Last time I mentioned IBM SPSS collaboration and deployment services and promised to talk more about it – so here we go:
Analytical Assets
Organizations positioning themselves to take full advantage of analytics will look to separate the effort of developing analytical assets and actually using them – between “creators” and “consumers”. Generally speaking, an individual (or team) can focus on the modeling process (creating) while key decision makers can focus on making decisions based upon analytical results (consuming).
The collaboration and deployment services product provided by IBM does more than offer a “virtual bridge” between analytical creation and consumption, it provides a “robust repository” where all parts that make up an analytical asset can be accessed and consumed. Additionally, this package is a centralized and searchable “enterprise analytical warehouse” (EAW) -designed to enable secure sharing and re-using of all of an organizations analytical assets. This allows identified users within an organization to explore and consume the analytical information available – which in turn will increase the effectiveness of decision making at every level of the enterprise environment.
The Bridge
The IBM documentation states that within the Collaboration and Deployment Service, a user can:
- Store analytical assets in a central, searchable repository, enabling the standardization and reuse of models to improve efficiency — and results
- Develop custom interfaces that run analytical processes – giving analysts and others greater control over how they access and use analytics
- Enable others to generate their own analytical output through a browser-based interface
- Operationalize analytical processes by initiating specific jobs – such as scores or reports — on demand, or at a scheduled time, or when triggered by other events, and by orchestrating complex jobs across multiple systems and applications
- Govern the environment in which the analytical processes occur and increase confidence in results
In an earlier post, I broke down basic data analysis into 3 steps:
- Identification (and preparation) of data,
- Selection of an analysis and summation method and
- Presenting the results
To that point, an intermediate step might be added here – “deployment” -which would involve “saving” all the artifacts of the first 2 steps to the organizations “enterprise analytical warehouse” using the collaboration and deployment service.
System Architecture
In general, the collaboration and deployment service consists of a single, centralized “analytical warehouse” (Collaboration and Deployment Services Repository) serving a variety of clients (using execution servers) to process and consume the available analytical assets.
An organizations functional analytical architecture would (should) include:
- Thin client (portals) and thick clients for CDS management and designing reports
- (Optionally) thick clients running “product collaboration” (allows direct access to the repository as well as file artifacts directly from their native product)
- A server (or server farm) acting as the analytical warehouse
- A database server and
- Multiple execution servers
Configuring the Environment
A typical installation and deployment is too detailed to explore in a blog post of course, but the following are important notes:
- It is advisable to utilize an experienced resource to ensure success with the installation and deployment
- Specific minimal hardware and software requirements must be met or (recommended) exceeded
- Prior to the install, verify that the necessary application server, database configuration, hardware, software, and permissions requirements have been met
- The installing user must have the appropriate file system permissions
- Before attempting installations, all required application servers and databases must be running and accessible
- Virtualization can be utilized but adds additional complexity and requirements to the installation
Scaling to the Future
The future of the analytical architecture is fully supported by the use and optimzation of the following:
- Migration tools,
- Optional add-on components,
- Clustering schemas,
- Logging services, and an
- Import tool…
Conclusion
The IBM SPSS family fully supports advanced analytics at an enterprise level with collaboration and deployment services. Other important members of this family include the Data Collector, the Modeler, the Decision Manager and (of course) Statistics.
One by one, I am going to expose them all!
______________________________________________________________
Thor: [walking into a pet shop] I need a horse!
Pet Store Clerk: We don’t have horses. Just dogs, cats, birds.
Thor: Then give me one of those large enough to ride.
