The current PubSub model presumes DataSetWriters are created based some logical structure within the source address space.

There are applications that have no logical structure and instead random variables from the source address space are streamed to the cloud.

Applications that use this model do not have clearly defined DataSetWriters that can be created and have to

1) Create a single DataSetWriter of all variables where the field name is a unique id for the Variable instead of a Name;
2) Create a DataSetWriter for each variable which causes the Connection object change every time a Variable is added or removed.

Proposal: add a new NetworkMessage flag: 'DynamicDataSetWriters'.

This flag indicates that DataSetWriters are dynamic and list of DataSetWriterIds is not known in advance. The DataSetMetaData will be published for each DataSetWriter and provides the DataSetWriterId for an active DataSetWriter. When using this mode NetworkMessage/DataSetMessage headers shall be present the PublisherId and DataSetWriterId shall be specified.

The PubSubConnnection metadata shall have a single DataSetWriter which is the template for all DataSetWriters. The MetaDataQueue name shall have the DataSetWriter Name as the last topic level. DataSetWriter name for the dynamically generated DataSetWriters shall be <templatename>.<datasetwriterid> (each unique DataSetWriter needs its own topic name since DataSetMetaData messages are sent with Retain=TRUE).