Database Reference
In-Depth Information
low-latency integration requirements. Data Replication has sophisticated
functionality for high-speed data movement, conflict detection, system moni-
toring, and a graphical development environment for designing integration
tasks. Furthermore, it's integrated with the set of IBM PureData Systems for
high-speed data loading/synchronization, and also with Information Server
to accumulate changes and move data in bulk to a target system.
Federation
involves accessing data that's stored in federated repositories
through a federated query. This is often used to retrieve data from multiple
systems or to augment data that's stored in one system with information
from another system. IBM InfoSphere Federation Server (Federation Server)
accesses and integrates data from a diverse set of structured data sources,
regardless of where they reside. It enables hybrid data warehousing by join-
ing data from multiple repositories, and also exposing information as a ser-
vice (IaaS) via InfoSphere Information Services Director. From a federation
perspective, it's important to be able to federate search (as the Data Explorer
technology we outlined in Chapter 7) across your enterprise assets, as well
as with a query API such as SQL. In the future, Federation Server may inte-
grate with Data Explorer to provide structured data source search and query
within an overall Big Data (structured and unstructured) federated search
and discovery.
We believe that organizations shouldn't try to solely deliver enterprise
integration with Hadoop; rather they should leverage mature data integra-
tion technologies to help speed their deployments of Big Data, whenever that
makes sense. There's a huge gap between a general-purpose tool and a pur-
pose-built one, not to mention that integration involves many aspects other
than the delivery of data, such as discovery, profiling, metadata, and data
quality. We recommend that you consider using IBM Information Server
with your Big Data projects to optimize the loading (via bulk load or replica-
tion) of high-volume structured data into a data warehouse, and extending it
with federation when required; the loading of structured or semistructured
data into Hadoop; and the collecting of information that's filtered and ana-
lyzed by stream analytics. You can then load the data into a relational system
(such as a data warehouse), federate queries across relational databases as
part of Big Data federation and navigation, and replicate data sources to a
Hadoop cluster or other data warehouse.
