CISL Annual Report 2005 > Production cyberinfrastructure plans

Data analysis and visualization plans

The Data Analysis Services Group (DASG) will continue to support the UCAR scientifc community through further enhancements in the VisLab environment, and significant progress on the design of a new advanced computing environment tailored to the specialized needs of interactive data post-processing, analysis, and visualization. Staff will continue to support the special needs of the UCAR scientfic community in the areas of visualization research and algorithm development and assistance with the production of animations and imagery. The following are key activities planned for FY2006.

Production visualization services

DASG will continue to provide visualization services for the most demanding scientific datasets generated by the UCAR community. Current planned activities for FY2006 include:

1. Continuing collaborations with CISL's Computational Sciences Section on the HOMME model. Specifically, DASG will generate new visualizations and work toward delivering simple visualization tools aimed at aiding diagnostics for HOMME.

2. Visualizing observed upper atmosphere ozone data for ACD.

Visualization Lab upgrade project

In FY2006, DASG will continue its efforts to further transition from proprietary to commodity technologies. These measures are aimed at both saving cost and exploiting rapid technology advances in many aspects of the commodity space. DASG will also begin exploring steps to better integrate visualization and data analysis resources as described in the analysis environment project below.

Specific items planned for FY2006 include:

1. Replacing the VisLab's five-year-old tiled, stereo projection system with current vendor offerings. Though it is unlikely that a commodity solution will present itself in this case, the current display environment suffers from a number of pitfalls that should be addressed (e.g. poor image registration, brightness and color balance between projectors, the need for multiple image generators, and the highly reflective and distracting silver screen required for passive stereo projection). Products anticipated to become available this year would allow the replacement of the current multi-projector display with a single projector that would address all of the present environment's shortcomings.

2. Further evaluating the small 64-bit visualization cluster deployed in FY2005, and pending results, acquire additional nodes allowing us to retire the lab's costly Onyx4-based remote visualization service.

3. Replacing the VisLab's five-year-old Fibre Channel RAID storage system with a RAID device based on commodity SATA drive technology.

Analysis environment project

The analysis environment project addresses several key components in moving us toward a high-performance analysis environment, on par with NCAR's renowned supercomputing capabilities. The project began in FY2005 with the research necessary to define the architecture for a new analysis environment. The next step is the installation of a testbed and subsequent analysis of test results. The project ends with recommendations for a production environment, an implementation plan, and the installation of new resources in FY2007.

The following project phases are planned for FY2006:

1. Activities in phase one focus on initial research into available technologies and the construction of a test environment. Specific goals in phase one include:
1. Basic architectural research, including reviewing previous research conducted by SCD.
2. Initial vendor contacts to ensure we choose promising resources for the testbed.
3. Visits to sites that have already integrated similar solutions.
4. Aquisition and installation of initial testbed equipment.
5. Development of benchmarks for analysis and visualization environments.
6. Development of testing criteria for storage and file system solutions.

Analysis environment project testbed phase 1 diagram.

 

2. Activities in phase two will focus on the evaluation and integration of a high performance, heterogeneous, shared file system and the evaluation of computational resources for use as analysis platforms. Specific goals in phase two include:
1. Identifying a suitable shared file system technology and characterizing the needs for its performance and capacity.
2. Identifying suitable computational nodes based on performance, supported software stack, interoperability, and maintainability.
3. Expanding the testbed to include all architectures supported by SCD.
4. Developing policies and mechanisms for allocating nodes and storage space.
5. Initial investigation of visualization integration.

Analysis environment project testbed phase 2 diagram.