|
Tuesday, June 18, 2002
Bad Software - Technology Review
Why Software Is So Bad: "In an amazingly short time, software has become critical to almost every aspect of modern life. From bank vaults to city stoplights, from telephone networks to DVD players, from automobile air bags to air traffic control systems, the world around us is regulated by code. Yet much software simply doesn't work reliably: ask anyone who has watched a computer screen flush blue, wiping out hours of effort. All too often, software engineers say, code is bloated, ugly, inefficient and poorly designed; even when programs do function correctly, users find them too hard to understand."
Web Services - News.com
Web services: Ready, set, wait:
Maybe we should begin bottom-up with web services: start with the small and medium enterprises of the world who need it and low-cost "Lego-like" software more than the bigger companies.
China Software Market
China Software Market Statistics (packaged software, 2001): 1. IBM - USD 111 million - 6.8% market share Writes Cnet: "According to IDC, China's software industry was valued at $1.62 billion last year. By 2006, it is expected to increase to $7.8 billion at a compound annual growth rate of 36.9 percent from 2001."
Emerging Markets
| PermaLink
| Comments (1)
Viagra
The Digital Divide
Reuters on the growing Digital Divide:
I wrote about The Digital Divide recently in Tech Samachar. Emergic is all about bridging it.
Disney using Linux for Animation
Disney Shifting to Linux for Film Animation: "Disney's animation division is announcing today that it plans to use Hewlett-Packard workstations and data-serving computers running Linux for digital animation work in the future....The Disney move is the latest commitment by major studios and special-effects houses -- including DreamWorks SKG, Pixar Animation Studios, Industrial Light and Magic and Digital Domain -- to Linux."
Linux-Windows TCO Study
A study done by CyberSource on the Total Cost of Ownership for Windows and Linux. Linux comes out 25-34% cheaper for a 250-node network over 3 years. This is the viewpoint from developed markets. In emerging markets, the difference is going to be much greater, because the cost of support goes down. In fact, while in developed markets, the aim is to minimise cost of support, the emerging markets need a lower cost of hardware and software because that is the entry barrier. Salaries are much lower so support costs are lower. But the cost of hardware and software as a percentage of people's salaries is very significant. The approach to be taken in emerging markets is that fof a Tech Utility: provide hardware, software, training and support for a fixed price per month. Think of it as an AOL for computing. In fact, even the price should be AOL-like (USD 20 per month). That is the price point at which the market will take-off.
Software
| PermaLink
| Comments (1)
In addition to what you have mentioned, there are also costs for training staff to use Linux. Linux is not altogether as similar to Windows as we (techie guys) want to believe. When we look at it from a lay person point of view, we must also factor in training costs. Plus, these costs will not be ONE time. For effective training of employees, and future re-training of new employees, a helpdesk would have to be setup within the organisation. The organisation would then have to contend with the cost of maintaining the helpdesk too. The 25-34% cost advantage of Linux (on 250+ machines) could very well be offset to a great degree, by these costs. Posted by Clinton Goveas
Linux in the Enterprise
A white paper from MetaSource. A good overview of applications available on the server and the desktop. We are using many of these applications for our Thin Client-Thick Server project. Things change a little when one looks at Thin Clients, rather than Thick Clients. Thats the challenge we have been working on.
Software
| PermaLink
| Comments (2)
Buy Meridia Buy Cialis
TECH TALK: Rethinking Enterprise Software: Software Components (Part 3)
The software components that serve as the building blocks of the new enterprise IT architecture are Enterprise Core, Enterprise OS, Information Bus, Databases, Adaptors and Digital Dashboard. As we go up the stack (from the OS to the applications), software vendors should be able to take pre-fabricated modules and assemble them together. An analogy from the MS-Windows world is how easy Microsoft made communicating with the various devices through its API. In fact, someone once described Windows as “a collection of device drivers.” It may be too simplistic an explanation, but think of how the world was before Microsoft came up with its APIs. The enterprise software world is in that state today. The big companies can get their own custom implementations or build on software from the large vendors. The smaller companies are caught in no man’s land between the two options: one of them is to buy existing software and the other is to get software developed from scratch. Both are expensive and impractical from the point of view of the SMEs (small and medium enterprises). What's needed is a low-cost, integrated eBusiness suite built out of simpler components. Creating simpler modules for accounting, payroll, prospecting, sales force management can help in two ways. First, as more companies build these modules, options will be higher for the intermediaries (the software assemblers who are akin to the hardware assemblers – the “Genuine Intel Dealers”). Second, the assemblers can now take a 70-80% solution and then quickly build the necessary interfaces for the end- customers, speeding up the application development process. Linux and Java should be the two pillars around which the enterprise software components industry needs to be built. For too long, the open source industry has focused on the lower levels of the stack – as a result, we now have multiple, different Linux distributions, desktops, word processors, spreadsheets and browsers. This is a waste of time. Stick to adopting one of each – Red Hat as the Linux distribution, Evolution as the Email-Calendaring client, Mozilla as the Browser and Open Office for the suite of productivity applications. These become the components for the Enterprise Core. The focus needs to be on creating the enterprise software modules. That is where variety is needed – for different industries, different countries. The middleware (between the Enterprise Core and the applications) is where Java needs to come in. Writes Ganesh Prasad: “J2EE is an agent of commoditisation, just like Open Source. Like Open Source, it transfers power from vendors to users and creates a buyers’ market. In practical terms, what this means is that users can follow an inexpensive prototype strategy (by using Open Source implementations in development and switching to commercial equivalents in production) or they can follow a mix-and-match strategy (by using the cheapest components that are good enough).” What is important in the SME case is the ability to mix-and-match. For the dream of enterprise software components to come true, what’s needed is standardisation. This is the interesting part. For the first time, not only are there standards for web services but also for business processes. By using XML,SOAP, WSDL and UDDI along with J2EE, it is possible to build the Enterprise OS on which developers worldwide can build. This will open up markets for enterprise software hitherto untapped. Imagine hundreds of software factories churning out low-cost, interchangeable components, built to a standard interface. These factories should work like the entertainment companies producing TV serials (as opposed to films). Films take 2-3 years to create and are “monoliths”. That is what the enterprise software giants are today. They come up with the new products or major upgrades once every few years. TV serial makers, on the other hand, come up with something new every week. They can change course quickly if some experiments don’t work. The TV channels are the “assemblers” piecing together 30- or 60-minute entertainment components to create prime-time programming. The entertainment world needs both – the films and the TV serials. What SMEs want is the equivalent of TV serials, and that is what is missing from today’s enterprise software world. What’s needed next is an “integrated development environment” for business processes to be configured through a graphical interface. Tomorrow: Visual Biz-ic
|
I feel one major reason is that not many companies are following the software engineering principles as given in books on s/w engg. Also, if they do try to follow a few priciples, a fool proof Analysis and Design is a must, which takes a lot of time and effort as well as co-operation from people who understand the functioning of the system for which the s/w is being written.
Posted by pankaj udasinThere are just too many parties involved in the making of software, it is also noted by many s/w gurus that the specs which are given to the programmers are most of the time not upto the mark or are not interpreted wisely enough by the coders, which result in Buggy s/w.
I think a relatively transparent policy which involves Project Managers, Analysts, Coders and the functional experts from the client could produce better results.But, in order to implement this policy there should be excellent knowledge sharing tools available for them to share their thoughts throughout the development, rather than just at pre and post coding stages of Analysis , Design and Testing.
I find this hard to believe. The "gloom" talkers in the article(all Americans, by the way), generalise on software without specifying which type of software they were so concerned about. Through reading the article, thereader can infer that they are concerned with the s/w quality of the OS, basically, and are trying to link the OS' failure to the failure of software in general.
The author then, predicatbly, goes in for quite a long bout of Microsoft bashing. In all this, the author talks of "standardising" components to reduce errors in making new components every time. Sure, while we are at it, we might as well build cars with the "time-tested" bullock cart wheels!
Besides, Microsoft designed its operating system with the intention of making it compulsory to for consumers to use only MS products on the OS. Due to this SMART business decision, the operating system now shares components between progams in a very efficient manner, making each program more faster responsive in return.
Don't get me wrong, I'm not pro-Microsoft. However, I couldn't just standby and watch a group of software quality proponents, who for obvious business benefits to their firms, are trying to prove how BAD software is written in the world today. Obviously, these guys are not in tune with software testing in today's scenario of unit testing.
Also, there are much better "iterative" s/w development lifecycles, such as "Extreme Programming", in use today. Following the process guidelines of these development life cycles will produce better results than the older SDLC.
I do not believe that involvement of too many parties is the problem, it is the management of information across the parties that is to blame. And this does not require any "knowledge" tools, it needs proper channels for "communications".

Posted by Clinton GoveasAmbien
Posted by ICOSPropecia
Meridia
Propecia
Buy Cialis
Order Cialis
Meridia
Propecia
Cialis