How promoted a prototype
of their solution to the cloud
Galileo, “The Data API for ML” is required to promote a prototype of their solution to the cloud. A set of AWS services was chosen to work together to integrate Java code, a python library and a demo website written in javascript with a backend in Java.
About Galileo
Galileo is “The Data API for ML”. They provide an abstraction layer for ML models using a python library. They provide this service to development companies interested in creating machine learning products without actually writing the models themselves.
Galileo Challenges
The customer product was in a very early stage. Galileo technically validated their product in local, small hadoop instances. Their challenge was to take that prototype and make it production-ready for its future customers in a reliable and cost-effective platform.
As technical challenges we can mention that they need to provide access to jupyter notebooks which can connect to a hadoop cluster with the Galileo library installed, and that the core of the library is written in Java (although is meant to be used from python).
Why AWS
AWS was chosen because we can deploy Galileo library quickly on an EMR cluster, deploy the core components as containers in EKS and have all the system to scale as needed. Also, having load balancing capabilities contribute to make the implementation simpler.
Partner Solution
The client required to solve different problems:
-
The libraries were not containerized. The first job was to create Dockerfiles to generate images and upload them to EMR so we can run them in EKS.
-
Migrate a Cassandra database which is an integral part of the library and the demo site to Amazon Keyspaces.
-
Create an Amazon EMR cluster for running demos and giving notebooks to potential customers. Here we carefully crafted Policies and Roles so access to EMR notebooks can be granted to customers without endangering the implementation and other customer’s data.
-
The customer now is able to give their potential customers a set of credentials to log in into EMR notebooks, and they can test the Galileo library with the full power of EMR and without security risks.
An architecture diagram follows:
Takeaways
The most important benefit for the customer is that they can now do demonstrations to potential customers with a Live system, and they don’t need to install anything locally. This will allow Galileo to start selling their services and platform.
Our approach
Teracloud is a fast-growing AWS Advanced Tier Services Partner created by certified cloud experts, with seasoned experience in migrating and deploying startups, enterprises, and everything in between to the cloud. We've worked with companies from many industries such as airlines, healthcare, education, and e-commerce designing, implementing, and managing Cloud workloads with HA architecture under 99.9% uptime SLA and PCI/HIPAA compliance requirements.
Our expertise ranges from designing state-of-the-art cloud architecture to deploying and managing cloud environments through Managed Services including AWS Consulting, Architects, DevOps, DevSecOps, MLOps, 24/7 Support, Automation, Disaster Recovery, and Security. We provide secure, scalable, and highly available solutions for our clients.
Get in touch
Take your business to the next level.
