Re:For most problems...
on
Computer Voodoo?
·
· Score: 4, Interesting
True, and throwing out equipment fried by lightning doesn't hurt, either. Seriously, what kind of questions are these? Hardware doesn't work? Isolate it and replace it. Other than freezing an old hard drive to free up bearings to get data off before you throw it out, it's not worth the aggravation.
Of course, as far as real Computer/Equipment Voodoo is concerned, there is always the Heisenbug. Just had a mechanical version of this today, the Bayer tech has spent 3 days on a machine to isolate a pump problem. To see the pumps, you have to open a panel either on the side or the front of the instrument. The past 2 days, he was working on it through the front, and the problem didn't occur. Today, after being called back because it happened again, he opened the side panel to watch it, and accidentally bumped the front panel while he was looking at it. As soon as the front panel closed, the problem occurred. It turns out that a zip-tie that holds some tubing from the pumps together was caught on the front panel, and when the panel door closed, it pulled on the zip-tie, which pulled on and pinched the tubing, causing a pressure sensor to throw a fault.
For Windows, use roaming profiles and default installations. For Linux, rsync works quite well for the base OS (say, a staggered start time at night based on IP), with OpenLDAP doing auth and home dirs stored on central server(s) and mounted via NFS. New system setup becomes a knoppix CD, partition the new drive, format partitions, mount them, rsync the distribution to the new machine, chroot, and setup boot loader. You could of course script all of this, and this is very similar to what I do for kiosk systems (Linux/Firefox setup), except the kiosks don't change, so it's just a big tarball via sftp instead of rsync. You could also do tarballs, and keep the last few versions as backups in case you screw something up. If the hardware is identical, use the distribution of your choice, but if there are several different systems, you may want to use one with good hardware detection (like knoppix).
VisiCalc was not included because it was buggy and lacked important features, but I have to agree with you that Lotus 1-2-3 belongs there, since Excel is buggier than VisiCalc was. Java belongs on the list for exactly the reasons specified in the article, it completes the source code portability concept started with other languages by adding cross-platform BINARY portability through the use of an intermediate language and VM. While Java may have its roots in C/C++, Smalltalk, and large included-library languages like Ada, Java has become the basis of and benchmark for all new general-purpose languages, including everything that has come from Redmond since the introduction of Java. Things like RIM's Blackberry and Microsoft's.NET are the best examples of why Java is great software, since the language is essentially irrelevant, but the VM concept and large, standardized, and stable core library are the reasons why Java has the influence it has. Simple ideas that are taken for granted now were revolutionary (although often originally introduced 30-40 years before) concepts when Java was introduced, like Swing, RMI, and the security manager (including the sandbox for browser applications).
To avoid this problem, you evaluate the tools you have (floats, double, fixed decimal, int, long, possibly BCD, etc) and see if they match your needs. If not, go get better tools (an arbitrary-precision number library, like java.math.BigInteger or java.math.BigDecimal). Similar libraries exist for just about any language. Decimal arithmetic is inefficient on binary systems (except for hardware-implemented BCD as in IBM S/370), as well as completely unnecessary, since the binary arithmetic functions already exist in hardware, and storing individual decimal digits will always take more space than storing the whole number in binary.
The RPG language also had an implementation of BCD, and probably any compiler for the IBM S/370 line would at least have a library for it, as I believe the IBM mini/mainframe architectures had implemented it in hardware.
no, although a double or larger increases the precision, all floating point-style numbers suffer from this problem. You might get away with it for a larger number, but you are wasting bits, and stil cannot guarantee that the number is always going to be accurate. In addition, as the number gets larger, more precision is lost. For instance, 782533.37 as a float is 493F0C55, or 782533.31, and as a rounded float is 493F0C56, or 782533.38. Neither would be acceptable, and even a double will lose precision at about 16 significant digits, which could be a problem for daily interest calculations or another high-precision calculation. What should be used is a packed decimal (BCD) or fixed decimal (int or long and an associated scale factor). The whole point is that for a financial institution, there is likely a fixed precision level that is acceptable, but floats, doubles, etc., cannot guarantee that any given precision will be maintained, as it depends upon all of the factors involved (value and size of each number in the calculation). For the same reasons, floating point numbers should never be used in program flow control, either, as doing a for(float x = 0.0; x 1.0; x = x + 0.01) will iterate 101 times, since the first time x is not less than 1.0 is when it reaches 1.00999..., and if you are looking for a specific value, it may never reach it (like for(x = 0.0; x != 1.0; x = x + 0.01)). If you instead assume that you will always deal with 2 decimal places (or 3, or whatever), you can guarantee that your addition or multiplication will be accurate, and can scale the answer later if necessary.
For the uneducated, the reason that this is stupid is that IEEE-754 floating point numbers cannot REPRESENT all values, they APPROXIMATE them. There is no way to properly represent the value 0.01 as a float (0.01 is best approximated by 3C23D70A, or 9.9999998e-3). So, for instance, if you were to add up 100 pennies, you would have 99.999998 cents, not 100. Repetitive additions (like credits and debits from an account) or multiplications (interest calculations, amortizations, etc.) simply make the problem worse, which is why floats should NEVER be used to track money. A fixed decimal system should always be used for financial systems.
You mean like these:
SCAT Sacramento Center for Assistive Technology
SCAT Sarasota County Area Transit (Florida)
SCAT School & College Ability Test
SCAT School and Community Action Team
SCAT Scout/Attack
SCAT Screen Capture and Transfer
SCAT Secure Campus Anonymous Tip
SCAT Security Control of Air Traffic
SCAT Senior Citizens Against Crime
SCAT Senior Citizens Against Telemarketing
SCAT Sensor Coverage Analysis Tool
SCAT Sewing Circle and Terrorist Society (women's group)
SCAT Sheep Cell Agglutination Test
SCAT Ship Certification Acceptance Team
SCAT Shoreline Cleanup Assessment Team (oil spills)
SCAT Shrewsbury College of Arts and Technology
SCAT Single Channel Autonomous Trunking
SCAT Somerset College of Arts and Technology
SCAT Sonar Calibration Alignment Training
SCAT South Carolina Association of Taxpayers
SCAT South Coast Agility Team (dog agility club; Costa Mesa, CA)
SCAT South Coast Aquatic Team
SCAT South Coast Area Transit (California)
SCAT Space Coast Area Transit (Brevard County, Florida)
SCAT Space Coast Area Transit (Florida)
SCAT Special Community Action Team
SCAT Specialized Community Area Transportation (service)
SCAT Sport Competition Anxiety Test
SCAT Street Cleansing Ambulance Truck
SCAT Street Crime Assertive Team (East Orange, New Jersey)
SCAT Street Crimes Arrest Team(s) (police unit)
SCAT Student Coalition Against Tobacco
SCAT Student Community Action Team
SCAT Sub-Category (US Navy system for categorizing test instruments)
SCAT Subcategory
SCAT Submarine Classification & Tracking
SCAT Submersible Cable-Actuated Teleoperator
SCAT Submersible Concentric, Axially Traversing
SCAT Supersonic Commercial Air Transport
SCAT Support Center for Advanced Telecommunications Technology Research, Foundation
SCAT Surface Controlled Avalanche Transistor
SCAT Suspense Control and Automated Tracking System
SCAT Switch Consolidation Action Team (Sprint)
SCAT Synchronized COA Authoring Tool
SCAT System Calibration and Test
SCAT System Commonality Analysis Tool
SCAT Systematic Causal Analysis Technique (systematic method of analyzing the basic causes of accidents)
SCAT Systems Consolidation of Accessions & Trainees
Oddly enough, it doesn't seem like anyone outside of Silicon Valley knows what a dictionary is.
...and don't forget Sun, if you feel like paying $20,000 for an 8-core Ultrasparc T1 chip that uses less power than an Opteron. Windows not allowed though. But since the original question is probably talking about managing a datacenter full of customer machines (which you can't control, unless you lease them), I would imagine that just changing the way you do cooling would make a rather large difference. Have you considered piping heat from the racks to the outside through the roof, and using a high specific-heat fluid (say water/ethylene glycol mix) that could be prechilled to an extremely low temperature during off-peak times, then used in a radiator-type arrangement for cooling. I would say that sealing the sides of your rack and replacing the front and rear doors with liquid-cooled radiators/heat collectors should work quite well. If you can pre-chill the air entering the front of the machines, it will be cooler coming out, and if you absorb the excess heat coming out the back, less heat will enter the room, meaning your air conditioners won't have to work as hard. Other good option would be to duct the exhaust heat (if it is significantly above room temp) up and out of the building with a few smaller fans.
Anyone else read that and think of the scene in the South Park movie where Gates comes in and says how much faster the new Windows is, then gets shot? Asking an MS exec to give statistics on how much better (in any way, speed, security, etc.) the new version is compared to everything else is like asking a preschool student how much better his/her fingerpainting is compared any other painting: even if they just smeared everything together to create a big brown blob, their "8 times better" masterpiece should be hung in the Louvre. According to Microsoft, each new version of Windows is faster than the previous, yet Windows 3.1 ran fine on a 386/16 with 8MB RAM, but XP crawls on a machine 50 times faster. XP may do more, be more intuitive, and be a major improvement, but it certainly isn't faster. Odds are pretty good that if you need a few hundred people to maintain a piece of software on a full-time basis, your project is bloated, inefficent, and likely to be full of security holes.
Of course, this isn't just a Microsoft problem, plenty of open-source projects also fit this description, especially Gnome or KDE. Fvwm may not have been the ideal window manager, but it ran well on the same 386/16, and Gnome/KDE struggle on the same hardware XP struggles on.
I'd say you're mostly right. I think Intel will be positioned as the desktop / lower-end / lower-priced processor, and AMD will dominate the market for servers. Sort of a role reversal from the case before. Unless Intel comes up with something like AMD's direct connect architecture and hypertransport, AMD will continue to dominate I/O performance, even if Intel does eclipse them in the standard user benchmarks. I don't personally care who I buy from at work, but at least as it is now, the Opterons are the only thing allowed in the rack for new systems, and Intel chips for laptops. Desktops don't matter as much, so it's mostly Semprons at the moment (best bang for little bucks). As soon as Intel really abandons the MHz/GHz = performance mantra and realizes that the northbridge is their biggest drawback, they'll turn it around. The mobile chips and the Core Duo are a good start, but we'll see if they stick with them.
I saw a quote somewhere from an exec at AMD, who was asked how the Pentium 4 could be improved on. He said: "Use the Pentium III". The sooner Intel realizes he was right (mobile/Core chips are more closely related to the PIII than P4), the better.
Slim might be correct, but it certainly CAN happen. What is important is that you realize that it is possible. Just because you have to lose 2 or more to lose data doesn't make it safe. Having 20 drives from the same lot number could mean that all 20 are affected by a manufacturing flaw that kills them at 20,000 hours. Also, there are power surges, fire, etc. Always back up (off-site), always keep spares both on and off-site, and buy drives from multiple lot numbers. Keep in mind that the MTBF for multi-drive systems is lower than a single drive (in fact, statistically it is the MTBF for a single drive divided by the number of drives). You should expect failures, and should expect more than you would a single drive. The idea is not to reduce the likelyhood of a failure (it does the opposite), but to reduce the damage caused by a failure. Keep in mind that failures can occur in the drive, controller, power supply, or any other component. Keep spares on hand, and use a high end (3ware 9500 series works excellent for SATA, Adaptec for SCSI, but not as good IMHO as 3ware for SATA) controller card.
As for the original topic, I would go with SATA II drives, on a 3ware controller, and make sure both the drives and the controller support NCQ. Faster drives are of course better, but more drives is also better (more spindles = more platters = more read heads = faster aggregate reads), and keep in mind that RAID level is extremely important, and the right choice has a lot more to do with your application and supporting hardware than anything else. Generally, though, RAID 5 is sufficient for storage, RAID 1 for smaller application servers that you can't afford to have outages on, and RAID 10 (1+0) on databases (or really anywhere you can afford it). RAID 10 on a high-end controller allows you to read at double (or better) the speed of the RAID 0 subarrays, since the controller can read from the stripe set that is closest to the data/least busy. Also, be sure to get a controller with plenty of memory onboard, and battery backed memory if power loss could be an issue. Personally, I'd go with the 3ware 9550SX line on an Opteron-based system (eliminate the Intel-based northbridge I/O / Mem bottleneck).
In a home environment, this is likely not a real problem. In a business environment, anything beyond about 15 active users is usually too much for a Linksys-type router, since the processor and memory capabilities of these are usually pretty low. I think Netgear has a few for small/medium businesses, but if all you want is a NAT box, Linux/*BSD work quite well on some pretty low-end hardware. 100 users on a DSL/Cable circuit could be handled by an old Pentium 133 picked up on eBay for $25. At work, we have a FreeBSD box (though on a much faster Opteron 244) doing NAT, firewalling, monitoring, load balancing, and intrusion detection for 2 Gigabit segments, 3 T1s, and a Frame Relay circuit. On average, this box is at 0.4% CPU utilization when you aren't actively monitoring something.
And given the future of the global economy, you might want to start calling the Yuan (China) real money. But certainly, the Euro is a much better global choice than the US Dollar, and if China ever stops artificially lowering their exchange rate, the Yuan will probably be the strongest currency. After all, if you have over 1/5th of the world population using it, it must be a pretty good candidate for "real money". The dollar may be relatively strong, but only around 1/20th of the population really use it. (Not counting international business transactions, since the currency used really makes no difference.)
Even better, get rid of the mouse. If you don't do detail work (graphics), an IBM trackpoint works quite well. I just picked up an old IBM Model M13 keyboard (black, buckling-spring, trackpoint), and it's great (though the people sitting around me may not enjoy the sound at 130wpm as much as I do). It also has a PS/2 mouse port for using a regular mouse with it, for when you do need to do detail work. The only problem is that you will find yourself reaching for the trackpoint on every keyboard you use.
While I don't agree with the idea of buying the cheapest crap you can find, spending twice as much just for a logo and testing is hard to justify. Rather than buying 1 IBM, buy 2 custom systems, but pick top quality components. At our department it's AMD processors (Opterons), Tyan motherboards, 3ware SATA RAID or Adaptec SCSI RAID controllers, Western Digital SATA or Seagate SCSI drives, Samsung, Kingston, or Micron memory, Zippy n+1 redundant power supplies (for 3U+), and Chenbro cases. Rounding out the cheap list is the Netgear Prosafe line for the LAN (the L3 managed switches work great, have more features than competitors at 3X the price), and a handful of Cisco 16xx, 25xx, and 26xx routers picked up on eBay (Do you really need the fastest router on the market to handle a few T1s?). With two of everything, it's still cheaper than buying IBM or HP, and I get the added bonus of being able to use the spare systems for development/load balancing/failover/testing/etc.
Sun is just trying to become a combined RedHat and IBM, selling service and support for free software (like RedHat), and hardware and support to run the free software (like IBM). Actually, it seems like a pretty good idea to me, since it certainly has worked for RedHat and IBM, and Sun still has a great reputation in the large business and telecom sectors. Besides, what better thing could Sun do? Companies aren't buying big software contracts anymore, and they have moved away from the $50,000+ server market down to clusters and blades. Sun needed to position themselves to sell that lower-end hardware, and the best way for them to do that is to give away the OS, management tools, and development tools, and get big applications moved to "their" solution. They'll even give away some hardware to people who bring applications (and those applications' users) to them, or make it affordable for people who might have a say in the future on what solution to buy. They've gone from being a provider of solutions that few can afford, to one that everyone can afford, with the same name and reputation as before. They will never get back to where they were in the dot-bomb days, but I think they just realized that and are actually doing something about it.
If you still don't understand the thinking behind this, look at where the other top dot-bomb high-end-only hardware company is. Remember SGI? Take a look at their financials sometime, (hint: they're on pinks now, they were delisted from NASD last month) you'll see just how well hanging on to selling $50,000 workstations was working out.
Not to mention you're still forgetting the best reason.... You don't have to pay the $1080 up front. People on more limited budgets (students) can afford $30/mo, but can't necessarily afford to drop $900 today. I did the same thing for a notebook in college.
Aside from the Eclipse example (which really didn't have anything to do with Java, other than that the Eclipse people wanted to make up for some of the weaknesses of Swing [most of which no longer exist]), there really isn't that much choice. J2EE = J2SE + a few items, JavaBeans is basically just a framework for certain tasks, and if you don't need any of the specialized features of the add-ons, there really isn't anything other than "pure" Java. The core language today isn't that different from what it was 10 years ago, and having a large number of add-on packages available isn't something that should be viewed as a bad thing. A Java app will still run on all Java platforms, provided that any additional packages are installed. Choice in things like SOAP implementations, XML parsers, and application servers is a good thing, and it doesn't mean that the language isn't "standardized". If anything, Java is more "standardized" than C/C++. The Java String class is always in java.lang.String, while a String in C++ is in a different place depending on the compiler used. C and C++ certainly do better in some areas than Java, and Java does better in some areas than C/C++, but that doesn't make either one bad, or either one good.
Also, poor business model != poor choice of language. There are plenty of viable Java development shops. Language is just a tool to get the job done, and your favorite language may not always be the best choice for what you want to do. For some things, LISP is a great language, but I wouldn't want to write a VoIP softphone in it.
Back on topic though, as a developer, I always love free development tools, but in this case, as a shareholder, I'm not quite sure how they're going to make money on this. Good will can be hard to deposit in your bank account. At least they have 8-core processors and Opteron servers to (hopefully) turn a profit.
Actually, for a launch to happen this week, I would expect that they must be VERY confident, as this weekend is already a worldwide media event (the final round of the F1 championship in Shanghai), and having your mission fail while media from every corner of the planet is in your country seems like a bad idea to me. In fact, I think it may just be more likely that the Chinese want to use this weekend to show off their own technical capabilities.
Along the pseudo-management line, what exactly do you mean by this? If you mean you manage less than five people, but have complete or near-complete control the direction of the company's IT business, then I'd say stay where you are until you run out of new ideas. Think about things like increasing productivity, automation, VoIP, and B2B e-commerce. If you have an interest in things like that, you can still grow in the position you are in, and will have an easier time quantifying your value to the company when it comes time for raises, spin-off companies/divisions, or moving on to the next job. I work in the healthcare segment as a CIO/CTO, and personally get much greater joy out of cutting costs by 20%, increasing business, and integrating diverse components of the company through technology than I would out of being a code monkey for Google. Getting your hands dirty integrating everything but the snack machine is supremely satisfying, especially when the end result is that both you and the company profit from it.
Keep in mind, though, that an engineer at a search company might be worse than where you are now. You may find yourself bored to tears, working on a project you think is useless, and lacking the power to do anything about it. In other words, power is sometimes more important than money, and in a lot of cases, both are more important than working for a specific company. Take a good look at what you will be doing on a daily basis if you switch. There are a few exciting areas in search, but a lot more that could be considered "programmer hell". As a side note, as a management type, I would be far more impressed with what assets you brought to your previous employer than who your previous employer was. I'd hire a programmer who made 10 people redundant at his previous job with Waste Management before I'd hire the one who wrote the code to display a different picture each day for Google/Yahoo/whatever. There are a lot of good people at search companies, but more than half of them are nothing more than programmer drones that code up the latest design from management.
Of course, that's just me. I can't stand working on the exact same piece of code for more than about 2 days. A low to mid-level programmer position at Google would be tedious to me in a week (except for the days for personal projects, of course, I think that part is absolutely brilliant of them). I like being able to work on different subjects in IT and choose the next subject on a continual basis, it keeps things from being monotonous. Monday I'll install new servers, Tuesday I'll do some web services, Wednesday is management, maybe accounts receivable for Thursday, and VoIP for Friday, whatever. I keep a list of things that have to be done, and areas I want to look into, and just mix it up a bit. Some things you will like, some you won't, but being in control of your career makes it a lot more enjoyable overall.
Not bad, but by booting from a remote image, you run the risk of the remote image server being unavailable, unless you have a redundant image server. Your best bet is to have the remote image server, but use/create a provisioning system that allows you to install the image you want to the machine OR run a network image. This eliminates the need for a redundant image server most of the time (machines can still run from local disk), but allows you to reprovision for temporary needs like a failed server or scheduled maintenance.
For (an overly simplified) instance, say you run two main sites, site x on 4 machines, and site y on 4 machines, with the image server set to do nothing (allowing the machines to continue to boot from disk). If site x is on/. today, you simply adjust your provisioning to have 2 of the machines that usually host site y to run site x from network image, and reboot the two machines. Now you have a 6/2 split, and when the load subsides, all you have to do is remove the rule and reboot the two machines. Of course, if you decide that site x needs 5 full time machines, you could just set the provisioning to load the host image for x onto one of the y machines, reboot it, and remove the rule. I don't know of any existing free/open source software to do this in a general case off hand, but I'm sure such an animal already exists, since there are dozens of commercial packages to do this. If not, it won't take more than an hour or two to whip up some shell scripts to do this in your specific case. All you need is a network boot system that writes an image file or copies a directory named in some configuration file to disk and reboots for the permanent case, and runnable network boot images for the temporary case (probably just a root NFS that uses the same directory as the permanent one).
Slashdot Mirror
True, and throwing out equipment fried by lightning doesn't hurt, either. Seriously, what kind of questions are these? Hardware doesn't work? Isolate it and replace it. Other than freezing an old hard drive to free up bearings to get data off before you throw it out, it's not worth the aggravation.
Of course, as far as real Computer/Equipment Voodoo is concerned, there is always the Heisenbug. Just had a mechanical version of this today, the Bayer tech has spent 3 days on a machine to isolate a pump problem. To see the pumps, you have to open a panel either on the side or the front of the instrument. The past 2 days, he was working on it through the front, and the problem didn't occur. Today, after being called back because it happened again, he opened the side panel to watch it, and accidentally bumped the front panel while he was looking at it. As soon as the front panel closed, the problem occurred. It turns out that a zip-tie that holds some tubing from the pumps together was caught on the front panel, and when the panel door closed, it pulled on the zip-tie, which pulled on and pinched the tubing, causing a pressure sensor to throw a fault.
For Windows, use roaming profiles and default installations. For Linux, rsync works quite well for the base OS (say, a staggered start time at night based on IP), with OpenLDAP doing auth and home dirs stored on central server(s) and mounted via NFS. New system setup becomes a knoppix CD, partition the new drive, format partitions, mount them, rsync the distribution to the new machine, chroot, and setup boot loader. You could of course script all of this, and this is very similar to what I do for kiosk systems (Linux/Firefox setup), except the kiosks don't change, so it's just a big tarball via sftp instead of rsync. You could also do tarballs, and keep the last few versions as backups in case you screw something up. If the hardware is identical, use the distribution of your choice, but if there are several different systems, you may want to use one with good hardware detection (like knoppix).
VisiCalc was not included because it was buggy and lacked important features, but I have to agree with you that Lotus 1-2-3 belongs there, since Excel is buggier than VisiCalc was. Java belongs on the list for exactly the reasons specified in the article, it completes the source code portability concept started with other languages by adding cross-platform BINARY portability through the use of an intermediate language and VM. While Java may have its roots in C/C++, Smalltalk, and large included-library languages like Ada, Java has become the basis of and benchmark for all new general-purpose languages, including everything that has come from Redmond since the introduction of Java. Things like RIM's Blackberry and Microsoft's .NET are the best examples of why Java is great software, since the language is essentially irrelevant, but the VM concept and large, standardized, and stable core library are the reasons why Java has the influence it has. Simple ideas that are taken for granted now were revolutionary (although often originally introduced 30-40 years before) concepts when Java was introduced, like Swing, RMI, and the security manager (including the sandbox for browser applications).
...and you expect most Slashdot readers to actually READ the LAST paragraph? Of the actual article? Are you kidding?
To avoid this problem, you evaluate the tools you have (floats, double, fixed decimal, int, long, possibly BCD, etc) and see if they match your needs. If not, go get better tools (an arbitrary-precision number library, like java.math.BigInteger or java.math.BigDecimal). Similar libraries exist for just about any language. Decimal arithmetic is inefficient on binary systems (except for hardware-implemented BCD as in IBM S/370), as well as completely unnecessary, since the binary arithmetic functions already exist in hardware, and storing individual decimal digits will always take more space than storing the whole number in binary.
The RPG language also had an implementation of BCD, and probably any compiler for the IBM S/370 line would at least have a library for it, as I believe the IBM mini/mainframe architectures had implemented it in hardware.
no, although a double or larger increases the precision, all floating point-style numbers suffer from this problem. You might get away with it for a larger number, but you are wasting bits, and stil cannot guarantee that the number is always going to be accurate. In addition, as the number gets larger, more precision is lost. For instance, 782533.37 as a float is 493F0C55, or 782533.31, and as a rounded float is 493F0C56, or 782533.38. Neither would be acceptable, and even a double will lose precision at about 16 significant digits, which could be a problem for daily interest calculations or another high-precision calculation. What should be used is a packed decimal (BCD) or fixed decimal (int or long and an associated scale factor). The whole point is that for a financial institution, there is likely a fixed precision level that is acceptable, but floats, doubles, etc., cannot guarantee that any given precision will be maintained, as it depends upon all of the factors involved (value and size of each number in the calculation). For the same reasons, floating point numbers should never be used in program flow control, either, as doing a for(float x = 0.0; x 1.0; x = x + 0.01) will iterate 101 times, since the first time x is not less than 1.0 is when it reaches 1.00999..., and if you are looking for a specific value, it may never reach it (like for(x = 0.0; x != 1.0; x = x + 0.01)). If you instead assume that you will always deal with 2 decimal places (or 3, or whatever), you can guarantee that your addition or multiplication will be accurate, and can scale the answer later if necessary.
For the uneducated, the reason that this is stupid is that IEEE-754 floating point numbers cannot REPRESENT all values, they APPROXIMATE them. There is no way to properly represent the value 0.01 as a float (0.01 is best approximated by 3C23D70A, or 9.9999998e-3). So, for instance, if you were to add up 100 pennies, you would have 99.999998 cents, not 100. Repetitive additions (like credits and debits from an account) or multiplications (interest calculations, amortizations, etc.) simply make the problem worse, which is why floats should NEVER be used to track money. A fixed decimal system should always be used for financial systems.
You mean like these: SCAT Sacramento Center for Assistive Technology SCAT Sarasota County Area Transit (Florida) SCAT School & College Ability Test SCAT School and Community Action Team SCAT Scout/Attack SCAT Screen Capture and Transfer SCAT Secure Campus Anonymous Tip SCAT Security Control of Air Traffic SCAT Senior Citizens Against Crime SCAT Senior Citizens Against Telemarketing SCAT Sensor Coverage Analysis Tool SCAT Sewing Circle and Terrorist Society (women's group) SCAT Sheep Cell Agglutination Test SCAT Ship Certification Acceptance Team SCAT Shoreline Cleanup Assessment Team (oil spills) SCAT Shrewsbury College of Arts and Technology SCAT Single Channel Autonomous Trunking SCAT Somerset College of Arts and Technology SCAT Sonar Calibration Alignment Training SCAT South Carolina Association of Taxpayers SCAT South Coast Agility Team (dog agility club; Costa Mesa, CA) SCAT South Coast Aquatic Team SCAT South Coast Area Transit (California) SCAT Space Coast Area Transit (Brevard County, Florida) SCAT Space Coast Area Transit (Florida) SCAT Special Community Action Team SCAT Specialized Community Area Transportation (service) SCAT Sport Competition Anxiety Test SCAT Street Cleansing Ambulance Truck SCAT Street Crime Assertive Team (East Orange, New Jersey) SCAT Street Crimes Arrest Team(s) (police unit) SCAT Student Coalition Against Tobacco SCAT Student Community Action Team SCAT Sub-Category (US Navy system for categorizing test instruments) SCAT Subcategory SCAT Submarine Classification & Tracking SCAT Submersible Cable-Actuated Teleoperator SCAT Submersible Concentric, Axially Traversing SCAT Supersonic Commercial Air Transport SCAT Support Center for Advanced Telecommunications Technology Research, Foundation SCAT Surface Controlled Avalanche Transistor SCAT Suspense Control and Automated Tracking System SCAT Switch Consolidation Action Team (Sprint) SCAT Synchronized COA Authoring Tool SCAT System Calibration and Test SCAT System Commonality Analysis Tool SCAT Systematic Causal Analysis Technique (systematic method of analyzing the basic causes of accidents) SCAT Systems Consolidation of Accessions & Trainees Oddly enough, it doesn't seem like anyone outside of Silicon Valley knows what a dictionary is.
Yeah, I was thinking his problem sounded like a good choice might be an ATA over Ethernet-type solution, like Coraid.
...and don't forget Sun, if you feel like paying $20,000 for an 8-core Ultrasparc T1 chip that uses less power than an Opteron. Windows not allowed though. But since the original question is probably talking about managing a datacenter full of customer machines (which you can't control, unless you lease them), I would imagine that just changing the way you do cooling would make a rather large difference. Have you considered piping heat from the racks to the outside through the roof, and using a high specific-heat fluid (say water/ethylene glycol mix) that could be prechilled to an extremely low temperature during off-peak times, then used in a radiator-type arrangement for cooling. I would say that sealing the sides of your rack and replacing the front and rear doors with liquid-cooled radiators/heat collectors should work quite well. If you can pre-chill the air entering the front of the machines, it will be cooler coming out, and if you absorb the excess heat coming out the back, less heat will enter the room, meaning your air conditioners won't have to work as hard. Other good option would be to duct the exhaust heat (if it is significantly above room temp) up and out of the building with a few smaller fans.
Anyone else read that and think of the scene in the South Park movie where Gates comes in and says how much faster the new Windows is, then gets shot? Asking an MS exec to give statistics on how much better (in any way, speed, security, etc.) the new version is compared to everything else is like asking a preschool student how much better his/her fingerpainting is compared any other painting: even if they just smeared everything together to create a big brown blob, their "8 times better" masterpiece should be hung in the Louvre. According to Microsoft, each new version of Windows is faster than the previous, yet Windows 3.1 ran fine on a 386/16 with 8MB RAM, but XP crawls on a machine 50 times faster. XP may do more, be more intuitive, and be a major improvement, but it certainly isn't faster. Odds are pretty good that if you need a few hundred people to maintain a piece of software on a full-time basis, your project is bloated, inefficent, and likely to be full of security holes.
Of course, this isn't just a Microsoft problem, plenty of open-source projects also fit this description, especially Gnome or KDE. Fvwm may not have been the ideal window manager, but it ran well on the same 386/16, and Gnome/KDE struggle on the same hardware XP struggles on.
I'd say you're mostly right. I think Intel will be positioned as the desktop / lower-end / lower-priced processor, and AMD will dominate the market for servers. Sort of a role reversal from the case before. Unless Intel comes up with something like AMD's direct connect architecture and hypertransport, AMD will continue to dominate I/O performance, even if Intel does eclipse them in the standard user benchmarks. I don't personally care who I buy from at work, but at least as it is now, the Opterons are the only thing allowed in the rack for new systems, and Intel chips for laptops. Desktops don't matter as much, so it's mostly Semprons at the moment (best bang for little bucks). As soon as Intel really abandons the MHz/GHz = performance mantra and realizes that the northbridge is their biggest drawback, they'll turn it around. The mobile chips and the Core Duo are a good start, but we'll see if they stick with them.
I saw a quote somewhere from an exec at AMD, who was asked how the Pentium 4 could be improved on. He said: "Use the Pentium III". The sooner Intel realizes he was right (mobile/Core chips are more closely related to the PIII than P4), the better.
Slim might be correct, but it certainly CAN happen. What is important is that you realize that it is possible. Just because you have to lose 2 or more to lose data doesn't make it safe. Having 20 drives from the same lot number could mean that all 20 are affected by a manufacturing flaw that kills them at 20,000 hours. Also, there are power surges, fire, etc. Always back up (off-site), always keep spares both on and off-site, and buy drives from multiple lot numbers. Keep in mind that the MTBF for multi-drive systems is lower than a single drive (in fact, statistically it is the MTBF for a single drive divided by the number of drives). You should expect failures, and should expect more than you would a single drive. The idea is not to reduce the likelyhood of a failure (it does the opposite), but to reduce the damage caused by a failure. Keep in mind that failures can occur in the drive, controller, power supply, or any other component. Keep spares on hand, and use a high end (3ware 9500 series works excellent for SATA, Adaptec for SCSI, but not as good IMHO as 3ware for SATA) controller card.
As for the original topic, I would go with SATA II drives, on a 3ware controller, and make sure both the drives and the controller support NCQ. Faster drives are of course better, but more drives is also better (more spindles = more platters = more read heads = faster aggregate reads), and keep in mind that RAID level is extremely important, and the right choice has a lot more to do with your application and supporting hardware than anything else. Generally, though, RAID 5 is sufficient for storage, RAID 1 for smaller application servers that you can't afford to have outages on, and RAID 10 (1+0) on databases (or really anywhere you can afford it). RAID 10 on a high-end controller allows you to read at double (or better) the speed of the RAID 0 subarrays, since the controller can read from the stripe set that is closest to the data/least busy. Also, be sure to get a controller with plenty of memory onboard, and battery backed memory if power loss could be an issue. Personally, I'd go with the 3ware 9550SX line on an Opteron-based system (eliminate the Intel-based northbridge I/O / Mem bottleneck).
In a home environment, this is likely not a real problem. In a business environment, anything beyond about 15 active users is usually too much for a Linksys-type router, since the processor and memory capabilities of these are usually pretty low. I think Netgear has a few for small/medium businesses, but if all you want is a NAT box, Linux/*BSD work quite well on some pretty low-end hardware. 100 users on a DSL/Cable circuit could be handled by an old Pentium 133 picked up on eBay for $25. At work, we have a FreeBSD box (though on a much faster Opteron 244) doing NAT, firewalling, monitoring, load balancing, and intrusion detection for 2 Gigabit segments, 3 T1s, and a Frame Relay circuit. On average, this box is at 0.4% CPU utilization when you aren't actively monitoring something.
And given the future of the global economy, you might want to start calling the Yuan (China) real money. But certainly, the Euro is a much better global choice than the US Dollar, and if China ever stops artificially lowering their exchange rate, the Yuan will probably be the strongest currency. After all, if you have over 1/5th of the world population using it, it must be a pretty good candidate for "real money". The dollar may be relatively strong, but only around 1/20th of the population really use it. (Not counting international business transactions, since the currency used really makes no difference.)
...and for healthcare organizations, HIPAA. Or, for publicly-traded healthcare organizations, both, which sucks.
Even better, get rid of the mouse. If you don't do detail work (graphics), an IBM trackpoint works quite well. I just picked up an old IBM Model M13 keyboard (black, buckling-spring, trackpoint), and it's great (though the people sitting around me may not enjoy the sound at 130wpm as much as I do). It also has a PS/2 mouse port for using a regular mouse with it, for when you do need to do detail work. The only problem is that you will find yourself reaching for the trackpoint on every keyboard you use.
While I don't agree with the idea of buying the cheapest crap you can find, spending twice as much just for a logo and testing is hard to justify. Rather than buying 1 IBM, buy 2 custom systems, but pick top quality components. At our department it's AMD processors (Opterons), Tyan motherboards, 3ware SATA RAID or Adaptec SCSI RAID controllers, Western Digital SATA or Seagate SCSI drives, Samsung, Kingston, or Micron memory, Zippy n+1 redundant power supplies (for 3U+), and Chenbro cases. Rounding out the cheap list is the Netgear Prosafe line for the LAN (the L3 managed switches work great, have more features than competitors at 3X the price), and a handful of Cisco 16xx, 25xx, and 26xx routers picked up on eBay (Do you really need the fastest router on the market to handle a few T1s?). With two of everything, it's still cheaper than buying IBM or HP, and I get the added bonus of being able to use the spare systems for development/load balancing/failover/testing/etc.
Sun is just trying to become a combined RedHat and IBM, selling service and support for free software (like RedHat), and hardware and support to run the free software (like IBM). Actually, it seems like a pretty good idea to me, since it certainly has worked for RedHat and IBM, and Sun still has a great reputation in the large business and telecom sectors. Besides, what better thing could Sun do? Companies aren't buying big software contracts anymore, and they have moved away from the $50,000+ server market down to clusters and blades. Sun needed to position themselves to sell that lower-end hardware, and the best way for them to do that is to give away the OS, management tools, and development tools, and get big applications moved to "their" solution. They'll even give away some hardware to people who bring applications (and those applications' users) to them, or make it affordable for people who might have a say in the future on what solution to buy. They've gone from being a provider of solutions that few can afford, to one that everyone can afford, with the same name and reputation as before. They will never get back to where they were in the dot-bomb days, but I think they just realized that and are actually doing something about it.
If you still don't understand the thinking behind this, look at where the other top dot-bomb high-end-only hardware company is. Remember SGI? Take a look at their financials sometime, (hint: they're on pinks now, they were delisted from NASD last month) you'll see just how well hanging on to selling $50,000 workstations was working out.
Not to mention you're still forgetting the best reason.... You don't have to pay the $1080 up front. People on more limited budgets (students) can afford $30/mo, but can't necessarily afford to drop $900 today. I did the same thing for a notebook in college.
Aside from the Eclipse example (which really didn't have anything to do with Java, other than that the Eclipse people wanted to make up for some of the weaknesses of Swing [most of which no longer exist]), there really isn't that much choice. J2EE = J2SE + a few items, JavaBeans is basically just a framework for certain tasks, and if you don't need any of the specialized features of the add-ons, there really isn't anything other than "pure" Java. The core language today isn't that different from what it was 10 years ago, and having a large number of add-on packages available isn't something that should be viewed as a bad thing. A Java app will still run on all Java platforms, provided that any additional packages are installed. Choice in things like SOAP implementations, XML parsers, and application servers is a good thing, and it doesn't mean that the language isn't "standardized". If anything, Java is more "standardized" than C/C++. The Java String class is always in java.lang.String, while a String in C++ is in a different place depending on the compiler used. C and C++ certainly do better in some areas than Java, and Java does better in some areas than C/C++, but that doesn't make either one bad, or either one good.
Also, poor business model != poor choice of language. There are plenty of viable Java development shops. Language is just a tool to get the job done, and your favorite language may not always be the best choice for what you want to do. For some things, LISP is a great language, but I wouldn't want to write a VoIP softphone in it.
Back on topic though, as a developer, I always love free development tools, but in this case, as a shareholder, I'm not quite sure how they're going to make money on this. Good will can be hard to deposit in your bank account. At least they have 8-core processors and Opteron servers to (hopefully) turn a profit.
Actually, for a launch to happen this week, I would expect that they must be VERY confident, as this weekend is already a worldwide media event (the final round of the F1 championship in Shanghai), and having your mission fail while media from every corner of the planet is in your country seems like a bad idea to me. In fact, I think it may just be more likely that the Chinese want to use this weekend to show off their own technical capabilities.
Along the pseudo-management line, what exactly do you mean by this? If you mean you manage less than five people, but have complete or near-complete control the direction of the company's IT business, then I'd say stay where you are until you run out of new ideas. Think about things like increasing productivity, automation, VoIP, and B2B e-commerce. If you have an interest in things like that, you can still grow in the position you are in, and will have an easier time quantifying your value to the company when it comes time for raises, spin-off companies/divisions, or moving on to the next job. I work in the healthcare segment as a CIO/CTO, and personally get much greater joy out of cutting costs by 20%, increasing business, and integrating diverse components of the company through technology than I would out of being a code monkey for Google. Getting your hands dirty integrating everything but the snack machine is supremely satisfying, especially when the end result is that both you and the company profit from it.
Keep in mind, though, that an engineer at a search company might be worse than where you are now. You may find yourself bored to tears, working on a project you think is useless, and lacking the power to do anything about it. In other words, power is sometimes more important than money, and in a lot of cases, both are more important than working for a specific company. Take a good look at what you will be doing on a daily basis if you switch. There are a few exciting areas in search, but a lot more that could be considered "programmer hell". As a side note, as a management type, I would be far more impressed with what assets you brought to your previous employer than who your previous employer was. I'd hire a programmer who made 10 people redundant at his previous job with Waste Management before I'd hire the one who wrote the code to display a different picture each day for Google/Yahoo/whatever. There are a lot of good people at search companies, but more than half of them are nothing more than programmer drones that code up the latest design from management.
Of course, that's just me. I can't stand working on the exact same piece of code for more than about 2 days. A low to mid-level programmer position at Google would be tedious to me in a week (except for the days for personal projects, of course, I think that part is absolutely brilliant of them). I like being able to work on different subjects in IT and choose the next subject on a continual basis, it keeps things from being monotonous. Monday I'll install new servers, Tuesday I'll do some web services, Wednesday is management, maybe accounts receivable for Thursday, and VoIP for Friday, whatever. I keep a list of things that have to be done, and areas I want to look into, and just mix it up a bit. Some things you will like, some you won't, but being in control of your career makes it a lot more enjoyable overall.
Not bad, but by booting from a remote image, you run the risk of the remote image server being unavailable, unless you have a redundant image server. Your best bet is to have the remote image server, but use/create a provisioning system that allows you to install the image you want to the machine OR run a network image. This eliminates the need for a redundant image server most of the time (machines can still run from local disk), but allows you to reprovision for temporary needs like a failed server or scheduled maintenance.
/. today, you simply adjust your provisioning to have 2 of the machines that usually host site y to run site x from network image, and reboot the two machines. Now you have a 6/2 split, and when the load subsides, all you have to do is remove the rule and reboot the two machines. Of course, if you decide that site x needs 5 full time machines, you could just set the provisioning to load the host image for x onto one of the y machines, reboot it, and remove the rule. I don't know of any existing free/open source software to do this in a general case off hand, but I'm sure such an animal already exists, since there are dozens of commercial packages to do this. If not, it won't take more than an hour or two to whip up some shell scripts to do this in your specific case. All you need is a network boot system that writes an image file or copies a directory named in some configuration file to disk and reboots for the permanent case, and runnable network boot images for the temporary case (probably just a root NFS that uses the same directory as the permanent one).
For (an overly simplified) instance, say you run two main sites, site x on 4 machines, and site y on 4 machines, with the image server set to do nothing (allowing the machines to continue to boot from disk). If site x is on