On a Red Hat Linux version 7 platform, HDFS can now be configured as a backend storage.
In this mode, storage no longer needs to be local to the machine. It can be in a remote cluster.

Benefits:

Note: HDFS storage is still not supported by the Consolidation server.

To configure HDFS:

Please contact us for advice on the architecture.

Several transformation and aggregation Java processors are now delivered out-of-the-box to help you cover basic use cases.

See "Consolidation Server Guide > Writing Transformation and Aggregation Processors > Java Processors > Define Java Transformation Processors > Use out-of-the-box transformation processors"

 

#attrsimilar is a query node in the index which returns all the documents matching the similarity query and calculates the similarity measure.
As it does not filter search results at all, you should combine it with a #filter to return only the documents having a similarity higher than a given threshold value.

Note: Similarity is the inverse of distance and is calculated as follows: similarity = 1 - distance

For more details about this function, see "Configuration Guide > Configuring Search Queries > Configuring and Searching Similarity Measures"

New operators have been added to perform chemical operations using an external library with chemical knowledge.

Previously, inside a slice, different queries were run in parallel but a single query was using only one core.
As of now a single query can also be run on multiple cores to reduce the time perceived by the end user.

So using more slices is no longer the only solution to reduce query time.

See the description of the nbVirtualPartitionPerSlice parameter in the XML Configuration Reference.

Cumulative hotfix tracability has been improved. Timestamps are now associated to our cumulative hotfixes to easily determine which hotfix version is installed.

The cumulative hotfix version can be seen in the Administration Console > About CloudView menu, or using the following command: KITDIR/tools/hotfix list

Installing a cumulative hotfix version is now automatically understood by the system (you no longer need to uninstall the previous version manually).

In the Administration Console, the Diagnostics menu indicates the number of outstanding issues detected by validators in the last applied configuration.

The Groups panel has been added to let you create your own group of configuration validators, by adding some builtin validators and your custom validators (if any).

In addition, deprecated components that are still used in the configuration are now reported in the Diagnostics.

See "Administration Guide > Monitoring CloudView > Identifying configuration issues".

For security reasons, it may be necessary to use a more recent JVM than the one provided with the kit. To do so, edit the JAVA_HOME variable in <INSTALLDIR>/ngroot.env

See "Installation Guide > Installing CloudView > Security Considerations > Use a different version of the JVM".

The XML Configuration Reference, the Semantic Reference (under "Configuration") and the CVConsole Commands Reference (under "Administration") are now part of the Online Help and their content can be searched for.

 

The R2018x GA only proposes the migration of the configuration in the install.sh -migrate procedure.

R2018xR1 also perfoms the automatic data migration with the following restriction.

Warning: Direct migration from 2012x, 2012x SP1, 2013, 2013 SP1, 2013 SP2 to 2018x is not supported.
To migrate from an old CloudView version to the R2018x version, first migrate to R2016x.R5 then upgrade from R2016x.R5 to R2018x.

In the following procedure, the original cvapp application was made on CloudView 2016x.R4.

Components tagged as deprecated in previous releases have been removed.

They will not be removed automatically during the migration operation, so you will need to remove them from your old instance before migrating.

Impacted components are: CustomCondition, FeaturesExtractor, MLNamedEntities, QueryMatcher in the <DATADIR>/config/Analysis.xml file.

For example:

Explanation: This issue is related to the Indexing Server and the slice(s) starting time. If the Indexing Server is up and running before the slice, this issue will not occur. If the Indexing Server is not started yet and the replica tries to start with the old index generation (before migration), this type of issue will occur.

 

This section provides information about end-of-support dates. For more details, see http://media.3ds.com/support/DS_LifeCycleInformation.pdf.

*FULL support starts when the release is available.
*EXTENDED support starts at the end of the FULL support for one additional year when customers are current with the applicable support charges and have paid the relevant additional charge.
For more details, see https://www.3ds.com/support/policies-offers/lifecycle-policy/