8 ways to successfully get AI and analytics into production
October 25, 2018 5:14 AM
8 ways to successfully get AI and analytics into production
Applications architect, author and Apache Software Foundation board member Ted Dunning outlines some of the habits and modern approaches highly successful teams use and others can leverage to get their own AI and analytics systems into production.
8 ways to successfully get AI and analytics into production
Getting a large scale system to work and getting it to work in production are very different, notes Ted Dunning, a board member with the Apache Software Foundation and chief application architect at MapR. Dunning is also the author of the newly released “AI and Analytics in Production ,” a short book that addresses finding value with AI and machine learning. He explains that value from big data becomes real when data-intensive applications go into production. "Although this can be challenging, there are ways to make it easier and more likely to succeed," he says. In the following slides, Dunning outlines some of the habits and modern approaches highly successful teams use and others can leverage to get their own AI and analytics systems into production.
Embrace true multi-tenancy
“This approach requires the ability to strictly and securely insulate separate tenants as appropriate while still being able to allow shared access to data when desired. Multi-tenancy should be one of the core goals of a well-designed large data system because it helps support large-scale analytics and machine learning systems while optimizing resource use both in development and in production.”
Simplicity and flexibility are key
“Big data systems don’t need to be cumbersome. If you find your design has a lot of workarounds, that’s a flag to warn you that you might not have the best architecture and technology to support large-scale data intensive applications in production.”
Build a Data Fabric
“One of the most powerful habits of successful data-driven businesses is to have an effective data fabric that spans the organization. A data fabric is not a product you buy. Instead it is the system you assemble in order to make data from many sources available to a wide range of applications developed and managed by multiple users. The data fabric is what houses and delivers data to your applications. This may exist across multiple data centers, on-premises or in a cloud or multi-cloud or even to the IoT edge.”
Orchestrate containers with Kubernetes
“When you build a production analytic or AI system, there are two parts of the problem. One is having the right data and data access, and the other part of the problem is the analytics: actually running the software to analyze the data. Analytics applications require a lot of coordination, and with the increasingly widespread containerization of applications, it’s essential to have a way to coordinate processes running in containers. Kubernetes, an open-source orchestration system for managing deployment of containerized applications, is emerging as a leading solution. But to avoid being limited as to which applications can be containerized, you need a data platform with the capability to persist data (state) from containerized applications as a variety of data structures. This powerful combination of Kubernetes and an appropriate data platform offer a big advantage for production systems.”
Extend applications to clouds and edges
“Many business goals drive a need to extend applications across multiple data centers—to take advantage of cloud deployments or multi-cloud architecture—and to go between IoT edge and data centers, making edge computing a practical option. That means having effective and easily administered ways to mirror data (files, tables and streams) and preferably to have efficient geo-distributed stream and table replication. The same platform should extend across multiple centers on premises or in cloud and be able to interact with Kubernetes across these geo-distributed locations.”
Use streaming architecture and streaming microservices
“Streaming can form the heart of an overall architecture that provides advantages that include the flexibility of a microservices approach along with fast response.”
Cultivate a production-ready culture
“Good social habits improve the chance for success in production. An overall data-aware organization is important. Also useful is a DataOps approach that extends the flexibility of DevOps by adding data-intensive skills such as data science and data engineering needed to deploy and maintain data-intensive applications in production. People also need to feel free to explore and experiment, making room for innovation.”