Case study

Bibblio

Bibblio offers a suite of tools that help digital publishers increase their audience and revenue without invasive advertising. The London-based startup uses natural language processing, recommendation algorithms and machine learning to suggest relevant content for any website. Over the past five years, Bibblio has grown rapidly, leading its development team to look for a replacement to their IBM Compose database engine. They wanted a more flexible solution that would perform well at scale without allowing costs to get out of control. After moving to Redis Enterprise, Bibblio cut its storage spending while maintaining low latency for content recommendations.

Breaking free from limitations imposed by IBM, AWS and Google

The company’s powerful solutions for content circulation, promotion and syndication are built around APIs that publishers use to manage their content onBibblio’s system. Bibblio runs natural language processing on the content text, applies metadata about interconnections between other content, and sends back recommendations about which related content a website should display within a certain article. All of this heavy lifting requires a highly performant database that can return recommendations to a customer’s website with negligible performance degradation.

Previously, Bibblio used IBM Compose to host its open source Redis data store, but wanted a more flexible platform with fewer limitations. One of the main driving factors was the development team needing the control to maintain their consistent delivery and low latency without breaking the bank as data volumes grew. After trying to use AWS ElastiCache, they grew frustrated with confusing billing models and various other constraints. Bibblio also explored using Redis on Google Cloud as an option, but realized it would require them to migrate other parts of their system to Google in order to meet their latency requirements.

“Every time we put additional data into our cluster, we ended up paying more with third-party hosted solutions, so we were looking for ways to take control and cap that spend as much as possible,” shared Michael Clarke, Head of Development for Bibblio. “We never had any trouble meeting application delivery expectations, but needed something that scaled the way we wanted it to without locking us into inflexible billing models.”

Requirements

Bibblio needed a flexible key-value and caching database that could meet its performance requirements for instant content recommendations while scaling cost-effectively

Redis Enterprise Benefits

  • Super-fast low latency
  • Minimal storage spend with increased capacity for growth
  • Scalable performance through a sharednothing cluster architecture
  • Zero operational hassle with 24×7 support

“Redis Enterprise is really quick and scalable, giving us the advantage of focusing on improving the quality of our recommendations to end users. Our data volumes are growing quickly, but Redis Enterprise helps us address our downstream delivery bottlenecks effectively. From a business perspective, it’s easier to focus on the big picture when we don’t have to be in the trenches fighting fires.”

Michael Clarke
Head of Development
Bibblio

Deploying Redis Enterprise for greater control and lower costs

Bibblio successfully moved off of the IBM Cloud and migrated its key-value and caching databases to Redis Enterprise, which was easy to configure and offered fixed payment plans with enough capacity to grow. Clarke remembers, “Moving to Redis Enterprise was the logical decision for us, having already experimented with open source Redis and built trust in the technology over time. I enjoy our conversations with the Redis Labs team, who are true allies and understand how critical Redis is to our business systems.”

Today, Bibblio uses Redis Enterprise to manage authentication-based tokens and cache various quasi real time analytics. The recommendation engine stores content interconnections in Sorted Sets – one of the fundamental data structures of Redis, and uses in-memory set operations to blend results upon final delivery of recommendations.

“Our goal is to deliver value to our commercial team, and keep our promises – we really make sure everything we develop is carefully orchestrated and fits into a larger plan so we can effectively provide the best possible solution to our customers.” said Clarke. “That’s why we’re happy with Redis Enterprise. It’s stable, doesn’t give us any issues and helps us stay focused on delivering value to our customers and commercial team.”

Download The Case Study

Architecting an instant experience that scales seamlessly

In order to fully benefit from the power of Redis Enterprise, Bibblio developed a simple, yet elegant architecture to run its recommendations database on AWS. The team uses the AWS CloudFormation configuration mechanism to deploy various infrastructure and topology, including spot instances, which reduces costs.

Redis Labs collaborated with Bibblio on a template for the Ansible application deployment tool, which runs automated scripts to install Redis on nodes – enabling the team to spin up another Redis cluster that looks identical to its production cluster in a matter of minutes when needed. All of this seamlessly integrates with Bibblio’s application, which is written in Clojure, ClojureScript, Node.js and Python.

Gaining freedom from the burdens of database administration

Since moving over to Redis Enterprise, Bibblio slashed storage spend and increased available capacity for future growth. On top of these benefits, the company’s development team enjoys Redis Enterprise’s dashboard interface that sums up all the key indicators they want to monitor. Since Redis Enterprise provides full automation of the day-to-day operations for their Redis databases – along with deep visibility into important metrics, triggers for auto-balancing clusters and seamless resharding – they can focus their energy on fine-tuning and evolving their core applications to support the business.