E4 Data Science ready-to-go offering
Urania is an end-to-end solution for large-scale Big Data Analytics and Artificial Intelligence orchestrated with Kubernetes, including leading distributed data processing systems and popular open-source frameworks for Machine Learning and Deep Learning.
It provides interactive computing services in Jupyter Notebook technology that interface with distributed Data Processing systems and includes a cloud-native batch scheduler ideal for launching distributed Data Processing jobs on multiple nodes, in multi-user scenarios.
Engineered, not assembled
Urania is the enabling solution for high-performance, scalable Data Science workloads.
Featuring an intuitive approach, it is easy to manage and provides the flexibility of state-of-the-art systems.
Urania is an open solution that is updated every six months to ensure cutting-edge, uncompromising use across all Data Science domains.
Urania approach
Urania is the cloud-native Data Science solution designed by E4 Analytics, which consists of two main modules: the infrastructure (Kaptain multi-node) that offers high-performance container orchestration services, and the platform (E4DS-PLATFORM) that enables Data Science specific services and features that can be used in both interactive and batch modes.
CLOUD-NATIVE
Urania is the end-to-end solution for large-scale Big Data Analytics and Artificial Intelligence orchestrated with Kubernetes, including leading distributed data processing systems and popular open-source frameworks for Machine Learning and Deep Learning
INTERACTIVE
Urania provides multi-user services for interactive computing in Jupyter Notebook technology that include support for popular open-source Data Science frameworks. These services are configured to use the underlying Data Processing systems
POWERFUL
Urania Includes a cloud-native batch scheduler designed for number crunching workloads, which are converted to Kubernetes workloads and submitted to batch queues to optimize throughput of results
FUTURE-PROOF
The world of Data Science is constantly and rapidly evolving. Through the subscription of E4 Analytics services, Urania’s software stack is periodically updated and enriched with the most innovative proposals from the open-source world
Designed to give the best: always
“Data is the new oil” (Clive Humby, 2006), but to be transformed from raw data to a source of value they require a sophisticated refining process: the techniques of Big Data Analytics and Data Modeling with Machine Learning.
With Data Analytics based on Machine Learning, the most innovative organizations are able to improve their internal processes, grow their product portfolios, enrich their customer services, optimize their supply chains, lower their operating costs, and more.
But effective results require a solution that provides high performance, scalability, ease of use, inherent flexibility and modern cloud-native technologies.
Urania brings all these features together and is designed to provide the user with a solution that is flexible, always up-to-date and adaptable to the new, diverse and growing needs of the organization using it, regardless of the specific field in which it operates.
Discover all the benefits
Solution Layout
Technical features
Urania consists of two main modules: the infrastructure (Kaptain multi node) offering high-performance container orchestration services, and the platform (E4DS-PLATFORM) enabling Data Science-specific services and functionalities.
Kaptain multi node
Kaptain is the multi node Kubernetes solution designed for workloads that are computationally intensive from a computational perspective, which includes a web UI
E4DS-PLATFORM
E4DS-PLATFORM is the software stack that integrates all the components needed to implement the entire Data Science workflow.
E4DS-PLATFORM ensures that different high-performance environments for distributed data processing (Apache Spark, Dask, and Ray) can coexist in the same infrastructure and supports the major frameworks for Data Analysis and Machine Learning
- ICE4DS is an Interactive Computing Environment configured for Big Data Analytics, Machine Learning and Deep Learning.
- ICE4DS is based on Jupyter Notebook technology and includes VSCode
- ICE4DS includes several working environments: pyData XXL (supports development in Python, Julia and R), Rapids.AI, PyTorch, Tensorflow, MxNet, HuggingFace, Spark, Dask and Ray
- ICE4DS is designed to use the distributed Data Processing systems built into E4DS-Platform
- ICE4DS can be configured to offer dedicated computing resources to the end-user
Architectural advantages
ARCHITECTURE THAT MAKES A DIFFERENCE
READY-TO-GO DATA SCIENCE
End-to-end solution for large-scale Big Data Analytics and Artificial Intelligence orchestrated with Kubernetes.
VERSATILE
Urania allows different up and running versions for each of the integrated work environments, and ensures end users the possibility to create new customized additional ones.
OPEN SOURCE
Urania integrates only Open Source technologies developed by the most relevant communities active in the field of Data Science.
SCALABLE
Urania’s architecture enables it to meet the growing demand for computing resources.
Why choose this E4 solution
END-TO-END
End-to-end solution for large-scale Big Data Analytics and Artificial Intelligence orchestrated with Kubernetes. Integrated to maximize the productivity of data scientists across multiple work environments.
VALIDATED
Performance tests are carried out on all nodes before the solution is released. In addition to the usual firmware, homogeneity, sanity and setup check, we use additional tools to verify whether performance levels correspond to those requested by the customer. Relevant tests include: HPL (High Performance Linpack) to test machine’s computing power, measured in FLOPs; STREAM to test memory’s bandwidth, measured in MB/s; and IOzone to test disk’s access speed, measured in MB/s and IOPS.
TESTED
According to a protocol developed by E4, each component undergoes a burn-in test for up to 120hrs to ensures properly engineered and functioning systems. This procedure reduces DoA (Dead on Arrival), decreases “early failure” rate and highly improves the overall reliability of E4 solutions.
SERVICED
E4 is one of the few companies currently providing services at the highest level in large academic and private infrastructures as well as in national and international research centres of great complexity and relevance.
System support and solution customization
• per day
• in packages of “x” days on a pay-as-you-go basis
• per project
*minimum billable 1/2 day
Data science consulting
• per day
• in packages of “x” days on a pay-as-you-go basis
• per project
*minimum billable 1/2 day
Functional training on the E4DS-Platform environment
Extra | New functionalities coming soon
Urania can host third-party containers, specifically the images available in NVIDIA NGC Catalog, a large collection of pre-trained models, AI toolkits, and development kits (SDKs) specific to different use cases. The content available in NGC simplifies the construction, customization, and integration into GPU-optimized software workflows, accelerating the time to solution for Urania end users.
In addition to the features found in the base configuration, Urania can be enhanced with a number of additional features, some built into the solution and others available on demand. Contact us for release dates.
CLOUD-NATIVE DATA SCIENCE
