Nexthink achieves less than 2ms latency and zero outages

THEIR CHALLENGE

Maximizing value from a critical investment

For large corporate enterprises, applications used by thousands of employees are major investments that must both enable employees and deliver results. Ensuring that staff use enterprise software to its fullest extent is critical. 

Adopt is a digital adoption platform operated by Swiss digital employee experience (DEX) management specialist Nexthink. It acts as an overlay to leading enterprise software, using embedded support to provide in-app guidance and personalized training that helps end users to navigate and learn complex applications and processes. The result is increased adoption, allowing staff to work more productively and deliver the necessary ROI. 

“Adopt guides employees through unfamiliar applications so they can understand and use them in the most efficient way possible,” explained Ladislav Gažo, Senior Director at Nexthink.

To add Adopt to its portfolio, Nexthink acquired its original developer, digital adoption platform AppLearn. It was a shrewd move, but not all of AppLearn’s technology stack, which included Couchbase and Elasticsearch, delivered the holistic performance that Nexthink was looking for. For Nexthink, there was a clear opportunity to simplify the environment, and in doing so, cut costs, increase ease of use, and significantly enhance AppLearn’s performance. 

“Our primary requirement for Adopt is that response times must be fast,” explained Matúš Bartko, Software Architect at Nexthink. “We don’t want users to open an application and see nothing happening for several seconds. That would be very, very disruptive.”

The previous setup often resulted in response times exceeding 100 milliseconds, putting pressure on the system. The system also had to handle high-frequency, concurrent partial updates to single-user objects from multiple sources; it was a pattern that was difficult to support efficiently with a complex mix of separate databases.

There was also an operational burden. The previous databases weren’t managed services, so Nexthink’s internal teams had to maintain them manually. This led to stability issues and, in some cases, outages. 

“Having a managed service was a secondary criterion, but still an important one. We didn’t necessarily want to maintain the solution on our own, so we opted to switch,” Bartko said. “We saw that MongoDB Atlas would not only relieve us of that maintenance responsibility, it would also provide better performance and enhance our development process. It wasn’t a difficult decision.”

OUR SOLUTION

Broad compatibility enables integration and cuts sprawl

As AWS is Nexthink’s key cloud partner, the business had taken time to assess native solutions such as Aurora and OpenSearch. 

“We’ve had experience with other databases too, and we considered one extensively,” Bartko added. “But it didn’t have the querying capabilities we were looking for, and we would have to use code logic instead of working directly through the database.”

MongoDB Atlas’s read-optimized model was also an important consideration. Adopt’s architecture reads data from multiple sources, such as user activity, external systems, and user profiles. This is then processed and consolidated into single documents for rapid access.

“We are potentially talking about gigabytes, maybe even thousands of gigabytes of data, so it’s not something we were keen to put somewhere in the cache,” said Bartko. “But we still needed quick access, and MongoDB’s document model allows that.”

Nexthink used MongoDB Atlas’ Terraform provider to integrate with its primary technology stack and also integrated with Datadog for monitoring, ensuring a seamless fit with its existing AWS ecosystem.

“MongoDB Atlas came with a predefined dashboard to monitor that cluster,” said Bartko. “We are using the AWS Identity and Access Management (IAM) tool, which is also compatible with MongoDB Atlas, and we’ve been able to map permissions with it too. Not having to introduce new technologies in that area is a major bonus.”

Also, MongoDB Atlas adds value with its ability to set locations for individual projects, which plays a key role in meeting growing corporate compliance requirements around data residency.   

“We’re active across the globe in 14 different production regions, rising to 25 if you count development environments,” said Gažo. “With MongoDB, we can map specific locations and deploy everything automatically.”

OUTCOME

A solution that is several orders of magnitude faster

For Nexthink, the transition to MongoDB Atlas has been a successful strategic move that simplified its technology stack and solved a specific, high-complexity engineering bottleneck: handling massive concurrent writes without sacrificing read speed. 

It has also given Nexthink the power to ensure that Adopt functions responsively and effectively, and has subsequently guided tens of thousands of users through learning journeys successfully. Performance has improved drastically, with latency for data retrieval dropping by 98% from over 100 milliseconds with its legacy tech stack to below 2 milliseconds with MongoDB Atlas. 

“Response times are several orders of magnitude faster with MongoDB Atlas,” said Bartko. “It takes so much of the pressure off the rest of the system that’s waiting for that data.”

MongoDB Atlas integrates seamlessly with Nexthink’s main programming languages, giving developers a fast, efficient working experience while avoiding technical debt and reducing overall costs.

“MongoDB Atlas has connectors for TypeScript, Python, and Java,” added Bartko. “That library of tools is something we’re leveraging more and more often.”

MongoDB Atlas also provides Nexthink with the stability and foundation to explore new, more advanced features that could further enhance Adopt’s offering. These include using Vector Search to scour files and empower AI-driven chatbots to answer user queries, while augmenting the general knowledge with company-specific documents (RAG), to increase accuracy and reliability.

“We can vectorize documents that a customer might have as a part of their internal documentation,” says Bartko. “Salesforce, for example, has an extended list of PDFs on how to use its solutions. We want to allow people to access that through a chat interface where they can ask questions. We’re currently prototyping that with MongoDB.”

More importantly, MongoDB Atlas’s managed service has eliminated outages and reduced significant strain on Nexthink’s internal teams, who can now focus on delivering outstanding products and experiences.

“With MongoDB Atlas, we’re solid. We can sleep well,” said Gažo. “That’s a key differentiator for us.”