Citus swirl

Chartbeat Delivers Historical Reporting & Content Intelligence to Digital Publishers with Citus Cloud

>10X
faster query responses
>2.6B
rows of data added per month

Delivering content intelligence to leading publishers worldwide

Today, both traditional and digital-native publishers are taking steps to become more data-savvy. By analyzing data on reader behaviors and preferences, publishers can optimize reader engagement and make smarter editorial decisions.

Chartbeat generates the insights these publishers need to connect with their readers and strengthen audience loyalty.

Partnering with over 50,000 media brands across 60+ countries, Chartbeat’s software and front-line tools help the world’s leading media companies understand, measure, and build business value from the attention earned by their content. Chartbeat’s customers range from the New York Times and Le Monde to National Geographic and Vice.

“To attract and retain readers, media publishers need to know which articles are most popular, which distribution method is the most effective, which mobile platforms are the most prevalent, and more,” says Wesley Chow, CTO, Chartbeat. “We strive to deliver those insights through dashboards and rich reports that are easy to use for editors and other content creators.”

Chartbeat dashboard
The Citus database powers Chartbeat’s historical dashboard, shown above

Searching for a database for a new historical analytics dashboard

Chartbeat had initially launched a real-time analytics solution for fast-paced breaking-news scenarios. Subsequently, the company decided to expand its offerings, building a historical analytics dashboard that could provide down-to-the-minute historical data about a publisher’s site and audience.

To support the new historical dashboard, Chartbeat needed a data store geared specifically for historical analytics.

“We had to take very fine-grained data from Redshift and a data warehousing system, and then produce aggregated versions of the data in an interactive dashboard,” says Chow. “The data store had to materialize certain analytical queries that are efficient in batch using column store but not fast enough for interactive use.”

Our goal was to shard on a per-customer basis to scale out and improve performance, but we didn’t want to spend our engineering time to shard at the application layer. The Citus database makes sharding simple: Citus has a multi-tenant data model that let us easily shard by customer—without having to spend time re-engineering our application.
Wesley Chow, CTO at Chartbeat

Selecting the Citus Cloud database for Chartbeat’s historical dashboard

While considering a homegrown solution, the Chartbeat team quickly gravitated toward the PostgreSQL database. “Several of our team members had strong Postgres backgrounds, so we decided to explore that route,” says Chow.

Open source was also a factor. “We wanted a database that would let us pull up the code and see how the thing works,” says Chow.

The need for multi-tenant sharding capabilities narrowed the selection. “We built out a proof of concept using Postgres, but didn't want to construct our own sharding layer, so we turned to Citus,” says Chow.

“Our goal was to shard on a per-customer basis to scale out and improve performance, but we didn’t want to spend our engineering time to shard at the application layer,” says Chow. “The Citus database makes sharding simple: Citus has a multi-tenant data model that let us easily shard by customer—without having to spend time re-engineering our application.”

Because Citus is a PostgreSQL extension rather than a fork, Chow recognized that Citus would do a better job of staying current with the newest Postgres releases compared with other databases that are forks of Postgres. And importantly, his team members would be able to work with Postgres, a database that was familiar to them.

As Chow’s team started its evaluation of Citus, team members found the setup to be fast. “We were able to get things going with Citus very quickly—much more quickly than we would have been able to with a homegrown solution,” says Chow.

Citus engineers helped speed the setup process. “We received a lot of help up front from Citus engineers,” says Chow. “They’ve been through a large number of installations and have more experience than our team. Being able to access people who know the internals of PostgreSQL well is a pretty big advantage.”

Through the evaluation of Citus, Chow’s team saw that Citus could deliver greater reliability than a homegrown solution. “When we run a `copy` statement to load data into the Citus database cluster, all either succeed or fail,” says Chow. “Whereas with a homegrown system, we might not have transactional guarantees, so we would wind up with an inconsistent import state. Citus makes it much easier to enforce correctness.”

Quick turnaround times from Citus engineers gave the Chartbeat team confidence to move forward with Citus. “When it was time to commit, we drew up SLAs with expected response times. And in our experience, Citus delivered even faster, better responses than we anticipated,” says Chow.

With Citus, we really don’t have to think much about our database. It’s Postgres, it scales out, it works, and when we need help, we can get it. Citus really is worry-free.
Wesley Chow, CTO at Chartbeat

Deploying Citus Cloud fast—without having to shard the application

Choosing the Citus Cloud database with its multi-tenant sharding capabilities enabled Chartbeat to get up and running quickly on AWS.

“If our developers had needed to engineer the application-level sharding themselves, that work would have pulled them away from other strategic projects,” says Chow. “With Citus, we were able to transition the new historical analytics dashboards to beta in just a couple months. We then turned on the new analytics dashboard for all our customers gradually over the next few months, with minimal fuss.”

In the past, it might have taken dozens of seconds or even minutes to materialize queries. With Citus, we can deliver sub-second latency, giving our clients a responsive, interactive experience and enabling them to make better decisions, faster.
Wesley Chow, CTO at Chartbeat

Powering an interactive dashboard whose data is growing by over 2.6 billion rows per month

With Citus, the team at Chartbeat has the performance they need for an interactive analytics dashboard whose historical data is growing by over 2.6B rows of data per month. Customer-facing analytics dashboards are different than in-house analytics tools: with customer-facing dashboards, you often have many customers querying the data at the same time—all of whom are expecting sub-second responses, even with complex queries.

“Dashboard interactivity is very important,” says Chow. “Every second a user has to wait to get insights, even if that request is crunching a lot of data, represents a potential disruption to our client’s workflow. In the past, it might have taken dozens of seconds or even minutes to materialize queries. With Citus, we can deliver sub-second latency, giving our clients a responsive, interactive experience and enabling them to make better decisions, faster.”

Staying focused on innovation with worry-free Postgres from Citus

By adopting Citus, Chow’s team has been able to stay focused on launching innovative, interactive analytics applications, rather than managing database infrastructure.

“With Citus, we really don’t have to think much about our database,” says Chow. “It’s Postgres, it scales out, it works, and when we need help, we can get it. Citus really is worry-free.”

Chartbeat + Citus Data Story Highlights

chartbeat.com

  • With Citus Cloud, Chartbeat launched a new historical analytics dashboard that generates deep content insights for digital publishers
  • Choosing a cloud database with multi-tenant sharding capabilities helped accelerate the Chartbeat launch and avoid time-consuming development work
  • By adopting Citus Cloud, Chartbeat was able to move operational risk from their code to the cloud team at Citus Data

With Citus, we really don’t have to think much about our database. It’s Postgres, it scales out, it works, and when we need help, we can get it. Citus really is worry-free.

Wesley Chow, CTO at Chartbeat

About Chartbeat

Chartbeat, the content intelligence platform for publishers, believes that today’s content creators need mission-critical insights to build a loyal audience. That’s why Chartbeat’s software goes beyond clicks and pageviews so content creators can understand what, within their content, is keeping people engaged in real time and across desktop, social, and mobile platforms. Partnered with over 50,000 media brands across 60+ countries, Chartbeat’s editorial tools and services help the world’s leading media companies understand, measure, and build business value from the attention earned by their content.

chartbeat.com

Recommended Next Steps