Doctrine cuts the cost of its research platform in half with AWS Graviton, without degrading performance
Thanks to its partnership with AWS since its inception in 2016, Doctrine has established itself as the leading legal intelligence platform. Performance gains linked to a reduction in costs have enabled them to focus on their core business: putting artificial intelligence at the service of legal professionals.
Doctrine cuts the cost of its research platform in half with AWS Graviton, without degrading performance
AI mobilized for lawyers and legal professionals
Doctrine's mission is to centralize legal information to make it understandable and relevant. “The quantity of legal documents which are put online increases year on year, and the documentary databases don’t list all of these documents, so lawyers and legal professionals used to have to search for them manually”, explains Jérémie Thomassey.
Doctrine aggregates this content and makes it available to various parties. They suggest content (court decisions, legal texts, parliamentary documents or even third-party comments freely accessible on the web) that is deemed relevant to their activity and allow users to share their own. The goal: building the first legal intelligence platform.
When it launched in 2016, Doctrine turned without hesitation to AWS for its hosting needs: The LegalTech startup was equally convinced by both the benefits linked to the performance of AWS services and the support from the teams made available to them.
A more efficient and less expensive platform thanks to AWS
Doctrine hosts the vast majority of its platform on AWS: 3 clusters for the web component along with 1 cluster for the data component related to the acquisition of data sites from which Doctrine aggregates the various content. Alongside this, LegalTech also indexes the content of all decisions and all laws on their Elasticsearch cluster, which includes 12 machines and 1.5 terabytes of data.
But that’s not all: Doctrine also regularly uses the AWS cost tracking platform. In this way, they have been able to reduce their AWS costs by 50% since March 2020 by configuring AWS alerts: as soon as a service over-consumes, they receive a notification so they can adjust the activity. These also allow developers to be more aware of the impact of their actions, and they can monitor them on a daily basis.
“In terms of performance, AWS Graviton instances are a real advantage: they enabled us to reduce the costs of these clusters by 50%, which represented 1/4 of our AWS budget,” says Jérémie Thomassey.
AWS as a partner on all fronts
In addition to the technical and financial benefits afforded by the use of AWS services, Doctrine particularly appreciates the unwavering support received from AWS: “Having a dedicated developer at AWS who helps us with any issues is a real plus, and if needed we can also contact experts on more technical issues or as part of training on specific points”, explains the infrastructure manager at Doctrine.
The other advantage of AWS according to Doctrine is their ability to show initiative, in particular by proposing improvement projects as was the case for AWS Graviton in February 2021. More recently, the LegalTech worked on setting up Amazon SageMaker, which allows notebooks to be launched in the cloud and extended on the fly, in as little as one week. Doctrine was thus able to have access to particularly powerful machines with ease, and at a lower cost for the data scientists, since they are switched off once the tasks have been executed, thanks to automatic mechanisms from AWS Lambda.
“We are currently planning the transition from our data cluster to Amazon Elastic Compute Cloud (Amazon EC2) Spot Instances, which are much cheaper and more suitable in terms of availability. Here, the flexibility allowed by AWS is another concrete and proven benefit,” states Jérémie Thomassey.
Setting ambitious goals – and achieving them
Ultimate guarantee of the quality of the relationship between AWS and Doctrine? AWS credits equivalent to the value of their Elasticsearch cluster offered in May 2021 in order to set up a pre-tested POC. The goal: to validate the possibility of running Elasticsearch on AWS Graviton instances, test their reliability, and validate the assumptions of technical and financial gains - all without impacting service, by duplicating their infrastructures to carry out the tests in shadow mode. Once the massive amount of data was indexed, within a week of going live, they were able to validate the performance - seeing a slight improvement in the process.
In the words of Jérémie Thomassey, the great success of their relationship over the last five years is the implementation of a hyper-ambitious internal SLO (service-level objective) of 99.99% availability: “The site is permitted 5 minutes of downtime per month, which is achieved almost every single time”, he concludes.
In view of the more than satisfactory results related to the use of AWS cloud services, Doctrine is already working on the launch of their next projects carried out with the support of AWS.
About Doctrine
Created in 2016, Doctrine is the first legal intelligence platform that analyzes, contextualizes and maps all legal information (court decisions, legal texts, doctrine, parliamentary documents, etc.) using artificial intelligence. Based in Paris, the company currently has more than 6,500 lawyer and legal professional users, as well as one million individuals who visit the Doctrine site each month.
Summary
- 50% billing savings thanks to optimization via AWS
- Significantly better cluster performance despite cost reduction
- Ambitious internal SLO of 99.99%, with 5 mins downtime per month
AWS Graviton
AWS Graviton processors are custom built by Amazon Web Services using 64-bit Arm Neoverse cores to deliver the best price performance for your cloud workloads running in Amazon EC2.
Amazon EC2 Spot Instances
Amazon EC2 Spot Instances let you take advantage of unused EC2 capacity in the AWS cloud. Spot Instances are available at up to a 90% discount compared to On-Demand prices.
Amazon Elasticsearch Service
Amazon Elasticsearch Service is a fully managed service that makes it easy for you to deploy, secure, and run Elasticsearch cost effectively at scale.
Amazon SageMaker
Amazon SageMaker helps data scientists and developers quickly prepare, build, train, and deploy high-quality machine learning (ML) models by bringing together a broad set of features designed specifically for ML.