Page 33 - Demo

Decoupling of storage, compute, and services
The separation of storage and compute resources is a funda- mental part of a modern data sharing architecture, as shown in Figure 4-3. All data is stored, in optimized form and without any loss of data fidelity, in the cloud. A single copy of the data stored in a modern cloud data warehouse — a single source of truth — can be accessed concurrently by any number of independent com- pute clusters, enabling an organization to perform any number of internal workloads, such as analytics.
FIGURE 4-3: A modern data sharing architecture built for the cloud with storage, compute, and services completely separate from each other.
Decoupling of storage and compute is also critical for sharing data. It enables data consumers to directly access shared data, using their own data warehouse compute power. But data consumers don’t pay for storage costs (because the shared data doesn’t move), and the data provider doesn’t pay for any of the compute that a data consumer uses to analyze shared data.
Multi-tenant metadata and transaction management
Making shared data usable requires access to data and coordi- nation across all data consumers to ensure consistency, security, and performance.
CHAPTER4 EnablingLiveDataSharingwithaModernCloudDataWarehouseArchitecture 27 These materials are © 2018 John Wiley & Sons, Inc. Any dissemination, distribution, or unauthorized use is strictly prohibited.