Currently, Internet fires (Spam, Viruses, Worms and DDoS) are being fought by reactive software and actions, and proactive efforts in new software designs that will prevent current firestarter techniques from working. Between the incomplete coverage of reactive techniques and the long lead time of proactive techniques lies a huge gap of vulnerability that we handle on a mostly ad hoc basis. We need to do better.
This article covers some basic steps we can take now. It also identifies more advanced steps to close the holes that Internet rats crawl through. Read Firefighting On The Internet to see the big picture, what we can learn from real firefighters, and find out what simple steps you could take to put out any local fires.
OS/2 and eComStation will get new support. The announcement of an OS/2 version of OpenOffice as well as that and other demonstrations will be available. Read more about Warpstock 2003 in OS/2 Updates.
Read about the powerful Opteron design and find out why Hans de Vries thinks Opteron is a real threat to Intel's high end processors. Read a summary in New Processors with links to the detailed analysis at Chip-Architect.
BBC and the Library of Congress are getting into digital libraries in a very big way. BBC is planning to make all of the BBC Archives available to the Internet, and the Library of Congress will next year host a Moving Images Collection, currently under development. In another field, the National Institutes of Health have announced the development of a Molecular Library, that will provide base information for development new drugs and nano-scale agents for an emerging "era of personalized medicine." Read more about all three digital libraries at Digital Library Updates.
Appro, a cluster and super supplier, has reduced the cabinet size for small cluster needs. It looks just right for a growing small business or university lab. Read more about this in Supercomputer News.
OS/2 and eCS users now have another way to handle MS Word documents. Rather than load Star Office, read about another method in OS/2 Updates.
The National Science Foundation ITR awards are made every year from a large set of proposals submitted from universities, corporations, non-profits and government agencies. This year, NSF awarded more than $169 million in new awards through the ITR program. In addition to the eight large projects, more than 175 mid-sized projects have been awarded up to $4 million for three to five years, and 180 smaller projects will each receive up to $500,000 for up to three years.
Large ITR projects focus on long-term innovations through coordinated research and education efforts at the intersection of computer science and other science and engineering fields. The eight large ITR awards were between $7.5 million and $12.5 million over five years. The projects were selected by merit review of more than 2,500 proposed projects.
Read more about the eight large ITR Awards. I think you will find "Predicting High Impact Local Weather" (big storms) and "100 Megabits per second to 100 Million Households" to be relevant to our future.
Real Reasons is Part I of a two part article addressing the problems that need to be solved in order to introduce a unix class (SCO, Linux, BSD, Darwin, OS X) system in a Microsoft Shop. Many of the problems are hidden in people's experience and assumptions. Part I identifies where and what these are. Read Part I in the Analysis and Commentary section.
IBM is at it again. Not content with building Cell chips for the Sony Playstation3, IBM is researching how to build chips that can perform a trillion operations per second. Read about 32 nanometer scale chips in New Processors.
In an excellent article on the Ace's Hardware site, Paul Murphy writes about the real disconnect between the way MSCE staff think and the right way to implement Linux (any unix) for maximum savings. He covers the real basis for why using Microsoft in an enterprise is so expensive and how to reduce the overall costs. Also worthwhile is his book "Unix Guide to Defenestration" and articles on TCO. Recommended for anyone looking for ways to reduce costs and increase stability. Read about Mad Hatter, Unix Guide and TCO.
In a step towards FTL (Faster Than Light :) processor interconnects, DARPA has awarded a four year, $30 million research effort to IBM and Agilent Technologies to reach the terabit per second level.
"We are looking beyond cabled rack-to-rack technology to very high levels of integration and to do that we need [optical] modules about two centimeters square right on the board next to the microprocessor. We will increase per-channel speed, but probably our most aggressive goal is to reduce power to 5-to-10 milliwatts per gigabit," said Marc Taubenblatt, senior manager of optical communications at IBM Research.
IBM and Agilent have even faster goals for the longer range plan - 40 terabits/sec in 2010 (only 7 years away). Read more in Supercomputer News.
SCO's lawsuit against Linux and the users of Linux has captured a lot of media attention and generated reams of speculation. SCO's objective is not clear, but there are already a number of effects in the server market with more to come. I've taken my own look at this situation and projected some possible outcomes.
If you evaluate this unprecedented attack against Linux as a publicity exercise, well it has to be called a resounding success. It's short term value for SCO's business is questionable, and it's impact on SCO's popularity has been very negative. I find it difficult but not impossible to believe that SCO's executive suite could not or did not see much of this coming.
You may be surprised at what SCO has in mind, and what outcomes are possible. Read more in "SCO vs Linux: What Happens Next?."
The state of the IT industry seems to be improving in a few areas. Semiconductor sales have inched up and experts are hailing this as the beginning of a recovery, the third annual prediction since 2000. I disagree with this optimism, based on the continuing announcements of layoffs and outsourcing.
The reason for this is the dot com bust has rippled through the economy, but we have yet to experience the housing bust, now just starting, or the credit bust whose leading edge is showing in the record number of personal bankruptcies. While the economy has mostly weathered the dot com losses, the housing and credit losses have not yet had much impact. But our current economy no longer has the reserves to tolerate both housing and credit problems that are coming, and the expense of keeping troupes in Iraq aggravates the current and future problems.
I wrote about the economy in more detail in Part I, and consequences and things you can do in Part II. If you haven't read it, take it as an antidote to the published optimism. I don't expect a depression, but even by comparison with current times, it could be quite difficult. Part II has some simple steps you can take to reduce your exposure and protect what you have. Of course you can go much further than these steps, but they are simple and free.
The merged HP and Compaq, now HPC, is a specific example of the short term mentality that is a major cause of our economic problems. In an effort to increase profits, Carly continues to cut employees from the merged company. Some redundant staff needed to be cut, but Carly has gone far beyond that and her actions are now setting the stage for a future HPC reorganization under the bankruptcy codes. By then Carly will have earned her multi-million dollar performance bonuses and be mismanaging another company.
Soon after the merger, I wrote "HP and Compaq - The Dust Settles." Since then, I've watched the reports of layoffs, outsourcing, reduced support for VMS and True64, and internal leaks about the real situation. The outcome is no longer in question, unless Carly and HPC do a fast return to former practices and accept the necessity of long term planning for solid payoffs.
The only good likely to come out of current mismanagement of HPC is as a bad example for others, and a number of high sounding Harvard Business Review articles. The employees, stockholders and customers will all be paying for Carly's mistakes for a long time. It is way past time for this kind of company looting to stop, and the HPC board is the only place that has that power. I don't expect them to act in time.
Two other facts to ponder. I saw this same situation happen at Memorex in the late 1970s and early 1980s as an employee. A new president was brought in with a $5 million bonus if he met certain goals, and 1 million shares of stock. He cut R&D and engineering, covered up internal quality problems and met his goals. He retired with over $30 million from stock sales, plus the bonus. One year later, Memorex was out of business, with a few parts sold off. This did not have to happen - Memorex products were very competitive.
The second item is an article I read this morning on the subject of HPC. The subtitle says it all - "Floggings Will Continue Until Morale Improves." Read Charlie Demerjian's look at HPC. "HP: A billion here, a billion there – pretty soon it adds up."
Virginia Polytechnic Institute and State University are building a big Power Mac G5 cluster, with an estimated 10 TeraFlop performance. Check it out in Supercomputer News.
Wonder why your job went to India? Check out these salaries at The Inquirer. I've also added another vendor to the Cluster Tools in the Supercomputer Index.
Monday morning, with over a dozen pages updated, I began the upload process. It went nowhere. Several experiments did not clarify the real cause, so I made my first mistake. I decided that Quanta was hung after months of use and shut it down. Restarting Quanta caused "No Mime Types Found" error, followed by a sig11 with no traceback. I proceeded to my second mistake. Figuring Quanta needed to be updated, I did a 'portupgrade quanta' and loaded the new version. Same problem.
Not satisfied with two minor mistakes, I trumped them by making a real whopper. I did a 'portupgrade -a'. This upgrades all of the ports in my system. Not only had I not bothered to check the count (over 240), but I had forgotten that several of them were really big. As in KDE3, ATLAS and a few others, each of which was good for several hours of compile time. Tuesday afternoon I fell back on FTP to update the site pages, really clumsy compared to Quanta.
The portupgrade process was also delayed by halts while the system waited for me to make option selections on three applications. All in all, the upgrade didn't finish until sometime Wednesday morning. Once again, I loaded Quanta. Same crash result. I was really confused, so I tried to find out why the kio was at V60 instead of 61. Searching gave no answer, so I fell back on my last resort - thinking.
Finally the light dawned. I had upgraded everything, but only loaded Quanta. To start with a clean slate, I needed to exit KDE and X, and restart them. I did a logout reluctantly, and waited while the couple of dozen apps closed. 'Startx' reloaded X and KDE, and KDE reloaded my applications cleanly. Sigh of relief. Now for the critical test - Quanta. It came up without problems, reloaded the dozen pages I had updated and was ready.
Now all I had to do was explain what happened. The original problem? During a web host upgrade, the ftp access file had not been restored. The real problem was not anywhere near my system, just in my head. Maybe this should have been titled as Stupidity Attack Delays Updates. :-}
Inspired by the close passage of Mars, I spent some of my weekend time updating Science Links to the new Science Index, in the new standard index style. Yes, there is a link in the Physics section under Exophysics to Hubble's amazing close encounter pictures. This is the largest single set of updates since I started BillsWrite.
OS/2 versions of Mozilla 1.5b and Firebird 0.6.1 have just been released. The OS2BBS site has much to offer as well. Details in the renamed OS/2 News and Updates.
The current Site Journal was trimmed back to start in July, with the older site journal material linked at the end of the current one. Each journal is linked back to the one before, all the way back to December 2001, when this site started. It's been an interesting journey. Feedback and suggestions from readers is always welcome - just use the 'Contact the Author' link in the left frame. Thanks.
DOE's Pacific Northwest National Labs just had its Itanium supercomputer upgraded with 1,400 Itanium Madison 1.5 GHz processors, making it currently the world's fastest Linux super. Performance jumped from 6.2 TeraFLOPS to 11.8 TeraFLOPS. Read more at Infoworld.
Wired has a good article on the impact of MIT's Open CourseWare - Lectures and tests from their courses available free on the web. This has good and not so good implications. It's good for those of us who want to become more knowledgeable, but it also increases the information going overseas that is hurting the economy. Read the Wired article, and visit the MIT OCW site.
The early morning of Wednesday was Mars closest approach, but it isn't moving away all that fast. Still good viewing and picture opportunities. Give it a try this weekend if you're one of the sensible people who stay near home during holidays.
Have you noticed the annual holiday gas price increases? No? You must not have a car. We don't have to worry about the oil companies going bankrupt. That oligarchy of oil producers who, coincidently of course, just happen to raise prices before and during holidays that involve heavy travel don't have to worry. But the car drivers do. Get used to $2+ gas and don't expect any government investigations to change that.
I'm not changing this web site into a political soapbox, but I can't help speaking out at the obvious abuses by the mega corporations on the citizens at large. Have you wondered why the auto companies are all pushing the gas guzzling SUVs? Is there a connection with the oil companies? I always know when I've pulled into a gas pump that just filled a SUV. The total amount is always over $40. When it's over $100, I figure it was an RV.
Don't expect any relief from our supposed representatives. Another $20 million in contributions will calm any outcry from government. Maybe we need to send new people in.
The future is always uncertain, more so now than usually for the US. The current and future effects of shipping jobs overseas will haunt us as we try to restructure jobs and save companies.
Meanwhile, individuals and families suffer the effects of this problem while having little control over their environment. Part II of this article ends with a list of steps people can take to meet the future with confidence. It's called "Planning: One Step at a Time." Prescription - Take these steps, one each weekend for relief from the stress of uncertain economies.
Read Part I first, then Part II.
Admittedly this is an area outside of my normal writing, and outside my area of technical expertise. But the problems are obvious and plentiful, and the probable consequences available to anyone who collects the data and applies logic and knowledge from similar events in the past. In short, most readers could duplicate my data and reasoning and I recommend exactly that. Don't take my word for it - it's too important for each of you and our country. Please write me if you have new or more current data.
Part I listing the problems is in place. Part II, discussing consequences and alternatives, will be soon. Here is an excerpt from the Overview.
"If you haven't noticed that the news is full of stories about companies laying off (firing) thousands of workers while they ship jobs overseas, you must have spent the last year on Mars. The employment picture for high tech is particularly bad, but we are not the only class of employees whose jobs and income are vanishing. In general, the service sector of large companies and much of manufacturing is vulnerable too."
Read the rest of Part I in Bill's Write.
Next Wednesday, August 27th, (corrected) will be Mars closest approach to earth in the last 5,000 years and for the next 60,000 years. A friend tells me he can see the ice caps now with just his binoculars. Mars appears near the moon as it rises. This is your chance for some great shots if you have a telescope or a camera with good telephoto lens. I'm not going to wait around for the next one.
IBM has detailed some of the design and reasons for it at the Hot Chips Conference. See our New Processors page for more. You can also check out IBM's 100 Teraflop super planned for LLNL in Supercomputer News.
Sun announced Gemini, a CMP dual processor on a chip that draws a max of 32 watts. This announcement introduces a new page that will cover significant new processor announcements outside of the commodity x86 processors. See our New Processors page.
Four ways to build faster supers is outlined in the Supercomputer News page, from the Hot Chips Conference this week. Read more in Supercomputer News.
Despite appearances, BillsWrite is not becoming a supercomputer announcement site. It's just that there has been a lot of activity in the large cluster and supercomputer area recently. There will be other subjects here Real Soon Now (RSN). Honest.
The new AMD Opteron chip is a clear hit with makers of large clusters and supercomputers. The two new supers will be built for Los Alamos National Labs. The large system will have 2800 processors and 11 Teraflop performance. Read more in Supercomputer News.See Supercomputer Index Cluster Hardware for updated information on clusters and interconnects.
Two articles in EE Times report that three government studies will recommend increasing funding for supercomputers. Both large arrays of commercial microprocessors and custom chips and designs have their place, according to the reports. Japan's powerful Earth Simulator, at 35 Teraflops, has gotten the attention of designers and users of supers, and now the government weighs in.
Overall, this is good news because of the benefits of projects like BIRN (Biomedical Informatics Research Network) and the NVO (National Virtual Observatory). Both these and a large number of other projects will accelerate the delivery of medical and scientific discoveries that benefit all of us.
More information and links in Supercomputer News.
I had to check this one out to be sure. The folks at ReactOS are building an NT compatible OS in Open Source, and have a base command line system up, with a release planned for the end of August. Compatibility with NT drivers is part of the plan.
The reason this is interesting to me is that I am in the process of slowly phasing out my NT box, now running just one app that requires Windows - Pagemaker 6.5. Just recently I moved from IBM Top Page on NT, an obsolete web site tool, to Quanta Plus on FreeBSD. Pagemaker is the last reason I run NT, my only Windows box. ReactOS will let me run other Windows based software without being tied to Microsoft, a really useful option. Kudos to the ReactOS people.
For those not familiar with FreeBSD, it is a powerful unix design OS with 20+ years of history. It supports over 9000 applications that run native in what is called 'Ports', plus Linux and SCO ABIs (Application Binary Interface) which will run those binaries directly. I've used Linux, but prefer FreeBSD as a workstation/server, and OpenBSD for a firewall. Check out the BSD information, and take a look at ReactOS.
A company named Golden Code Development Corp has announced the release of its Sun Java2, version 1.4.1 VM port (J2SE v1.4.1 in Sun's Marketingspeak) to the IBM OS/2 operating system. Read about it in OS/2 Updates.
Terrasoft and Marathon have teamed up to deliver Mac supercomputers. More in Supercomputer News.
Supercomputer News (SN) is now long enough to make it difficult to find resources without scrolling slowly down a long file. As the supply of supercomputer, software and vendors of same grows, the problem will only get worse. In addition, SN has resided in the Science section where more support for science information is planned.
Organizing and expanding supercomputer information in a new directory will help solve both issues. The old SN will continue in a new space, and will be linked from a Supercomputer Index, which will contain a wider range of links and data about supers.
You will see me user the short word 'supers' to mean supercomputers more frequently in the future. So take a look now at the new Supercomputer Index and the new old Supercomputer News. Of course there is YASA (Yet Another Supercomputer Announcement).
Adobe's version 4 PDF generator has a subtle problem. It defaults to using managed colorspaces even when that is not needed. This causes colorspace errors in Acrobat Reader version 3 which cannot handle managed colorspaces. Because it is a default setting that is not reset by selecting V1.2 format for Acrobat 3 compatibility, a large number of PDF files are partially or wholly unusable in Acrobat 3.
This has been a major problem for OS/2 and eCS users because Adobe never went beyond V3 for OS/2. Until now, the only answer for this was to install Ghostscript and use that under OS/2 with emx.
InnoTek, as of July 29, has released an OS/2 kit for Acrobat Reader 4.05. For more information on how to avoid PDF colorspace errors and links to the InnoTek site, check out OS2 Updates. This isn't open source, but I'm awarding InnoTek a Software Gem Award for this support of OS/2.
The precipitous decline in the fortunes of the Mozilla Project has a lesson worth learning. Entirely aside from illegal Microsoft bundling of IE, Mozilla might not have been successful even in a normal competitive environment.
Consider what they had going for them:
Why then did they fall so far, to about 5% of browsers? It wasn't just Microsoft's typical tactic of bundling. Bundling, in this case is equivalent to selling the product under cost, which is a violation of the Sherman antitrust act. I think Mozilla had major problems with lack of Focus and excessive Time To Market.
Their lack of focus caused them to exhibit what Brooks' book "The Mythical Man-Month" called "The second system effect." This is a situation where after successfully creating a first version of software, caution is tossed out and every enhancement, hook and gewgaw is included in the second release, regardless of cost or value. The result of this is a predictable explosion of complexity and the resulting delays, costs and bugs.
Their problem with excessive Time To Market is that it assumes that your competitors will take just a long, not valid in software anymore. This also reduces the sense of urgency that usually restrains the introduction of too many new functions.
I think it is clear to most firms that success with a product requires Focus, both on the product and its market. In addition, it's really hard to grow success in software when you are not one of the early entrants. Software is a tough environment - when you come late to the party, all you get is leftovers.
As I said earlier, by next year supercomputer announcements will be listed as bullet points. I'll start the process here, and list some details and links in Supercomputer News
In a June Netcraft survey, the five lowest failed request systems were running FreeBSD. In all, ten of the top 50 sites used FreeBSD. The great majority of the other sites were powered by Linux, Solaris and Windows 2000.
I've started to make changes to the web site to make it easier to find information. My first step is to add a new Topic index to the Byte Columns section. This is in addition to the annual indexes, and is by topic then date. Each entry has a brief excerpt and a link to the article. There are ten topics ranging from Changing Times to Supercomputers.
I've changed the Byte Columns link in the top frame to invoke the Topic Index. The annual indices are still available in the link frame on the left. I've also added the 2003 Index and the February column on Computational Science.
Temporarily, the return indexes on the articles will still bring up the annual indices, but the annual indices will have links to the Topic Index. These links will get added to the articles as work continues on the site. Click on the 1999-2003 Topics in the top frame and have a look.
This one is a little off the beaten path. It answers the question "Are game machines really powerful?" Find out more in Supercomputer News.
You may have noticed that most of the recent Journal updates are about supercomputers of one kind or another. There are a number of reasons for this.
The range of supercomputer capabilities has reached a critical mass and I expect it to accelerate substantially in 2004. By then I expect such announcements to be so common they just get listed as bullet points.
Two new announcements today are based on AMD's Opteron. One is a 10 Teraflop system the Chinese will build, the second is a 40 Tflop system for Sandia National Labs built by Cray. These systems and the recent Appro cluster announcement make the Opteron the newest hot supercomputer chip. For the full list of reasons why this is happening and these two supercomputer announcements, check out Supercomputer News.
In my June column for Byte, I listed four open source gems which are outside of the operating system area. These four are in addition to the well known Star Office and Open Office and desktops KDE and Gnome. They are:
For overviews on these four and links to the sites, you need to read the whole article on Byte.com in the June Advanced Software and Technology column. Byte charges a $20 fee for a one year subscription. It's worth it even if you don't read my columns. :-}
In addition to these four, I am adding a fifth to the list:
This product sets new standards for flexibility. Originally an HTML web editor, Quanta has been extended to edit for almost any DTD, including SGML and XML. The project, template and editing facilities are powerful and well organized.
Short of emacs, you won't find any editor as extendible and customizable as Quanta Plus. Recommended.
This is the first large Opteron cluster that I have seen published. What is especially interesting is the very dense blade arrangement and the use of the new 10 gigabit Infiniband technology.
While HPC clusters have been the major source of new supercomputer systems, I predict that Blade and Infiniband technology will replace traditional clusters in many areas. Take a look at the future in Supercomputer News.
Computers may calculate complex results to difficult questions in physics, chemistry and medicine, but until a human can understand the results, it is of little value. So the final step of machine to human is vital, though often difficult.
Enter now the new visualization systems from SGI. Begin with desktop or deskside systems that drive a mere 20 million pixels. Grow up to 64 MIPS processors driving multiple high end ATI Radeon GPUs which can drive up to 100 million pixels in real time for full immersive systems. All of this delivered at a cost/performance improvement factor of 40!
More on this in Supercomputer News. Think of the high end system as a prototype for the HoloDeck in Star Treck: TNG.
While I am very careful not to fall into the conspiracy theorist trap of "No evidence = a conspiracy of silence," I was impressed with the thorough reporting and evidence in a report from New Zealand called "Sludge Report #154." In this they provide a convincing case for manipulation of votes in a widely deployed electronic voting system. Here is a link to their introduction, and another to the full Sludge #154 report.
In addition, here is another link to our government suppressing free speech in reporting from Iraq. Not exactly the things the US is supposed to be famous for, is it?
"Justice's Final Transmission From Baghdad
1 July 2003, 4:02 pm
Opinion: Scoop Link'Gag' order contradicts U.S. value Iraqis like
Judge Gilbert S. Merritt
By GILBERT S. MERRITT For The Tennessean"
Dr. Kaku talks about physics and the nature of the universe with an unusual clarity. He manages to take complex physics concepts and make them clear to non-scientists. Here are a few of his concepts from the Guest of Honor talk.
Later this week I'll be adding notes from Dr. Kaku's panel on Time Travel, plus an excellent overview by two scientists from JPL on past, present and future robotic explorations.
The Linux Networx Evolocity(tm) system, called MCR by LLNL, can process 7.6 trillion calculations per second (teraflops) running the Linpack benchmark. Only two other supers are faster as of this date: The Earth Simulator in Japan and ASCII White, an IBM SP Power3 system at Lawrence Livermore National Labs.
Three scientists at Los Alamos National Labs have increased the search speed of BLAST (Basic Local Alignment Search Tool), an Open Source project, by up to 140 times by splitting the DB over multiple computers, eliminating disk accesses.
Really, really big supers, like the planned Q at Los Alamos, have set new size records. Three megawatts for power, two megawatts for cooling (about 5,000 homes worth), 330000 square feet of floor space, costing $93 million. Yikes!
Compare this to Green Destiny. Cost $335,000, size 2 x 3 feet by 6 feet tall. No special A/C required. Read more about these supers in Supercomputer News.
Contact Info: To contact the author, click on the mailto link below.