11.01.2017 Views

A Technical History of the SEI

ihQTwP

ihQTwP

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

AFSC/ESD conducted a source selection to find contractors to develop a model generic command-center<br />

architecture and functional specifications with supporting tools. ESD awarded research-and-development<br />

contracts to Ray<strong>the</strong>on and Hughes to build, test, and validate a “product<br />

line” approach to systems development [Hughes 1991]. TRW was later added as a third contractor.<br />

Eight line programs used <strong>the</strong> PRISM model architecture and supporting tools to realize an estimated<br />

56 percent average savings in cost and an average 66 percent savings in time. These eight<br />

systems were early examples <strong>of</strong> s<strong>of</strong>tware product lines.<br />

At <strong>the</strong> time, <strong>the</strong> <strong>SEI</strong> was engaged in developing <strong>the</strong> Feature-Oriented Domain Analysis (FODA)<br />

[Kang 1990] that analyzes a problem domain across multiple similar systems to identify common<br />

and variable features. FODA serves as <strong>the</strong> basis for a vast number <strong>of</strong> subsequent feature modeling<br />

approaches and dialects still in use today. At <strong>the</strong> <strong>SEI</strong>, FODA later evolved into product line analysis,<br />

which extended <strong>the</strong> analysis <strong>of</strong> commonality and variability beyond features to quality attributes.<br />

The <strong>SEI</strong> investigation into product lines was also made possible by its concurrent focus on<br />

s<strong>of</strong>tware architecture. <strong>SEI</strong> contributions in architecture definition, documentation, and evaluation<br />

were an important part <strong>of</strong> a s<strong>of</strong>tware product line approach. Serendipitously, <strong>SEI</strong> staff members<br />

traveled to Sweden to interview staff at CelsiusTech Systems AB, ostensibly to do an architecture<br />

case study; what <strong>the</strong>y found was that CelsiusTech had taken a product line approach that was<br />

achieving significant results in ship systems built for national defense, a domain <strong>of</strong> interest to <strong>the</strong><br />

DoD. Those results included systems completed in days instead <strong>of</strong> years, order-<strong>of</strong>-magnitude<br />

productivity gains, and mass customization where 20 s<strong>of</strong>tware builds were parlayed into a family<br />

<strong>of</strong> over a thousand specifically tailored systems. The promise <strong>of</strong> product lines that was documented<br />

in <strong>the</strong> CelsiusTech case study [Brownsword 1996] led <strong>the</strong> <strong>SEI</strong> to pursue an initiative in<br />

s<strong>of</strong>tware product lines. The <strong>SEI</strong> recognized that when developing multiple similar products, <strong>the</strong>re<br />

will be some degree to which <strong>the</strong>y are <strong>the</strong> same, but <strong>the</strong>re will also be some degree to which <strong>the</strong>y<br />

vary. Economic advantage is achieved through a systematic product line approach that effectively<br />

manages this variation. Creating a s<strong>of</strong>tware product line depends on establishing a s<strong>of</strong>tware architecture,<br />

or product line architecture, for <strong>the</strong> entire set <strong>of</strong> systems.<br />

Broad Use <strong>of</strong> <strong>SEI</strong> Approaches to S<strong>of</strong>tware Architecture<br />

The influence <strong>of</strong> <strong>SEI</strong> work in s<strong>of</strong>tware architecture on <strong>the</strong> DoD has been broad and pervasive.<br />

Major defense contractors, such as Boeing and Ray<strong>the</strong>on, now have architecture evaluation teams<br />

and architecture evaluation as part <strong>of</strong> <strong>the</strong>ir architect certification processes. Also, U.S. Army staff<br />

have reported repeatedly that use <strong>of</strong> scenario-based architecture evaluation methods reduces risk<br />

in schedule and cost, improves documentation, and results in higher-quality products.<br />

Moreover, in 2009, <strong>the</strong> U.S. Army mandated that all Project Executive Offices appoint a chief<br />

s<strong>of</strong>tware architect (CSWA) to be responsible for oversight and management <strong>of</strong> s<strong>of</strong>tware development<br />

within each PEO. The memo specified that <strong>the</strong> CSWA must earn a S<strong>of</strong>tware Architecture<br />

Pr<strong>of</strong>essional Certificate from <strong>the</strong> <strong>SEI</strong> (or ano<strong>the</strong>r certificate-granting organization with s<strong>of</strong>tware<br />

architecture expertise). The decision was based on an understanding <strong>of</strong> <strong>SEI</strong> work in s<strong>of</strong>tware architecture<br />

and its impact and, in particular, a recent impact study <strong>of</strong> <strong>the</strong> use <strong>of</strong> <strong>SEI</strong> architecture<br />

evaluation techniques in <strong>the</strong> Army [<strong>SEI</strong> 2009]. Also, <strong>the</strong> <strong>SEI</strong> conducted a study <strong>of</strong> <strong>the</strong> impact <strong>of</strong><br />

<strong>the</strong> Army Strategic S<strong>of</strong>tware Improvement Program (ASSIP)-sponsored QAWs and architecture<br />

evaluations using <strong>the</strong> ATAM. Ten out <strong>of</strong> 11 programs that responded to <strong>the</strong> survey indicated that<br />

ATAM/QAW produced better results than <strong>the</strong>y traditionally obtained.<br />

CMU/<strong>SEI</strong>-2016-SR-027 | SOFTWARE ENGINEERING INSTITUTE | CARNEGIE MELLON UNIVERSITY 237<br />

Distribution Statement A: Approved for Public Release; Distribution is Unlimited.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!