4.7 ESPDS Scalable and Secure Infrastructure

Tuesday, 8 January 2013: 5:00 PM
Ballroom G (Austin Convention Center)
Rich Baker, Solers, Greenbelt, MD; and D. M. Beall

As part of NOAA's Environmental Satellite Processing and Distribution System (ESPDS) Development program, Solers is leading a team to build an Enterprise Architecture based system for the Environmental Satellite Processing Center (ESPC) that provides robust data ingest, processing, product generation, and distribution functions by employing enhanced computing methodologies and the latest technology improvements, including a Service Oriented Architecture (SOA). These enhancements, currently being developed, provide a secure, scalable, and adaptable infrastructure in response to changing demands of the ESPC.

This new Enterprise Architecture makes use of a virtualized environment for efficient “on-demand” management of resources, leveraging VMware technology. This flexible architecture allows for effective monitoring of the services and intelligent dynamic resource management in a cloud environment. The existing (legacy) ESPC is a stove piped system where entire individual servers are statically reserved for single purpose needs. This new approach virtualizes the services and dynamically resources services to support the network based on its loading. This simplifies the hardware requirements for the system in that commodity computing servers (such as x86 blades) can be added to the network, shared across the system and tasked as needed, rather than just added to augment a specific stove pipe process. In some cases, servers can be re-used from other systems to augment the services offered in the Enterprise solution. The system logs are aggregated in a tool called Tripwire, which provides centralized logging, report generation, and Security Incident and Event Management (SIEM). Another tool called SolarWinds Orion allows the system administrator to monitor the system loading, and provides oversight to the hardware, services, and network in order to identify which services and users have the heaviest demands. Virtualization and clustering technologies such as VMware enable the dynamic distribution of services to balance the load depending on the time criticality and latency sensitivity.

In the future, High Performance Computing (HPC) technology will be leveraged to enable parallelization of Product Generation algorithms, which will speed up the processing time significantly while making the most effective use of computing resources. This is anticipated to result in an order of magnitude speed enhancement and reduce some algorithms' runtime from hours to minutes.

Migrating the ESPC from a stove-piped architecture to an enterprise architecture has significant payoffs in terms of speed, performance and cost. Great effort has been taken to analyze the ESPC's needs and express them in the form of services, with an eye for the system architectures of the future.

While it is impossible to know the shape of things to come, it is imperative that we design the services to leverage the architectures we see coming in the commercial world in the future. Leveraging SOA, virtualized infrastructure, and parallel processing will enable the next generation of services to be fast, low cost and upgradable.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner