Dedact: Building a personalized learning platform

About Dedact

Netherlands-based Dedact develops digital learning tools that make personalized learning possible, leading to better, more efficient education.

Industries: Education
Location: The Netherlands

With Google Cloud Load Balancing and Google Kubernetes Engine, Dedact built a resilient infrastructure that can withstand a barrage of DDOS attacks and continue its 100% year-over-year growth rate.

Google Cloud Results

  • Secures its platform from a barrage of DDOS attacks with Google Cloud Load Balancing
  • Maintains a 100% year-over-year customer growth rate, with no compromise in performance
  • Cuts costs with autoscaling features in Google Kubernetes Engine
  • Enables Dedact to deliver smooth performance to 50,000 students during school hours and scale back down in the evening

Eliminates outages due to DDOS attacks

Netherlands-based startup Dedact aims to improve education by personalizing the learning process for school students and teachers. With its digital platform, Learnbeat, students can read textbooks, take tests, or do exercises which are marked instantly, lightening the administrative burden for teachers as well as providing immediate insight into how the class is performing. “We make the feedback loop very quick,” says Dedact CTO Frank van Rest. “It means teachers can quickly adapt to the needs of their class.”

“We needed to find a solution as soon as possible, so we started talking with different cloud providers. Google was very quick with its reaction and had very insightful, specific ideas about how to deal with the DDOS attacks. It was obvious that Google would be a great partner for us.”

Frank van Rest, CTO, Dedact

What started as a pilot with a handful of classes in 2010 has become a platform serving over 50,000 students in the Netherlands and has expanded to primary schools in Italy. In February, 2016, however, Dedact suffered the first of what would become a series of distributed denial of service (DDOS) attacks, causing outages across its infrastructure and disrupting school lessons. With its existing infrastructure provider already at its limit, Dedact needed to find a robust alternative and find it quickly. To do that, it turned to Google.

“We needed to find a solution as soon as possible, so we started talking with different cloud providers,” says Frank. “Google was very quick with its reaction and had very insightful, specific ideas about how to deal with the DDOS attacks. It was obvious that Google would be a great partner for us.”

Lasting protection

Creating a cloud-based digital publishing platform like Learnbeat requires an enormous amount of computing power. As well as displaying digital textbooks, Learnbeat takes in exercise answers, marks them, and feeds the results back to teachers in a very short amount of time. As Dedact grew, it developed profiles of the students and used them to make adaptive learning algorithms. This meant that Learnbeat could automatically raise or lower the difficulty of questions to keep each individual student in a sweet spot of learning.

To provide this kind of capability, Dedact needs to collect, store, analyze, and act on hundreds of thousands of data points every day. Doubling its customer base every school year, by 2016 the company’s growth rate was already starting to strain its existing infrastructure. When the DDOS attacks started causing outages, the company knew it was time to choose a new solution.

Dedact chose Google Cloud Platform for its new infrastructure, thanks to a rapid and effective response from Google representatives. Within three months, Dedact had to migrate its architecture over to Google Cloud Platform before the new school year started in late August, with double the number of users. In addition, Dedact rewrote its architecture to run on clusters to take advantage of clusters runing with Google Kubernetes Engine, and to take full advantage of the flexibility and powerful autoscaling features of Kubernetes.

“Google Kubernetes Engine gave us more possibilities to use different programming languages and operating systems than we had before. It also made operations very easy. We just have software engineers. We don’t need a systems engineer.”

Frank van Rest, CTO, Dedact

To deal with the DDOS attacks, the company made extensive use of Google Cloud Load Balancing to protect its internal network from the effects of heavy external traffic. Dedact used Google Cloud SQL to manage its huge database, and Google BigQuery and Google Cloud Datalab to interrogate the data for insight.

With Google Compute Engine instances, Dedact ran its own secure, internal Gitlab, to store code and continuously integrate, test, and deliver rolling updates. Meanwhile, the company used Google App Engine as a separate infrastructure for its social platform, allowing teachers to share material that could generate heavy traffic without affecting the main Learnbeat platform.

“Google Kubernetes Engine gave us more possibilities to use different programming languages and operating systems than we had before. It also made operations very easy,” says Frank. “We just have software engineers. We don’t need a systems engineer.”

Resilient, scalable, evolving

Since the migration, Dedact has not had a single infrastructure outage due to DDOS attacks, delivering a consistently smooth service to its users. Scalability on Google Kubernetes Engine allowed Dedact to easily double its number of users during the migration period and then double them again the following school year. Learnbeat currently handles 500,000 data points every day without compromising on performance. To keep costs down, Dedact worked closely with the Google team to help ensure the best possible autoscaling solution to handle the heavy workloads during classroom hours and scale down in the evenings and weekends.

“It’s great to see Google innovating the whole time we’ve been with them and we know that we can benefit from those innovations. We’re very confident that they can help us grow in the years to come.”

Frank van Rest, CTO, Dedact

Not content with building a bigger, more resilient learning platform, Dedact is constantly working to improve Learnbeat and add new features. Until now, Learnbeat exercises for students have used closed questions or multiple choice to accommodate the limits of a machine-based marking system. Dedact is currently testing a way to assess and score essay-style questions using models trained on teachers’ previous marking data. To do that, Dedact is working with TensorFlow, Google Cloud AI machine learning and Google Cloud Pub/Sub to build a model that can give an accurate, useful assessment of students’ answers for a wider variety of subjects.

Meanwhile, as the company expands and teams up with traditional publishers who want to digitize their products with Learnbeat, Dedact is looking to work with Google Data Studio to easily visualize data for its partners, helping to provide them with insight and improve their own products. As Dedact continues to expand and improve its platform, it can rely on Google to keep providing the right tools to help it deliver the best learning experience it can deliver.

“It’s great to see Google innovating the whole time we’ve been with them and we know that we can benefit from those innovations,” says Frank. “We’re very confident that they can help us grow in the years to come.”

About Dedact

Netherlands-based Dedact develops digital learning tools that make personalized learning possible, leading to better, more efficient education.

Industries: Education
Location: The Netherlands
Google Cloud Platform logo

12 Months FREE TRIAL

Try Kubernetes Engine, BigQuery, and other Cloud Platform products with $300 in free credit and 12 months.

TRY IT FREE
Google Cloud Platform logo

12 Months FREE TRIAL

Try Kubernetes Engine, BigQuery, and other Cloud Platform products with $300 in free credit and 12 months.

TRY IT FREE