HiCLAS1  HiPERiSM Consulting, LLC, Linking with AS1MET Services

Cost Effectiveness FAQ

Is it cost effective to purchase licenses for AERMOD-HPCS?

The issue of cost effectiveness revolves around two metrics of value: (a) workload throughput and, (b) application performance. In response to the first metric the argument currently accepted is that buying more hardware is the most cost-effective solution to AERMOD throughput. Historically this was the only alternative because there has been no concerted effort to improve the application performance. While adding more hardware  does improve the thorough-put (i.e. increase the frequency of AERMOD runs) this choice does nothing to reduce the run time of AERMOD. The HiCLAS1 approach has been to make progress based on metric (b) application performance. As a result the current release of AERMOD-HPCS does reduce the run time of AERMOD whereas increasing hardware capacity brings with it new issues (as discussed in the next FAQ). As a footnote: HiCLAS1 now offers a response to metric (a) with release of the HiCLAS1 Task Manager (HCTM) to enable concurrent execution of either AERMOD (the Preferred model), or AERMOD-HPCS (an Alternate model) and further information is available at the Products page (click tab at left). Back to top of page.

Why not just buy more hardware to increase workload throughput?

The idea that buying more hardware resources solves the workload throughput issue has hidden costs associated with it. These costs include (a) salary for system administration, (b) depreciation, (c) hardware and software replacement costs over the lifetime of the machine (typically less than four years). These are real considerations that, in the HiCLAS1 experience, needed to be addressed. Over the life time of a processor generation typical maintenance issues have included problems such as memory failure, hard drive controller failure, motherboard failure, and operating system corruption. Recovering from each of these types of problems costs time and money. Therefore, the larger the number of machines to maintain, the shorter the mean-time-to-failure for any one of them. Back to top of page.

What is the cost benefit of AERMOD-HPCS?

The current product (AERMOD-HPCS v1.8) is the first step in improved performance for AERMOD and is 2 to 3 times faster than the U.S. EPA's AERMOD. Not only that, but it also provides superior work-load throughput performance when compared to the U.S. EPA AERMOD distribution (see HPC-2007-3). Future releases will offer even more performance enhancement. At this time, EPA's AERMOD workload throughput requires 2 to 3 times more machine capacity to equal the throughput of AERMOD-HPCS. This argument applies no matter how many machines or CPU's there are to begin with. So the cost benefit question reduces to two parts: (a) "how many machines are needed to, say, double the throughput of the workload with U.S. EPA's distribution?", and, (b) "what is the relative cost of adding more hardware versus buying AERMOD-HPCS licenses"? For some answers to these questions see the FAQ's on license issues. Back to top of page.

Send mail to george@hiclas1.com with questions or comments about this web site.
Copyright © 2007 HiCLAS1
Last modified:  04/15/2009