Deployment options for Tamr include:
- Single node, hosted on a single dedicated server on the cloud or on-premises at a customer location. In single-node deployments, Tamr supports use cases with up to 10M records. See Single-Node Deployments.
- Cloud-native, hosted on the customer’s cloud infrastructure. Performance can be tuned by scaling out resources.
Tamr can be deployed on these cloud platforms:
- AWS (Amazon web services). See Deploying Tamr on AWS.
- GCP (Google cloud platform). See Deploying Tamr on Google Cloud Platform.
- Microsoft Azure. See Deploying Tamr on Azure.
The following diagram provides a general illustration of the software stack.
Tamr (top) and processing (bottom) layers
- The Tamr application layer is composed of an array of loosely-coupled RESTful microservices that provide a broad array of capabilities, while simultaneously allowing flexibility in how the application is deployed, and scalability of individual application components (heterogeneous scaling).
- The data processing layer assembles a set of external microservices. These highly-scalable components provide both high-volume data processing and low-latency search and filtering.
The complete software stack can run in cloud environments or be installed on premises.
Note: Typically, Tamr deployments include three identically configured environments for training and iteration (DEV), validation (TEST), and implementation (PROD). For best practices for implementing this environment structure, Tamr customers can consult the Tamr knowledge base.
Updated about a month ago