Working with SAS Viya Platform
06/12/2017 by Ivan Gomez Modernization - Infrastructure
Being the analytics software industry leader, and as part of its commitment to deliver products that exceed customer needs, SAS unveiled a completely revamped analytics framework during the most recent SAS Global Forum called SAS Viya. Now after a few months of intense anticipation by the analytics community, SAS Visual Data Mining and Machine Learning, the first of many native SAS Viya products, has been released. DMML targets professional data scientists and includes SAS’s newest machine learning and text analytics algorithms to perform powerful analysis and gain insight using supervised and unsupervised learning techniques on structured and unstructured data.
While SAS Viya does provide new algorithms and features, SAS Viya has a lot more to offer than an array of new procedures and a catchy name. It is in fact a game changer and this post covers key aspects you must know about it.
The deployment of SAS Viya products departs from the operating-system independent SAS depot towards standard native tools. Currently, the installation in Red Hat Linux uses the YUM package manager to retrieve the software from SAS’s private repositories and perform the installation of the RPM packages on the customer’s environment. These RPMs contain the source code along with all the information needed to deploy it in the server, such as directory layouts, permissions, etc.
This approach simplifies the installation by removing the need to transfer large depot files and to create deployment plans while allowing SAS to rely on feature rich deployment tools. SAS Viya uses Ansible, a configuration management and IT automation tool, to automate the deployment. Ansible playbooks provided with the SAS order express the desired final states of servers, and Ansible communicates with the servers over SSH to achieve these states.
CAS is the new analytics engine that powers the SAS Viya ecosystem. It is composed of an in-memory engine with ability to overflow to disk to work with tables larger than the total available memory and microservices that provide admins with user and data management capabilities.
CAS has an elastic and scalable design in line with today’s need to perform infrastructural adjustments based on system load and demand. It’s easy to add or remove worker nodes from an existing CAS deployment. This makes it possible to leverage platforms such as AWS to spin up new worker VM’s during times of peak usage and shut them down when the demand for resources decreases.
CAS sessions are the link through which users connect to and interact with CAS. When a CAS session is started, a new process is created in all members of the CAS cluster and all subsequent CAS actions happen through that session. CAS sessions also provide other functionality such as user authentication and authorization, resource tracking and management, and isolated execution of actions.
Instead of relying on the traditional metadata server, CAS has built-in authentication supporting Active Directory, LDAP, PAM, etc. and authorization to configure how users are granted access to data and resources.
Because each session creates a new process in all the CAS servers, it is easy for admins to see and limit how much computer resources a session can use.
Sessions are independent of each other and therefore users do not have to worry about what other users are doing. If one session fails, only that session dies and all other sessions and the server keep operating without interruptions.
SAS Viya targets a wider audience than previous SAS products. Users no longer need to learn the specifics of SAS interfaces or the SAS language. Instead, they can use their existing skills and leverage Python, Java, Lua, and REST API’s to communicate with SAS Viya and gain access to its analytics algorithms and premium performance. This means that non SAS users can think of SAS Viya as an addition to their analytics toolset and not necessarily as a complete replacement that requires them to switch away from their preferred tools.
Of course, SAS Viya also provides native SAS interfaces. The default coding application shipped with SAS Viya products is SAS Studio. In addition, existing SAS 9.4 deployments can also cohabitate with SAS Viya deployments and execute code in CAS using SAS/Connect.
You can visit the SAS Viya Homepage to find more information or talk to Zencos about a free trial.
We help you set the right course, develop the right solution, and transfer the right knowledge. Put our experts to work on your next Analytics project.