Domino Data Lab announced Domino 4.3, adding support for the popular Red Hat OpenShift distribution of Kubernetes to make it easier for its customers to scale data science workloads on any platform. 4.3 also improves Domino’s model monitoring capabilities, and extends its IT security features for enterprises via new reporting capabilities.
Domino offers a data science management platform that centralizes predictive analytics and machine learning (ML) research and development based on an open ecosystem that lets data scientists choose their preferred tools and algorithms while reducing the burden on IT.
“Large, sophisticated data science organizations demand flexibility in how they build and deploy their data science stacks. Adding Red Hat OpenShift to our wide variety of deployment options gives customers even more flexibility to run on almost any cloud provider or on their own on-prem hardware,” said Nick Elprin, co-founder and CEO at Domino Data Lab.
“We’re obsessed with delivering enterprise-grade security, control, reliability, and observability in a central platform that helps our many Fortune 100 customers unleash the power of data science. We continue to focus, with this release, on helping them accelerate and confidently manage their demanding data science operations.”
Expanded Elastic scaling with Red Hat OpenShift Kubernetes support
Kubernetes (K8s) is quickly becoming the IT standard for flexible containerized application orchestration across clusters with the ability to automatically deploy, scale capacity up and down on demand, and manage production workloads.
Red Hat OpenShift Kubernetes Engine, popular with IT teams, offers an attractive Kubernetes option for many customers since it can run on virtually all major cloud providers, as well as on-premise deployments.
With this release, Domino can now take advantage of intelligent Kubernetes orchestration on OpenShift clusters for efficient management and smart utilization of computing resources.
Rapidly scaling containerized workloads is particularly important as the demand for high-powered CPUs, GPUs and RAM can spike dramatically when training models or engineering features, and then quickly scale down once completed.
For organizations that have invested in large, centralized Kubernetes clusters to improve hardware utilization across a large pool of users and application workloads, Domino now supports multi-tenant Kubernetes clusters so a dedicated cluster for installation is not required.
Domino Model Monitor (DMM) enhancements
Domino Model Monitor (DMM), introduced in June 2020, now has powerful new capabilities that make it easier for enterprises to maintain high-performing ML models on any platform.
DMM lets organizations automate the monitoring of model inputs and outputs to detect changes in production data that could signal when a model is no longer producing results that are consistent with current business conditions.
Undetected data and model drift are especially problematic during a pandemic, since drastic changes to the economic environment and human behavior increase the likelihood of model inaccuracy and the associated risks of financial loss and a degraded customer experience.
The latest update includes new trend analysis capabilities that offer better insight into how the quality of a model’s predictions have been changing over time. It also includes new traffic charts to track the volume of model predictions and ground truth data (actual results) over time.
Advanced enterprise-grade authentication and security
Domino broadens its enterprise-grade authentication capabilities to include options for certification of Domino APIs and third-party services via short-lived Domino identity (OpenID) tokens to connect to any external authentication service.
When combined with its robust SSO capabilities, these enhancements make it easier for Domino administrators to grant or revoke user access while limiting where users are able to connect from.
Domino has also significantly enhanced its internal processes and tooling to comply with enterprise application monitoring and security reporting requirements, for example:
- Domino logs can be exposed to Fluentd-compatible aggregation tools
- Application health metrics can be integrated into Prometheus monitoring systems
- Container and dependencies support vulnerability scanning and remediation