You should have the course outline in front of you. The CSSE office has a master copy available for photocopying. I will now go over the important points.
Submission of Assignments Students are required to submit their assignment by 12 NOON on the due date. Shortly thereafter, the box will be emptied and sealed. Late assignments must be handed directly to staff in the Enquiries Office, where they will be stamped with the date and time received. Late submissions will be penalised at the following rate: Submission on the due date but after 12 noon: -10% Submission the day after the due date -20% Submission two days after the due date: -40% Submission 3 or more days after the due date will not be accepted. Extensions Extensions will not be given for group assignments. Given the size of the groups, the illness of individual students will not be accepted as a reason for extensions. In the case of extensions for individual work, extensions will be granted if valid medical certificates are produced. Group Assignments All members of the group will rate all other members of the group and this rating will modify the mark that each individual receives. If a group is having trouble with an individual member and your efforts to resolve the issue has been fruitless, then the group should approach myself to assist in resolving the problem prior to submission of the assignment.
CS90 - How Westpac wasted $250 million In 1988 Westpac decided to replace their entire banking system, using lots of cutting edge technology (OO etc.) “Core System 90”, and $89 million, five year, project: “ Westpac Banking Corp. had high hopes for Core System 90 (CS90…In 1988, when Westpac launched the $85 million, five-year, soup-to-nuts project, CS90 was slated to redefine the role of information technology at the Sydney, Australia based bank giant. The plan was to create a showcase of decentralized data systems, allowing branch managers to spin out new financial products rapidly using CASE (computer-aided software engineering) tools and expert systems. … But those hopes have been dashed. Last month, more than three years after CS90 was begun, Westpac reluctantly ponied up to the worst: The project is out of control. CS90 has already drained nearly $150 million from Westpac's coffers-and the bank has virtually nothing to show for it. What's more, CS90s scheduled 1993 completion date is now just a pipe dream.” (Information Week, Dec. ’91) Early 1992: “Westpac banking, Australia's largest bank, will eliminate 500 jobs from its development technology area, saving $37 million a year. The action follows the termination of a runaway systems project that cost $150 million-nearly twice the original estimate when the project started in 1988.” Important to remember that the banks are amongst the best software developers, since it is absolutely critical to their business. Therac-25 - Radiation death courtesy of the computer http://sunnyday.mit.edu/papers/therac.pdf Between June 1985 and January 1987, a computer-controlled radiation therapy machine, called the Therac-25, massively overdosed six people. Worst in 35 year history of medical accelerators. First in Therac series to rely on software for safety. Earlier models had separate circuits and hardware interlocks. Cryptic error messages “MALFUNCTION N” (1<=M<=64). Not explained in User Manual. No indication of patient risk. Operators could simply “proceed” after a malfunction, by hitting the “P” key Concurrent access to shared memory. Race conditions. Logic error meant that parameters entered by the operator could appear on the screen, but not cause the corresponding internal variables to be modified. Patients received doses 1000s of times higher than prescribed. There was more than one independent, serious software fault. Over-reliance on software – assumption that only hardware fails. Some faults had been present in software reused from Therac-20, but hardware interlocks had prevented accidents, and thus their detection.. McKinsey’s PeopleNet from http://www.prenhall.com/divisions/bp/app/alter/student/useful/ch8mckinsey.html The most important resources for an international consulting company such as McKinsey & Co., are its people. Accordingly, McKinsey set out to develop PeopleNet, a company-wide human resources information to keep track of staffing assignments and related information about the firm's employees. During the project McKinsey encountered many of the common setbacks consultants often help their clients with. As a first foray into client-server computing, the project was a departure from the company's previous focus on mainframes and PCs, and some of the people in the IS group never agreed with the technical approach. The expected infrastructure based on a $20 million telecommunications project was not available because that project was later judged too expensive. Consequently the PeopleNet project had to bear the unexpected burden of building infrastructure incrementally. McKinsey's highly independent branch offices worried that a uniform set of guidelines would reduce their traditional local control over staffing. After many meetings, control of the part of PeopleNet related to tracking staff assignments was given to local offices, which agreed to share with headquarters only a limited amount of personnel data, such as languages spoken, skills, and experience. There were also problems with data discrepancies between the Scandinavian offices and headquarters in New York, resulting in a massive cleanup. Originally seen as a 12 month, $4.9 million project, the revised projections were that it would run 30 months, cost $12 million, and produce a system somewhat different from the one originally envisaged New Jersey Department of Motor Vehicles http://doi.acm.org/10.1145/88636.87943 Figures unverified: In 1984, they decided to replace their database system using 4GL modern techniques (4GL etc). When the new system was delivered, they cut across to it and the old one was decommissioned. On the first day of use, the overnight run took ~18 hours. By the end of the week, it was taking ~28 - clearly an impossible situation. The system “worked”, but performance testing had not been done. Failure to scale up from prototype. The old system had to be recommissioned, at a cost of &quot;wasn't revealed but in the millions&quot;. Microsoft’s first Windows database - Omega http://www.avdf.com/nov96/acc_jet.html http://www.xoc.net/works/architecting_large_programs/default.asp “ Studies have shown that somewhere between 15% and 40% of large software projects fail. The larger the project, the more likely it is to fail. By failure, I mean they are cancelled or postponed, or are never used. DeMarco and Lister in Peopleware found that in over 500 projects they studied, 15% of them failed. Of projects larger than 25 person-years, more than 25% of them failed. Other studies have shown much higher percentages. Entire books have been written about some of these monumental software failures, discussing such projects as the baggage handling system at Denver International Airport. There are many reasons for these failures. These numbers are appalling. This paper addresses technical issues revolving around the design of programs in Visual Basic to address design methodologies s that will give a large project a better percentage chance of succeeding. I'll leave it to others to discuss how to solve management and other issues that contribute to the failure of projects. Just to give you a firsthand example: I worked on one such failure when I worked at Microsoft. I was a Software Design Engineer at Microsoft from 1988 until 1992, working on what became Microsoft Access. However, the first two years, I worked on a Project code-named Omega. Unless you've seen one of the demos of it that Tod Neilson of Microsoft shows from time to time, you've probably never heard of Omega. It was Microsoft's first attempt at a Windows database program. I would guess that somewhere around 200 person-years were discarded when the project was cancelled. I won't list all the reasons for it's demise and the subsequent revival as Microsoft Access, but part of the problem was technical: The program was just too big. If we applied the techniques that I describe in this paper to that project, it might have had a better chance of succeeding. The techniques I'll discuss use the tools supplied by Visual Basic to avoid the technical pitfalls of many software projects. ” Omega didn’t make it, but its scripting language went on to become VB. Australian Customs Service - Intelligence Gathering System Decided to replace their intelligence system. The resultant system, at a cost of ~$A 3 million, proved to be so difficult to use that the only people who used it were the data entry operators. It was effectively a write-only system! (The intelligence people had decided to develop it themselves, rather than have the systems people do it) A later review decided that it might be worth spending another ~$100,000 to try to save it, but that it would probably have to be written off. (from Martin Dick) Denver International Airport http://www.geocities.com/kschottland/dia.html http://www.prenhall.com/divisions/bp/app/alter/student/useful/ch13denver.html The first large-scale airport built in the US for 20 years. The airport was completed behind schedule, and due to complications with the baggage handling system, it was significantly over budget. The initial estimate of the baggage handling system was $193 million (Gibbs, 86). Ultimately the project ballooned to $311 million and only serves one of the three concourses it was designed to handle (Schloh, Comparative). Further, the system serves only half of the airport's eighty-four gates and is only being utilized for flights departing or terminating in Denver. Statistically, this means the baggage handling facility is currently operating at twelve percent of its designed capacity. London Ambulance Service http://legacy.eos.ncsu.edu/eos/info/computer_ethics/risks/safety/medical/study.html http://catless.ncl.ac.uk/Risks/14.48.html#subj3 The Ambulance Chief of London retired in October of 1992 after a new computerized ambulance dispatch service caused delays resulting in the deaths of up to 20 people. Employees were against the use of the new system. Within hours of going online, calls were being missed and multiple ambulances were dispatched to the same site. Incoming calls were delayed longer than usual, as long as 30 minutes. The old system would have handled the load without breaking down. After 36 hours, they reverted to the manual system. Not enough consideration was given to system overload; there appears to have been no backup procedure. Operators of the system were quoted as saying &quot;there was no way to scroll back through the list of calls to ensure that a vehicle had actually been dispatched&quot; and &quot;the exception list just kept growing&quot;.
A Cautionary Tale In E-Trade's Glitch Crash Shows Internet's Vulnerability By Mark Leibovich Washington Post Staff Writer Monday, February 22, 1999 The chief technology officer was working on 90 minutes' sleep, a fitful night even by her three-hour norm. Now, shortly after the stock market opened Debra Chrapaty was seeing red on her monitor, and red meant crisis. It meant hundreds of thousands of online investors could not buy or sell stocks through E-Trade, the Internet brokerage whose computing operations she directed. She took a Tagamet pill to soothe her ulcer. Meanwhile, in a Silicon Valley schoolyard, E-Trade communications chief Lisa Nash was pledging allegiance to the flag, a ritual she performs with her two daughters every morning. Her beeper pierced the formality with the Code Red news and Nash kissed her daughters goodbye. On Wednesday, Feb. 3, one of the Internet's highest fliers was effectively grounded by a software glitch. E-Trade's 700,000 account holders were locked out of their online portfolios as the stock market was dropping in heavy trading. Every second E-Trade didn't work could threaten its viability, like a body deprived of oxygen. Internet time is unforgiving, and so are subscribers with money at stake. &quot;Get it together E-Trade,&quot; a subscriber railed in an online chat forum, &quot;Or you're going to the E-Grave.&quot; E-Tolerance? The Internet engenders little when problems occur, a reality that was slammed home at E-Trade this month. Online services now are perceived as utilities, not luxuries. In short order, subscribers have come to demand light-switch reliability when they tap their keypads, never mind the unfathomably complex technology underpinning those commands. As the Internet has bred exhilaration, user expectations likewise have soared, and when something goes wrong, look out. Already, two class action lawsuits have been filed against E-Trade in conjunction with this month's breakdowns. &quot;Technology fails, and anyone who promises otherwise is full of [it],” Chrapaty said. &quot;Things happen on any given day. My job is not to let our customers notice.” Yet over parts of three days, as its customers could only agonize, E-Trade was laid bare as a public microcosm of Internet frailty. It also became a corporate case study of crisis management in the Internet age, the desperate side of the exhilaration. Chrapaty and Nash were steeped in the ordeal : Chrapaty, 38, a former technology chief for the National Basketball Association, led the high-tech paramedics from E-Trade's offices in Alpharetta, Ga.; Nash, 40, a former State Department intern, headed damage control from the company's headquarters in Palo Alto, Calif. &quot;We need to perform with militaristic excellence,&quot; said Chrapaty, invoking the preferred imagery of chief executive Christos M. Cotsakos, a Vietnam War veteran who was awarded the Purple Heart: &quot;We need to operate with the souls of warriors.&quot; 'Wired to Wall Street’ E-Trade is one of the emerging Internet brands to invade contemporary lives, a dual product of the online boom and raging bull market. Now the third -largest Internet brokerage, E-Trade Group Inc. processes 60,000 trades a day. Some subscribers have given up their jobs to ride the gamblers' surge of &quot;day trading,&quot; tying fortunes to the stock market and seizing some control of its fluctuations through E-Trade. They pay nothing to subscribe and keep a minimum $1,000 trading balance in their accounts. E-Trade receives $14.95 or $19.95 commissions on each trade, depending on what exchange the stock is listed. Traditional brokerages typically charge twice that. &quot;Good morning is being replaced by 'Did you buy any Yahoo today?,' ” said Jerry Gramaglia, E-Trade's vice president of marketing, whose three children -- ages 10, 12 and 16 -- each have E-Trade portfolios. &quot;We offer the empowerment and thrill of being wired to Wall Street.&quot; Unless subscribers can only see &quot;Error&quot; messages. Which is why, on that Wednesday morning, Chrapaty's red eyes were fixed on a hideous number. Her monitor revealed that E-Trade's transaction &quot;per second&quot; ratio was well below normal -- and dropping fast. She deduced access problems. Chrapaty called in her &quot;super-smarts,&quot; a team of eight engineers who would root out the cause of the interruption. They laid out the symptoms, analyzed abnormalities and ate Krispy Kreme doughnuts. It took 30 minutes to isolate the problem, a software upgrade performed the night before. E-Trade technicians had added lines of computer &quot;code” to speed the process by which subscribers had their trades confirmed. But instead, the new coding infected the site with a most toxic side effect: no one could trade. Going Into Battle Lisa Nash arrived at E-Trade's headquarters at 7:30 a.m., Palo Alto time, on Wednesday. She made a beeline for the &quot;Wall Street&quot; conference room, where 12 executives crowded around a speaker phone as the Alpharetta team explained the situation. Nash wrote &quot;talking points&quot; for customer and media inquiries. This service interruption had nothing to do with site volume, Nash emphasized, and E -Trade could easily handle its daily trade capacity. What's more, only the trading portion of E-Trade didn't work; users could still find stock quotes, financial news and other data. A &quot;reserve team&quot; of 40 customer service managers was activated downstairs, near the cot where Chrapaty often sleeps. Chief operating officer Kathy Levinson appeared on CNBC early in the afternoon; Cotsakos was flying and essentially out of pocket. Chrapaty explained the technical arcana to Levinson over the phone. Chrapaty then watched Levinson a few seconds later on television. “ It made me realize how closely people were watching us,&quot; Chrapaty said, calling the experience &quot;slightly horrifying.&quot; Not as horrifying as what some E-Trade customers were telling company officials, reporters and online chat groups. Nash launched a crisis-specific e -mail address, which received 2,500 inquiries that day. She also led an outbound phone blitz, where company executives personally called hundreds of &quot;high-volume&quot; customers. Trading was halted for 75 minutes from 10:15 a.m. and 11:30 a.m., New York time. The technical staff placed a &quot;patch&quot; over the bad code to let trading resume. After the market closed, technicians removed the bad code entirely -- or so they thought. Chrapaty ordered pizza for her Alpharetta staff Wednesday night, while Palo Alto staff members ate sushi by the takeout bag. Nash, who did 50 interviews Wednesday, also took repeated calls from her daughters, ages 6 and 10. &quot;We saw E-Trade on TV,&quot; they kept telling her. Nash slept four hours Wednesday night, Chrapaty two. Code Red -- Again They awoke to a bitter deja vu. Trading was down again Thursday. E -Trade blamed Thursday's troubles on another &quot;aftershock&quot; from the Tuesday upgrade. Some of the troubled code had bled into the site's trading area again, although not all subscribers were affected this time. Whose fault was this anyway, a reporter asked? &quot;We don't do fault,” Chrapaty said briskly. Chrapaty, who is based in Palo Alto, had previously led her team in &quot;team- building&quot; exercises to instill battle readiness. Rope courses, trust falls and more elaborate sessions, such as giving out water pistols, donning a bull's -eye T-shirt and telling everyone to soak her. This drained aggression and promoted harmony, Chrapaty said. It also forged cohesion and composure during crises. Thursday's crisis lasted more than two hours. Nash's routine changed little, except that she skipped &quot;flag salute.&quot; But this morning reporters were asking about something else: The New York state attorney general's office had just announced it was investigating the online trading industry after &quot;dozens of complaints&quot; from customers. Such regulatory inquiries, while certainly unwelcome, invite an easy response to the press: &quot;E-Trade is cooperating fully with the investigation,&quot; Nash said in many variations Thursday. She emphasized that New York was investigating several online brokers, not just E-Trade, and that it was unrelated to that week's troubles. After the market closed, Chrapaty held a &quot;town meeting&quot; for her staff. She said she was proud of everyone, told a few they were &quot;champs&quot; and &quot;stud muffins,&quot; and gave out hats and T-shirts. Then she signed autographs. A Shock to the Stock Friday dawned with sublime quiet and smooth E-trading throughout the morning. It all exploded again at 12:20 p.m. with a 29-minute interruption. &quot;We believe we've got it all fixed,&quot; Nash assured Bloomberg News. &quot;But then again, we believed that this morning.&quot; E-Trade staffers complained that the press was picking on them. If they had a 29-minute failure six months ago, no one would have noticed. The attention was a perverse rite of the company's rapid growth and popularity. &quot;I find the attention frustrating,&quot; Chrapaty said. &quot;And also flattering.&quot; Either way, Feb. 3, 4 and 5 offered a cautionary lesson in the fallibility of Internet commerce coupled with the volatility of stock-market livelihoods. Wall Street's reaction was not flattering -- or forgiving. Shares of E-Trade closed the week at $48.93 3/4, down from the 52-week high of $62.43 3/4 it reached that Monday. The stock closed Friday at $40.12 1/2 on the NASDAQ Stock Market. Nash split her weekend between debriefing calls and a family hike. Chrapaty flew to Charlotte Saturday for her great-nephew's birthday, spent $1,000 on presents at Toys R Us, flew back to Alpharetta that afternoon and returned to Silicon Valley on Sunday. There, Chrapaty found a gift assortment of cookies and oranges, and some neighbors sent over roses. &quot;Listen guys,&quot; she said by way of thanks. &quot;I'm not dead yet.&quot;
Example of expenditure on generic software: Windows 2000. Launched Feb. 17, 2000 The long-awaited program has been more than three years in the making and cost an estimated $US1 billion ($A 1.58 billion) in development costs. The program itself is one of the largest works of software ever commercially marketed.
Software Engineering: Analysis and Design - CSE3308 David Squire [email_address] Room 5.23A B Block, Caulfield 9903 1033 101A , Building 26 , Clayton 9905 3295 (thanks to Martin Dick for initial development of course resources) CSE3308/DMS/2003/3
Similar to waterfall but uses a very short development cycle (60 to 90 days to completion)
Uses component-based construction and emphasises reuse and code generation
Use multiple teams on scaleable projects
Requires heavy resources
Requires developers and customers who are heavily committed
Performance can be a problem
Difficult to use with new technology
Rapid Application Development Business modelling Data modelling Process modelling Application generation Testing and turnover Team 1 Team 2 Team 3 Business modelling Data modelling Process modelling Application generation Testing and turnover Business modelling Data modelling Process modelling Application generation Testing and turnover