Slashdot Mirror
IBM Turns to Open Source Development
mmmbop writes "'Is open source changing the way that software is made? It is at IBM. BetaNews sat down with Doug Heintzman, IBM Software Group's VP of Strategy and Technology, to discuss the adoption of a hybrid development model called Community Source that combines the best elements of the open source model with decades of IBM programming practice - avoiding a top down approach that IBM says could make Microsoft's Longhorn obsolete upon arrival.' A long read, but well worth it."
Sorry, got a system with 3500 users on 14 x86-servers (no clusters) geographically spread out, with about 200 *different* applications outside e-mail.
One main administrator, a few local part-time ones who only do things like create new users.
One developer.
Works like a charm. For sure, the Notes UI is Idiosyncratic at best, extremely frustrating at its worst, but for the kind of things it does well, nothing comes close!
They plan on employing all those people in all those labs. They just figure that there will be massive inefficiency (and they're probably right, seeing as they saw the problem years back, apparently) if they don't manage it in a way that mimics open source. It's not a complete open sourcing of all IBM's applications he mentions. In fact, he seems to refer to it in terms of open-sourcing the codebase within the company, rather than open-sourcing to everyone. I have read somewhere around here that the same kind of thing goes on at Google.
My little site.
When I read what he is saying they want to do, it sure sounds a lot like reading Fred Brooke's "Mythical Man Month". They are realizing that writting high quality, re-usable components with good documentation is very expensive. I believe "MMM", discussed this in the very first chapter. There are two orthognal qualities (I believe re-usable, and quality documentation, but it's a really long time since I read the book) that add a factor of 3 in each direction. If you wanted both it would take 9 times as long.
Stratigically, it makes sense to invest in creating those, if you have features that will be needed in enough different projects and areas. It sounds like they are planning on breaking down internal barriers, and providing highly re-usable, and well documented internally, and then ensuring that people know of it's existance.
Providing the source is a good idea. Ensuring that the fixes get moved upstream is a good idea. However, this sounds like good Engineering Practices (which I suppose is what a lot of "Open Source" advocates say you get, where as "Free Software" advocates, say it's purely an ethical issue).
It sure seems like this has little to do with "Open Source", and a lot to do with solid Engineering. It just so happens that Open Source has a lot of solid engineering behind it. It sure looks like a no brainer to re-use source you already have access to. If you are going to re-use it, it should probably be designed for that. If people who didn't write it are going to use it, it should probably be documented fairly well. Some how this seems fairly obvious, as opposed to, "we add features as we need them, to resolve some personal niche", which is the crux of "Open Source" according to CatB.
Kirby
4758 Secure Coprocessor Driver for Linux
This project is a Linux device driver for the IBM 4758 PCI Cryptographic Coprocessor, which is a tamper-sensing and responding, programmable PCI card. It provides a highly secure subsystem in which data processing and cryptography can be performed.
ATM on Linux
ATM support for Linux is currently in pre-alpha stage. There is an experimental release, which supports raw ATM connections (PVCs and SVCs), IP over ATM, LAN emulation, MPOA, Arequipa, and some other goodies.
Abstract Machine Test Utility (AMTU) for Linux
Abstract Machine Test Utility (AMTU) is an administrative utility that checks whether the underlying protection mechanism of the hardware is being enforced. This is a requirement of the Controlled Access Protection Profile (CAPP) FTP_AMT.1.
Ananas Project: Summary
This is the source for Working XML, a column on developerWorks with companion project code that demonstrates the evolution of full-fledged XML applications. This is distributed under the artistic license.
Apache HTTP Server
The Apache project develops and maintains an open-source HTTP server for various modern desktop and server operating systems.
BlueHoc simulator
BlueHoc is a tool that predicts the performance of Bluetooth wireless hardware technologies. BlueHoc simulates the baseband and link layers of the Bluetooth specification.
COIN (Common Optimization INterface)
Developers can use Common Optimization INterface (COIN) to build optimization solutions. IBM mathematical optimization researchers opened the code they use in finding the optimal allocation of limited resources. The code has many applications in a variety of industries.
Channel Bonding
The Channel Bonding project works on methods to join multiple networks on Linux into a single logical network with higher bandwidth. The project team works with the Beowulf Ethernet Channel Bonding project, where bonding work began.
Consensus prototype
Consensus is a joint European project carried out by six companies. The project is partially funded by the European Commission. The project goal is to provide technology to support single-authoring for mobile devices. developerWorks hosts the open source implementation developed by the Consortium. Detailed information about the project is at the Consensus Project home page (http://www.consensus-online.org./
Content Query System (CQS) Project: Summary
Content Query System (CQS). CQS is a distributed peer-to-peer query system for the purpose of discovering content or data. XML messages are passed between systems and query "engines" are used to access the data that is being made available on the system.
Crypto Accelerator Driver
Device Driver Support for the IBM eServer Cryptographic Accelerator.
Crypto Interface Library
Generalized Interface library for the IBM eServer Cryptographic Accelerator Device Driver. Note, this is a low level api for the Specified adapter, it is not intended to be an interface which is written to by applications. Applications should use the openCryptoki PKCS#11 api for interfacing to the token.
Dynamic Probe Class Library (DPCL)
DPCL is an object-based C++ class library that allows tool developers and sophisticated tool users to build parallel and serial tools using a technology called dynamic instrumentation.
Embedded IBM PowerPC 4xx Linux Support
This project contains packages which enable add
Err, if they open sourced their big moneymakers, wouldn't they go out of business and then be unable to help us at all because their consultancy business was hurt by the loss of direct revenues from the licensing fees?
No, I doubt their motives are "pure" in the sense that they're not trying to get ahead of the competition. But so long as our interests are aligned, the fact that they have financial incentive (even in terms of hurting their rivals) is precisely *why* we can trust them to help us--because IBM will do what's good for IBM. And if that's good for us, too, it's best to be quiet, take advantage of that while we can, and make sure there are contingency plans in place in case that ever stops being true.
So I don't understand the complaint. No, it's not done out of purely altruistic motives, I suppose, but just why shouldn't we make the most of it while we can? Now if you see evidence that IBM's interests are beginning to diverge to a point where they'll start acting against OSS, by all means, gripe. Until then, doing so seems counter-productive, at best.