryption at rest and in-transit (with the possibility of maintaining the encryption key outside the cloud), using Google’s high-speed network for information traffic, and a reduction from 400 ms to 180 ms in the latency for delivering the startup’s platform’s scripts with Cloud CDN.

Emmanuel Martins, Tail’s BizDev head, highlighted the migration as a new selling point for prospects, since companies regard Tail’s use of Google infrastructure as a positive sign for their own digital transformation.

“We are part of a story that is being told as we speak, with tools that our customers know and trust. With that, we gained scale, allowing us to do more prospecting, take part in more meetings and close more deals as a result.”

Emmanuel Martins

Recommendations for other startups

Moving to Google Cloud allowed Tail to optimize its infrastructure, its costs, and the IT team’s work routine. Thanks to that, now the company plans to expand usage of the tools as the projects evolve, like migrating other services to BigQuery.

For other startups that want to migrate to Google Cloud on their own, Babadopulos believes the takeaway from Tail’s experience is to be prepared and plan through certifications and migration tests, making sure to test everything on real databases, VMs, and services. “Migration tests helped us to get confident for the real migration,” he says.

Tell us your challenge. We're here to help.

Contact us

About Tail

Tail is a data intelligence SaaS platform that brings the power of big data analytics and machine learning to every business. This startup helps brands in every industry optimize their marketing strategy by giving instant valuable insights through analysis of customer and prospect data as well as exclusive web-behavior and geo-behavior data. Tail is based in Sao Paulo, Brazil, and operates in Latin America and the U.S.

Industries: Technology
Location: Brazil