IBM hosted an Open Mic webcast with Cognos Support Engineers from India on 29 January 2014. The topic was "Cognos Dynamic Cube".
|Q & A|
Q. Is Dynamic Cube Designer a separate install?
A. Yes ,Cube Designer is installed separately in the same directory as Framework Manager.
Q. Can Dynamic cube be deployed on multiple dispatchers?
A. Yes. Dynamic cube be deployed on one or more dispatchers
However deploying it on multiple dispatchers can cause an overhead on the database.
Example - If Administrator refresh the cube on all instances(dispatchers) at once, then the load on the database will increase as simultaneously number of queries will be fired to fetch the data.
Q. What if developers do not have Admin rights? Can they still publish the cubes ?
A. Dynamic cube will be published by the Dynamic Cube designer Modeler or Framework Manager Modeler who have the rights to publish the Cube.
Q. Does Dynamic Cube support only Dynamic Query mode (DQM) ?
Q. What is the minimum hardware requirement to set up Dynamic Cube environment ?
A. Kindly refer details in the Developer Work Article :
Q. Is there any size limitation to Dynamic cube ?
A. Dynamic Cube has been designed for very large volume of data running into terabytes.
Q. If Dynamic cube is in-memory OLAP cube, then why there is need to model and publish it again?
A. Metadata is modelled in Cube Designer as per end user's requirement.
This model is then published so that it is made available for the end users to create analysis and analyze the data.Data,Aggregates and results are loaded and refreshed in-memory.
Q. How to refresh data in the cube? Does it require a downtime?
A. Refreshing data in a cube is nothing but refreshing the caches related to that cube.
This does not require a downtime.The member cache and data cache are updated live.
So when a refresh of the Caches is run, the existing running reports keep referencing the old cache and at the same time a new cache with latest data is created. Once the request on the old cache are completed , it is discarded and the new cache becomes active.
Q. If the 64 bit report server enabled, how does it affect the standard Framework Manager package?
A. If Report Server is set to 32 bit, it can serve both Compatible Query mode (CQM) and Dynamic Query mode (DQM) .
However if Report server is set to 64 bit, then the CQM package based reports will not work.
The dispatcher dedicated to dynamic cube should have the Report server set to 64 bit.
The other dispatcher should have the report server set to 32 which will host CQM packages.
Package routing then need to be set by the Administrator so that DQM and CQM query can be appropriately routed to the specific dispatcher.
Q. Any performance impact on the Dynamic Cube if using 32 bit Report Server ?
A. No performance impact because Cognos decides when to use 64 bit process and when to use 32 bit process while serving a request .More details in the following technote :
Q. What is the source for Dynamic cube
A. Any supported relational database modeled as Star or Snowflake schema can be a source for the Dynamic cubes
Q. Is there an alternate strategy recommended if there is a database where Snowflake or Star schema is not practical?
A. No. To create Dynamic Cubes the datasource must be in Star or Snowflake schema.
Q. Can a Dispatcher handle more than one Dynamic cube.
A. One dispatcher = 1 DQM server.
One dispatcher can handle multiple cubes if it has sufficient hardware resources but better to have dedicated Dispatcher for one Dynamic Cube.
Q. If the cube is configured to only one dispatcher then all dispatchers can reference the cube?
A. If the cube is configured to one dispatcher then only that dispatcher can serve it not the others.
so the cube needs to be configured to every dispatcher where you want it to be used.
Q. Does Dynamic cube have capability to generate relative time dimension (YTD, QTD, MTD, etc) like in PowerPlay Transformer?
Q. Does cube start automatically if the Dispatcher is restarted from the Windows services or does cube has to be manually started after a server restart?
A. Cube can be set to start in auto mode which means start it when the Query service starts(dispatcher starts )
It can also be started manually. If choose to start it manually then the auto start option in the settings should be unchecked.
Q. Can workload logs be directed to Audit database and build reports against it?
Q. Is it possible to convert the old model DMR built in Framework Mannager to Dynamic cube ?
Q. How to apply data level security in Cube designer? Can row level security be applied in Dynamic cubes?
A. Dynamic Cube has a different concept of applying security. This is because data is not pre-aggregated in Dynamic Cube as in Transformer Cube.
Security is applied in the form of Security filters and Security Views.There can be one or more than one Security filter within Security View.Each Security filter has Level members.
These Security filter/Security Views are assigned to the users and hence users have permission to view data corresponding to only those members defined within Security View/Filter.The data is aggreagted in run time corresponding to the applicable members.
Q. Does the Aggregate Advisor also provide information for regular Framework Manager packages running DQM (64-bit)?
A. No. Aggregate Advisor is not applicable to Framework Manager packages.
Q. How to refresh a Dynamic Cube ?
A. Refreshing a Dynamic cube involves refreshing a number of caches that are related to Dynamic cube. Administrator can define the task and schedule it to run at specified intervals.
More details in the Presentation attached.
Q. What are the parameters required for creating data source connection for Dynamic Cube?
A. Pre-requisites are jdbc drivers for that database. While creating the data source connection the parameters required are Server name, port number and database name.
More details in the Presentation attached.
Q. What is the actual difference between Framework Manager and Cube Designer because there is option of Dynamic Query Mode (DQM) in both.
A. Framework Manager is a modelling tool that can enable a CQM and DQM mode for the package.
It can not model the cube.
Cube Designer models the Dynamic cube which by default uses DQM mode.
Q. Why would use Dynamic Cube if TM1 Cube already exists?
A. Dynamic Cube can analyze very large volume of data due to its feature of In-database and In-memory aggregation.
But if write-back functionality is required then TM1 cube should be used.