I can't answer the specifics of the question (whether or not the quality is in fact different), but I have found that DVD players, much like CD players, are all really the same. You pay for features, and maybe a little better error correction, but otherwise they're all pretty much the same quality.
I have an Apex DVD player (the one that lets you run without copy protection or encryption), which cost me $180 (at the time, most players were selling for around $250, so it was unbelievably cheap). It's very good quality, and works very well (though I hear the special edition Abyss is one of the few DVDs it has trouble with).
If you're interested in saving a few bucks, I'm pretty sure the DVD player in the XBox or PS2 is as good as any basic player out there (though I'm not sure if a remote is included with the unit). The Gamecube is $100 cheaper, but if you pick up a $100 DVD player, you're about even. If you want something special, like progressive scanning, then you'll need to get a seperate player.
1. There is the Cancer aspect. It's obvious that no one understands the issues that may arise from having a radio transmitter around you all the time (Cel phones) let alone multiple ones around you all the time at your computer.
Although it (appears) true that RF energy can cause problems with cells in the body, the amount of power in an 802.11b or Bluetooth antenna is FAR less than cellular equipment. A microware cooks food at 1000 watts @ 2.5 GHz, but long term cell phone use at 1 watt is debatable (probably not good for you, anyway), and long term use of 802.11 equipment at 0.05 to 0.1 watts is most likely insignificant. Plus, RF signals degrade at 1/(n^4), so even 1 foot away from an antenna is significantly less power (though I won't debate how much radiation my left leg receives from my Lucent PCMCIA wireless card in my laptop - still, the convenience to me is with the miniscule risk).
There are three sets of possible issues you might encounter: Hardware, Software, and Compatibility. It looks like you've considered some of the software issues, though I'm not sure you've considered the harware or compatibility issues yet.
Hardware issues probably center around reliability, more than anything. If you're purchasing a 3 year old disk array which received constant use, then you can expect that you WILL encounter hardware issues - if not now, sometime during its useful life. You might just encounter a bad disk, or there may be larger problems (like with the head unit). I've seen NetApps which are only a few WEEKS old lose disks (not much EMC experience here).
You may need/want to update firmware or patch the internal OS of the unit. Be warned that you may need to purchase some sort of support contract to get patches and updates. Also, hardware like this is notoriously touchy about being moved - so once you get it set up, don't tear it down and set it up somewhere else, unless you're prepared to fight with it.
Finally, compatibility issues may arise. I've noticed that NetApps in particular are very touchy when using Linux NFS clients - very often the clients will need to be rebooted when they hang. I think this is more of a Linux NFS client code issue than a NetApp issue (Linux NFS is *slowly* getting better), but its something to keep in mind. Without knowing what hardware you are going to purchase, it's very difficult to tell you what to look out for, but my primary experience has been with NetApp (fairly reliable, and a huge pain to work with in a cross platform environment) and a little EMC (nifty, a little faster, and a lot more expensive).
Does everything in the world need to be "internet accessible" or "web-enabled"?!?!?
How long before phones start getting hacked or spread MS LookOut worms? How long before phone spamming becomes the norm?
*Sigh* I want the web for convenience. I want to web to make my life easier. I don't need the "cool" factor of every internet-capable device. I don't need my refrigerator ordering food for me, and I certainly don't need IRC on my phone. Frankly, I find it easier to CALL someone rather than attempt to type on a frickin' phone (or follow an IRC session with 50 people on one of those little phone LCDs). But I digress...
This sounds fairly trivial, actually. From his specs, assuming a standard set of hardware, it wouldn't be too hard to build a stock kernel, X with standard SVGA, use the splash screen patches to hide the messages, and install rdesktop (which, for those of you who are unfamiliar, is similar to the Citrix client).
In fact, he could even do a thin-client boot from the network and go diskless (which would make maintenance even easier), though for his specs, mounting most the partitions read-only and running reiser or ext3 on anything which changes would probably suffice).
I'm fairly sure I could do this in very little time, but I'm sure someone with more free time (since this project isn't paying enough for those of us who need to pay our rent) will beat me to it:-)
Altogether, probably $400-500 with OEM parts. IMHO, this would kick the ass of the X-Box, but you'd have to spend more money for it.
I'm not so sure this is true. Despite the fact that the majority of the hardware is based on x86 architecture, there are a number of differences, including the shared system/video memory architecture. A "comparable" Intel system won't perfom nearly as well - x86 hardware is designed to perform every task in a mediocre fashion instead of performing a few, specialized tasks very well. The X-box probably would do a TERRIBLE job of running MS Office, but it does a much better job of playing games.
In addition, I think there are several benefits of a console system (though I desperately despise both Microsoft and the X-box):
Standard hardware architecture for companies to design software - none of this "your Riva128 is too slow, and your GeForce2 is incompatible without a drive update, but the original GeForce or Matrox G400 will work..." Game companies like developing for consoles because there is a standard system, and they KNOW how their product will perform.
Decent game controllers - something normally not available for the PC.
The ability for multiple people to easily play together on a decent size "monitor" (TV).
I fully agree that a real computer seems to be a better investment than the X-box, but if all you're doing is playing games, a console like this really is better suited to the task.
Enough already! I don't need yet another web-enabled device. Okay, firewire from a DV cam is useful if I want to edit video, but I don't need to cruise the web on my handycam, and the last thing I want to hear is how the next Outlook worm has managed to hack my camera!
Sendmail is neither buggy nor slow nor full of security holes. I've worked with it on systems handling up to ~2 million messages per week on just a few ultra 2's, and it rarely broke a sweat. And, there hasn't been a sendmail security issue in something like 4 years. Yes, the configuration syntax is difficult if you're not used to it (although it's MUCH easier if you DON'T EDIT THE sendmail.cf FILE and instead EDIT THE sendmail.mc FILE AND REGENERATE THE sendmail.cf FILE FROM IT), but once you learn it you understand *why* the majority of MTAs are sendmail (or some variant). I've tried qmail and postfix, and I still use sendmail, so go figure.
Bind has had its share of security holes (and will probably have more), but is also neither slow nor buggy. I know PLENTY of ISPs that run BIND on literally hundreds of thousands or even millions of entries without any trouble. I've read dan's docs, point of view, preaching, etc, and I just don't buy it.
The beauty of Unix is that there are always multiple ways to get the job done. I prefer to use the time tested, proven method. I know when I have trouble, I can find an answer to my Sendmail, Apache, and BIND questions *without* having to e-mail the development lists for help - and that's helpful when you're doing an install or upgrade on a weekend when those people aren't around. IMO, I shouldn't have to go to the product-specific mailing list for answers, and I like knowing that my company will have an easy time finding a successor who understand the products when I move on.
Get Dan to change his ridiculous license and get a reasonable percentage of the syadmin population using his products, and then I'll consider trying them out again.
Get the bastille firewall toolkit. You can configure your rules by editing one very well documented file, and it can generate the rulesets for IPChains or IPTables (depending on your kernel version). It's available at:
http://www.tux.org/~peterw/#fwall
I've used it for IPChains and then migrated right to IPTables - it does a great job, IMO (well, none of the systems I've configured it on have been hacked, so I suppose that's a good sign). You may still want to review the rules once it's complete, but it's an excellent starting point
Because it's easy to find support - there are MANY companies who can support BIND, and even more DNS admins who can help.
Because troubleshooting is easier when you can search the web for answers (unlike many alternatives) - so many people have used BIND that even the toughest questions can be found using google.
Because it has a good license - you can modify it ad redistribute modified versions WITHOUT having to send your patches to the author.
Because the documentation is better - I tried to install djbdns once, and the included documentation said "go to the web site for documentation", and yet the website documentation couldn't answer my questions.
I'm not saying BIND is perfect, but it's robust, scales well, is easy to support, and has a proven track record. I've never had BIND up and die on me or choke even under the heaviest of loads. In addition, BIND 9 was a "complete" rewrite - I don't believe it's shown any security holes, and it offers all the features required in a DNS server.
I'll take Sendmail, Apache, and Bind over the alternatives any day.
Does anyone know if Apple intends to offer firmware updates to current Airport owners (or if this is even possible), or if those who previous shelled out $299 for an Airport will be out in the cold?
Rather than multiple competing standards for the *next generation* of devices, why won't someone devise a bus technology which won't be obsolete in two or three years? While PCI-X doubling the bandwidth to 1GB sounds great, with upcoming 10 Gb network adapters, you've ALREADY reached its limit WITHOUT ANY OTHER PERIPHERALS! Imagine what will happen when 100 Gb is standard!
In the mean time, how about using some proven technology now? Five years ago, SGI was using multiple crossbars on their workstations to improve throughput, reduce contention, and decrease latency.
Why can't something like this be implemented with the current PCI architecture?
Okay, PCI-X sounds neat, and backwards compatibility is a plus, but if you read the article it states that PCI-X will run at the lowest common denominator - if you have three 133MHz/64-bit devices (up to 1.06 G/sec transfer) and one 33MHz/32-bit PCI device you're stuck at 33MHz. Four PCI cards, four crossbars, and you can get any speed you damn well like on any slot. Otherwise, the current crop of PCI modems and low-speed network adapters will end up limiting this anyway!
The remaining protocols sounds nifty, but I don't see any performance specs. As hopeful as I am that *something* good will replace PCI, I'm going to be very disappointed if there are too many standards and vendors can't agree on something.
Someone asked me this same question before, and he's what I told them:
> I would like to start working toward relevant certifications, but I am not
> sure where to start. I have been researching the RHCE certification,
> which would seem to me to be a good place to start, but I am not really
> sure. What do you think? What would your suggestions be? MCSE, Cisco,
> RedHat, or something else?
Really depends on what you want to do. I'm a Unix guy. I love Unix. If
I had the time (and someone to pay for it), I'd get certified in RedHat,
Solaris, AIX, HPUX, and anything else I could find.
But, there are more MSCE jobs out there. That's not to say that having an
MCSE is better. If you don't mind working with Windows, and can (at least
superficially) buy in to the Microsoft (learning) mentality, then it's not
such a bad deal. I personally lost faith in the MCSE exams when I took
the 2000 exam without ever touching 2000 or opening a book and passed it
with around an 80%.
There are a lot of employers that won't consider you for an NT/2000 job
without one. Do they realize that the exam is really a joke for
experienced IT guys? Probably not. Especially for recruiters, it's easy
to say "well, person A must be qualified because they're certified, even
though person B has more experience".
Then again, as the economy slows, there are a lot of people who were never
really qualified to be IT guys who are MCSE and are out of work. There
are a lot fewer qualified unix people out there, so even though there are
less jobs, it's easier to find work. Don't know if that makes sense, but
think like this:
If there are MCSE 10,000 jobs, and 15,000 MCSEs, only the top 66%
can find work. However, if there are 1000 Unix jobs, and 1500 unix
people, well, it's easier to get in that top 66%, because you only have to
elbow your way above the other 500, not 5000. Sure there's still a
percentage who are incompetent, but if you get enough experience with
different Unix flavors, you will always be needed.
Just to give you an idea, I've worked with Linux for 3.5 years now,
FreeBSD for a year, dabbled in OpenBSD, Irix for 4 years as an end-user
and 6 months as an admin, Solaris for about a year (SPARC and x86), HP-UX
and AIX for 6 months, and some VAX/VMS in college. Sounds much more
impressive than "3-4 years of Windows 9x/NT/2000". And the truth is, once
you've learned the concepts of one Unix, the others are pretty easy to
guess.
So, back to your original question: What do you want to do? Do you want
to get certified to work with Unix, for added job security, extra pay,
etc? Are you comfortable with Windows? Do you like working with it? You
don't really have to hate Microsoft just because 5% or 10% of us do. One
of the things I love about Unix is that Unix guys typically make $15k more
than MCSEs because there's more of a demand.
If I was going to be a Windows guy forever, I'd get an MCSE, possibly +I,
and probably a CCNE (I think the path is CCNA->CCNP->CCNE now). At the
very least, take the first couple MCSE tests and be a basic MCP, and get
your CCNA.
If you like Linux, and think that's the path you want to follow, then I
have some better advice for you: Go buy a Sparcstation 10 or 20 off ebay
and some other hardware. Be prepared to spend $1000, but get yourself a
P200 you can play with, a Sparcstation 10, and maybe an SGI Indy and an HP
A-class (or some other basic HP box). Do a little research first, to find
out what OSes run on what hardware (google is wonderful for this kind of
thing), and check out the spec.org pages for speeds (so you don't buy
something too old). Make sure you get copies of media (OSes), or buy
them, install the OSes on each box, and start working with them.
Dedicate an hour a day to doing things like trying to build SSH, learn how
thei admin tools work on that OS, and everything else. Configure the
network. Convert your P200 to a firewall, configure firewalling, and use
it to cruise the web. My firewall at home is a Pentium 133 with 32 megs
of RAM,and it uses a modem for the external internet connection.
Your goal should be to be able to build a little network at home, make
sure all the machines can communicate, and can interoperate. Try to set
up your own DNS. Configure Apache. Play with Sendmail. Try to get a
feel for basic admin things. Once you get really comfortable, teach
yourself a little Perl or shell scripting. A little Perl goes a long way.
You don't have to be a guru, but knowing how to read a file, parse it for
key words, and reformat it can go a long way.
Figure it will take a year to really learn the systems, but you'll learn
very quickly being forced to work with them yourself. Six months of
administering your own systems, and trying to run all your own services is
like 2 years of work experience. Don't be afraid to ask for help, but
always try to find it yourself - Unix gurus usually don't mind being asked
once where something obvious can be found, but when they have to
repeatedly point someone to websites which could be found on google, man
pages, and howto's which are included in the distribution, they tend to
get bitter:-)
After you learn everything, you can go take the exams for a few hundred
dollars, and you'll be able to pass them with your eyes closed. The key
is not spewing out what you read in the book - it's understanding the
systems. Once you understand the underlying questions, you may very well
know the answer without ever reading the books or touching the system.
Did I answer your question, or confuse you more?
Is this actually useful?
on
RLX Gets Denser
·
· Score: 2, Insightful
Is shrinking equipment like this actually useful to most companies? I'm curious, because my prior employer, had lots of rack space, but the colo site only had ~20amps per rack available. With 9' racks, it was very easy to run out of power BEFORE space was an issue. They (above.net) claimed their colo was designed for 6-8U "servers", not 1U "pizza boxes".
Aside from power, how about heat dissapation? After all, a bunch of laptops theoretically would make EXCELLENT servers, since they even have 1-2 hour battery backups built right in, are extra small, and don't even require a KVM (they could just be pulled out as needed). BUT, a laptop runs waaaaaaay too hot to be used as a server, and their price/performance ratio is aweful, even if you do save on rack space.
Finally, do these have any redundancy built in? Is there anything else special about them? Personally, I'd rather have a 1U rack unit which is dual CPU, dual NIC, and dual power supply than two of these. It would probably be cheaper, last longer, and be less hassle.
Perhaps a better (and more professional) solution is something like Citrix. VNC limits you to one connection at a time, while Citrix allows multiple sessions simultaneously. In addition, Citrix is probably the fastest VNC-style thing I've used, and the Metaframe protocol works reasonably well, even over a modem. It's also more secure - rather than a VNC login, you use your Windows login, just like it's local (except it isn't). It's about the closest thing available for Windows to a remote XDM.
however like most people I talk faster then I type.
I should hope so! Incredibly fast typists max out around 90 words per minute. Sit down and read 90 words out loud, and tell me how long it takes. 15 seconds?:-)
If you only read aloud as fast as you typed, you'd probably ride the short bus
Perhaps an important question is *why* do you need 64GB of addressable memory? Would it be possible to rewrite some of the code such that you could get by on a few gigabytes? It is possible to parallelize the task and spread it over several smaller systems (cluster?)
I worked for a company which was hitting the limit on Intel processors (1GB addressable per process). In their case, they were reading in huge GIS terrain databases. A better solution was for them to read from disk as needed, rather than load the entire database into RAM. While this may not be possible in your case, it is worth considering.
You know, I'd like to think geeks, especially Slashdot geeks, were beyond getting sucked in by marketing buzzwords.
Fine. Call it emulation. Regardless it serves the same purpose, and doesn't detract from the argument. I'm simply arguing that there is a valid use for hardware emulation.
Or, call me crazy, you could just use an actual, existing RISC layer and do away with all the overhead of emulation.
Ahhh yes, but you would get the benefits of mass production (lower cost) because EVERYONE would use one of these chips. Instead of paying $100 for 10,000 StrongARMs, you could pay $50 for 10,000 Transmeta CPUs with a RISC emulation engine, since production costs could be spread throughout 10,000,000 processors.
In addition, as bugs are discovered, or optimization made, new revisions of code could be updated on the current CPUs. I bet Intel would've saved quite a bit of money and embarrassment over their FPU "bug" back in the Socket 5 days.
I don't think Transmeta's technology is as beneficial today as it could be in 5 or 10 years as processor speeds continue to increase and their code is optimized. I can't possibly believe a 230MHz StrongARM chip will come anywhere near a 2 GHz Transmeta CPU with RISC emulation.
That being said, as processors get faster, I think there's an inherent advantage to additional abstraction layers. Perhaps I'm wrong (wouldn't be the first time), but lets face it - few people code in assembly anymore, but everyone uses C or C++ - languages which _abstract_ the low level details - things the compiler takes care of.
No doubt you expect Java to have some sort of lasting impact on the industry as well . . .
You mean the idea of writing code which works on any platform without regard to hardware or operating system? Like it or not, that's the future. C#, Java... heck, even HTML and Javascript to an extent all work on that principal. Unless you are willing to ceed to Microsoft and use their _one_ platform without regard to interoperability, better face the fact that platform-independant code is the non-Microsoft world's best friend.
I don't understand your point. You say "the objective is to use as little hardware as possible to keep the price down" and then note that the "cost of making software is significantly less than adding extra hardware". I'm not an embedded engineer, but if doing software emulation is *cheaper* than specialized hardware, why not do software emulation?
Assuming you can get comparable performance, you can design a subsystem which meets your needs without necessarily having to deal with the design limitation imposed by others. Since we've already determined that the underlying hardware is FAST, and than most of the processing occurs in the software, if you optimize the software, you optimize the hardware. Plus (in a sense), if the code morphing technology was open, you would essentially end up with an open source processor. What could be cooler than that!
It would be a real shame to lose the most important piece of Transmeta's technology - the code morphing. Lower power was just a side benefit.
Image being able to design a totally new architecture unique to your specific application. Utilizing Transmeta's technology, you could design a specialized interface to the hardware, unique to your application, and then build a software platform around it.
Sure, that doesn't seem useful to someone running Windows applications, but think about how easy it would be to create specialized embedded devices. If you needed a processor with only 30 instructions, instead of the 4 billion provided by present-day CISC technology, you could create a pseudo-RISC layer on top of the chip and write software optimized for those procedures.
I'll be very disappointed if, in 30 years I find myself thinking how it should've revolutionized the industry, but was instead forgotten about.
Simple answer: Find something non computer science related to do.
What aspect of CS do you dislike? Programming? There are tons of non-programming jobs out there. As a sysadmin, other than the occasional Perl script, I don't write any code.
Really, just because you graduate with a CS degree does NOT mean that you need to go out and become a programmer, or even need to find a CS-related job. Ironically, I know a few English majors who are now brilliant sysadmins.
I don't mean to sound like a Troll, but if you're that close to graduation, then finish college and look for something which you like. A college degree is just proof that you can learn.
Why remote control? Can't I use my laptop and a browser? Especially with wireless home network...
Because we're talking about consumer electronics. Because consumer electronics means consumers can just plug it in and go. Let's face it, computers are NOT consumer electronics. This box is supposed to be a stereo component, and as such, you need an easy way to sit on your sofa and turn it on, skip tracks, and at least manage basic CD-style functions.
IMO, a web browser is good for creating playlists, but it's not practical for a component system. Heck, you can't even *assume* everyone who will purchase this will have a computer or a network configured.
5. The United States will publish a notice informing the public of the proposed Final Judgment and public comment period in the Washington Post and the San Jose Mercury News, for seven days over a period of two weeks commencing no later than November 15, 2001.
6. Members of the public may submit written comments about the proposed Final Judgment to a designated official of the Antitrust Division of the United States Department of Justice for a period of 60 days after publication of the proposed Final Judgment and Competitive Impact Statement in the Federal Register.
7. Within 30 days after the close of the 60-day public comment period, the United States will file with the Court and publish in the Federal Register any comments it receives and its response to those comments.
Here's your opportunity to submit intelligent comments about their anti-competitive actions which non/less-technical people will read. (no offense to/., but the Post has a much larger and more diverse readership).
Slashdot Mirror
I can't answer the specifics of the question (whether or not the quality is in fact different), but I have found that DVD players, much like CD players, are all really the same. You pay for features, and maybe a little better error correction, but otherwise they're all pretty much the same quality.
I have an Apex DVD player (the one that lets you run without copy protection or encryption), which cost me $180 (at the time, most players were selling for around $250, so it was unbelievably cheap). It's very good quality, and works very well (though I hear the special edition Abyss is one of the few DVDs it has trouble with).
If you're interested in saving a few bucks, I'm pretty sure the DVD player in the XBox or PS2 is as good as any basic player out there (though I'm not sure if a remote is included with the unit). The Gamecube is $100 cheaper, but if you pick up a $100 DVD player, you're about even. If you want something special, like progressive scanning, then you'll need to get a seperate player.
Although it (appears) true that RF energy can cause problems with cells in the body, the amount of power in an 802.11b or Bluetooth antenna is FAR less than cellular equipment. A microware cooks food at 1000 watts @ 2.5 GHz, but long term cell phone use at 1 watt is debatable (probably not good for you, anyway), and long term use of 802.11 equipment at 0.05 to 0.1 watts is most likely insignificant. Plus, RF signals degrade at 1/(n^4), so even 1 foot away from an antenna is significantly less power (though I won't debate how much radiation my left leg receives from my Lucent PCMCIA wireless card in my laptop - still, the convenience to me is with the miniscule risk).
There are three sets of possible issues you might encounter: Hardware, Software, and Compatibility. It looks like you've considered some of the software issues, though I'm not sure you've considered the harware or compatibility issues yet.
Hardware issues probably center around reliability, more than anything. If you're purchasing a 3 year old disk array which received constant use, then you can expect that you WILL encounter hardware issues - if not now, sometime during its useful life. You might just encounter a bad disk, or there may be larger problems (like with the head unit). I've seen NetApps which are only a few WEEKS old lose disks (not much EMC experience here).
You may need/want to update firmware or patch the internal OS of the unit. Be warned that you may need to purchase some sort of support contract to get patches and updates. Also, hardware like this is notoriously touchy about being moved - so once you get it set up, don't tear it down and set it up somewhere else, unless you're prepared to fight with it.
Finally, compatibility issues may arise. I've noticed that NetApps in particular are very touchy when using Linux NFS clients - very often the clients will need to be rebooted when they hang. I think this is more of a Linux NFS client code issue than a NetApp issue (Linux NFS is *slowly* getting better), but its something to keep in mind. Without knowing what hardware you are going to purchase, it's very difficult to tell you what to look out for, but my primary experience has been with NetApp (fairly reliable, and a huge pain to work with in a cross platform environment) and a little EMC (nifty, a little faster, and a lot more expensive).
Does everything in the world need to be "internet accessible" or "web-enabled"?!?!?
How long before phones start getting hacked or spread MS LookOut worms? How long before phone spamming becomes the norm?
*Sigh* I want the web for convenience. I want to web to make my life easier. I don't need the "cool" factor of every internet-capable device. I don't need my refrigerator ordering food for me, and I certainly don't need IRC on my phone. Frankly, I find it easier to CALL someone rather than attempt to type on a frickin' phone (or follow an IRC session with 50 people on one of those little phone LCDs). But I digress...
Am I the *only* person who feels this way?
This sounds fairly trivial, actually. From his specs, assuming a standard set of hardware, it wouldn't be too hard to build a stock kernel, X with standard SVGA, use the splash screen patches to hide the messages, and install rdesktop (which, for those of you who are unfamiliar, is similar to the Citrix client).
:-)
In fact, he could even do a thin-client boot from the network and go diskless (which would make maintenance even easier), though for his specs, mounting most the partitions read-only and running reiser or ext3 on anything which changes would probably suffice).
I'm fairly sure I could do this in very little time, but I'm sure someone with more free time (since this project isn't paying enough for those of us who need to pay our rent) will beat me to it
I'm not so sure this is true. Despite the fact that the majority of the hardware is based on x86 architecture, there are a number of differences, including the shared system/video memory architecture. A "comparable" Intel system won't perfom nearly as well - x86 hardware is designed to perform every task in a mediocre fashion instead of performing a few, specialized tasks very well. The X-box probably would do a TERRIBLE job of running MS Office, but it does a much better job of playing games.
In addition, I think there are several benefits of a console system (though I desperately despise both Microsoft and the X-box):
I fully agree that a real computer seems to be a better investment than the X-box, but if all you're doing is playing games, a console like this really is better suited to the task.
Enough already! I don't need yet another web-enabled device. Okay, firewire from a DV cam is useful if I want to edit video, but I don't need to cruise the web on my handycam, and the last thing I want to hear is how the next Outlook worm has managed to hack my camera!
Huh?
Sendmail is neither buggy nor slow nor full of security holes. I've worked with it on systems handling up to ~2 million messages per week on just a few ultra 2's, and it rarely broke a sweat. And, there hasn't been a sendmail security issue in something like 4 years. Yes, the configuration syntax is difficult if you're not used to it (although it's MUCH easier if you DON'T EDIT THE sendmail.cf FILE and instead EDIT THE sendmail.mc FILE AND REGENERATE THE sendmail.cf FILE FROM IT), but once you learn it you understand *why* the majority of MTAs are sendmail (or some variant). I've tried qmail and postfix, and I still use sendmail, so go figure.
Bind has had its share of security holes (and will probably have more), but is also neither slow nor buggy. I know PLENTY of ISPs that run BIND on literally hundreds of thousands or even millions of entries without any trouble. I've read dan's docs, point of view, preaching, etc, and I just don't buy it.
The beauty of Unix is that there are always multiple ways to get the job done. I prefer to use the time tested, proven method. I know when I have trouble, I can find an answer to my Sendmail, Apache, and BIND questions *without* having to e-mail the development lists for help - and that's helpful when you're doing an install or upgrade on a weekend when those people aren't around. IMO, I shouldn't have to go to the product-specific mailing list for answers, and I like knowing that my company will have an easy time finding a successor who understand the products when I move on.
Get Dan to change his ridiculous license and get a reasonable percentage of the syadmin population using his products, and then I'll consider trying them out again.
Get the bastille firewall toolkit. You can configure your rules by editing one very well documented file, and it can generate the rulesets for IPChains or IPTables (depending on your kernel version). It's available at:
http://www.tux.org/~peterw/#fwall
I've used it for IPChains and then migrated right to IPTables - it does a great job, IMO (well, none of the systems I've configured it on have been hacked, so I suppose that's a good sign). You may still want to review the rules once it's complete, but it's an excellent starting point
I'm not saying BIND is perfect, but it's robust, scales well, is easy to support, and has a proven track record. I've never had BIND up and die on me or choke even under the heaviest of loads. In addition, BIND 9 was a "complete" rewrite - I don't believe it's shown any security holes, and it offers all the features required in a DNS server.
I'll take Sendmail, Apache, and Bind over the alternatives any day.
Does anyone know if Apple intends to offer firmware updates to current Airport owners (or if this is even possible), or if those who previous shelled out $299 for an Airport will be out in the cold?
Rather than multiple competing standards for the *next generation* of devices, why won't someone devise a bus technology which won't be obsolete in two or three years? While PCI-X doubling the bandwidth to 1GB sounds great, with upcoming 10 Gb network adapters, you've ALREADY reached its limit WITHOUT ANY OTHER PERIPHERALS! Imagine what will happen when 100 Gb is standard!
In the mean time, how about using some proven technology now? Five years ago, SGI was using multiple crossbars on their workstations to improve throughput, reduce contention, and decrease latency.
Why can't something like this be implemented with the current PCI architecture?
Okay, PCI-X sounds neat, and backwards compatibility is a plus, but if you read the article it states that PCI-X will run at the lowest common denominator - if you have three 133MHz/64-bit devices (up to 1.06 G/sec transfer) and one 33MHz/32-bit PCI device you're stuck at 33MHz. Four PCI cards, four crossbars, and you can get any speed you damn well like on any slot. Otherwise, the current crop of PCI modems and low-speed network adapters will end up limiting this anyway!
The remaining protocols sounds nifty, but I don't see any performance specs. As hopeful as I am that *something* good will replace PCI, I'm going to be very disappointed if there are too many standards and vendors can't agree on something.
Someone asked me this same question before, and he's what I told them:
:-)
> I would like to start working toward relevant certifications, but I am not
> sure where to start. I have been researching the RHCE certification,
> which would seem to me to be a good place to start, but I am not really
> sure. What do you think? What would your suggestions be? MCSE, Cisco,
> RedHat, or something else?
Really depends on what you want to do. I'm a Unix guy. I love Unix. If
I had the time (and someone to pay for it), I'd get certified in RedHat,
Solaris, AIX, HPUX, and anything else I could find.
But, there are more MSCE jobs out there. That's not to say that having an
MCSE is better. If you don't mind working with Windows, and can (at least
superficially) buy in to the Microsoft (learning) mentality, then it's not
such a bad deal. I personally lost faith in the MCSE exams when I took
the 2000 exam without ever touching 2000 or opening a book and passed it
with around an 80%.
There are a lot of employers that won't consider you for an NT/2000 job
without one. Do they realize that the exam is really a joke for
experienced IT guys? Probably not. Especially for recruiters, it's easy
to say "well, person A must be qualified because they're certified, even
though person B has more experience".
Then again, as the economy slows, there are a lot of people who were never
really qualified to be IT guys who are MCSE and are out of work. There
are a lot fewer qualified unix people out there, so even though there are
less jobs, it's easier to find work. Don't know if that makes sense, but
think like this:
If there are MCSE 10,000 jobs, and 15,000 MCSEs, only the top 66%
can find work. However, if there are 1000 Unix jobs, and 1500 unix
people, well, it's easier to get in that top 66%, because you only have to
elbow your way above the other 500, not 5000. Sure there's still a
percentage who are incompetent, but if you get enough experience with
different Unix flavors, you will always be needed.
Just to give you an idea, I've worked with Linux for 3.5 years now,
FreeBSD for a year, dabbled in OpenBSD, Irix for 4 years as an end-user
and 6 months as an admin, Solaris for about a year (SPARC and x86), HP-UX
and AIX for 6 months, and some VAX/VMS in college. Sounds much more
impressive than "3-4 years of Windows 9x/NT/2000". And the truth is, once
you've learned the concepts of one Unix, the others are pretty easy to
guess.
So, back to your original question: What do you want to do? Do you want
to get certified to work with Unix, for added job security, extra pay,
etc? Are you comfortable with Windows? Do you like working with it? You
don't really have to hate Microsoft just because 5% or 10% of us do. One
of the things I love about Unix is that Unix guys typically make $15k more
than MCSEs because there's more of a demand.
If I was going to be a Windows guy forever, I'd get an MCSE, possibly +I,
and probably a CCNE (I think the path is CCNA->CCNP->CCNE now). At the
very least, take the first couple MCSE tests and be a basic MCP, and get
your CCNA.
If you like Linux, and think that's the path you want to follow, then I
have some better advice for you: Go buy a Sparcstation 10 or 20 off ebay
and some other hardware. Be prepared to spend $1000, but get yourself a
P200 you can play with, a Sparcstation 10, and maybe an SGI Indy and an HP
A-class (or some other basic HP box). Do a little research first, to find
out what OSes run on what hardware (google is wonderful for this kind of
thing), and check out the spec.org pages for speeds (so you don't buy
something too old). Make sure you get copies of media (OSes), or buy
them, install the OSes on each box, and start working with them.
Dedicate an hour a day to doing things like trying to build SSH, learn how
thei admin tools work on that OS, and everything else. Configure the
network. Convert your P200 to a firewall, configure firewalling, and use
it to cruise the web. My firewall at home is a Pentium 133 with 32 megs
of RAM,and it uses a modem for the external internet connection.
Your goal should be to be able to build a little network at home, make
sure all the machines can communicate, and can interoperate. Try to set
up your own DNS. Configure Apache. Play with Sendmail. Try to get a
feel for basic admin things. Once you get really comfortable, teach
yourself a little Perl or shell scripting. A little Perl goes a long way.
You don't have to be a guru, but knowing how to read a file, parse it for
key words, and reformat it can go a long way.
Figure it will take a year to really learn the systems, but you'll learn
very quickly being forced to work with them yourself. Six months of
administering your own systems, and trying to run all your own services is
like 2 years of work experience. Don't be afraid to ask for help, but
always try to find it yourself - Unix gurus usually don't mind being asked
once where something obvious can be found, but when they have to
repeatedly point someone to websites which could be found on google, man
pages, and howto's which are included in the distribution, they tend to
get bitter
After you learn everything, you can go take the exams for a few hundred
dollars, and you'll be able to pass them with your eyes closed. The key
is not spewing out what you read in the book - it's understanding the
systems. Once you understand the underlying questions, you may very well
know the answer without ever reading the books or touching the system.
Did I answer your question, or confuse you more?
Is shrinking equipment like this actually useful to most companies? I'm curious, because my prior employer, had lots of rack space, but the colo site only had ~20amps per rack available. With 9' racks, it was very easy to run out of power BEFORE space was an issue. They (above.net) claimed their colo was designed for 6-8U "servers", not 1U "pizza boxes".
Aside from power, how about heat dissapation? After all, a bunch of laptops theoretically would make EXCELLENT servers, since they even have 1-2 hour battery backups built right in, are extra small, and don't even require a KVM (they could just be pulled out as needed). BUT, a laptop runs waaaaaaay too hot to be used as a server, and their price/performance ratio is aweful, even if you do save on rack space.
Finally, do these have any redundancy built in? Is there anything else special about them? Personally, I'd rather have a 1U rack unit which is dual CPU, dual NIC, and dual power supply than two of these. It would probably be cheaper, last longer, and be less hassle.
Just my $0.02 from my past colo experience.
So, when will the government fund a probe to uranus?
Perhaps a better (and more professional) solution is something like Citrix. VNC limits you to one connection at a time, while Citrix allows multiple sessions simultaneously. In addition, Citrix is probably the fastest VNC-style thing I've used, and the Metaframe protocol works reasonably well, even over a modem. It's also more secure - rather than a VNC login, you use your Windows login, just like it's local (except it isn't). It's about the closest thing available for Windows to a remote XDM.
however like most people I talk faster then I type.
I should hope so! Incredibly fast typists max out around 90 words per minute. Sit down and read 90 words out loud, and tell me how long it takes. 15 seconds?
If you only read aloud as fast as you typed, you'd probably ride the short bus
5. Your wife is on line 1
4. Your ex-wife is on line 1
3. Your proctologist is on line 1
2. You've got mail... pattern baldness
1. You've got spam
Perhaps an important question is *why* do you need 64GB of addressable memory? Would it be possible to rewrite some of the code such that you could get by on a few gigabytes? It is possible to parallelize the task and spread it over several smaller systems (cluster?)
I worked for a company which was hitting the limit on Intel processors (1GB addressable per process). In their case, they were reading in huge GIS terrain databases. A better solution was for them to read from disk as needed, rather than load the entire database into RAM. While this may not be possible in your case, it is worth considering.
You know, I'd like to think geeks, especially Slashdot geeks, were beyond getting sucked in by marketing buzzwords.
Fine. Call it emulation. Regardless it serves the same purpose, and doesn't detract from the argument. I'm simply arguing that there is a valid use for hardware emulation.
Or, call me crazy, you could just use an actual, existing RISC layer and do away with all the overhead of emulation.
Ahhh yes, but you would get the benefits of mass production (lower cost) because EVERYONE would use one of these chips. Instead of paying $100 for 10,000 StrongARMs, you could pay $50 for 10,000 Transmeta CPUs with a RISC emulation engine, since production costs could be spread throughout 10,000,000 processors.
In addition, as bugs are discovered, or optimization made, new revisions of code could be updated on the current CPUs. I bet Intel would've saved quite a bit of money and embarrassment over their FPU "bug" back in the Socket 5 days.
I don't think Transmeta's technology is as beneficial today as it could be in 5 or 10 years as processor speeds continue to increase and their code is optimized. I can't possibly believe a 230MHz StrongARM chip will come anywhere near a 2 GHz Transmeta CPU with RISC emulation.
That being said, as processors get faster, I think there's an inherent advantage to additional abstraction layers. Perhaps I'm wrong (wouldn't be the first time), but lets face it - few people code in assembly anymore, but everyone uses C or C++ - languages which _abstract_ the low level details - things the compiler takes care of.
No doubt you expect Java to have some sort of lasting impact on the industry as well . . .
You mean the idea of writing code which works on any platform without regard to hardware or operating system? Like it or not, that's the future. C#, Java... heck, even HTML and Javascript to an extent all work on that principal. Unless you are willing to ceed to Microsoft and use their _one_ platform without regard to interoperability, better face the fact that platform-independant code is the non-Microsoft world's best friend.
I don't understand your point. You say "the objective is to use as little hardware as possible to keep the price down" and then note that the "cost of making software is significantly less than adding extra hardware". I'm not an embedded engineer, but if doing software emulation is *cheaper* than specialized hardware, why not do software emulation?
Assuming you can get comparable performance, you can design a subsystem which meets your needs without necessarily having to deal with the design limitation imposed by others. Since we've already determined that the underlying hardware is FAST, and than most of the processing occurs in the software, if you optimize the software, you optimize the hardware. Plus (in a sense), if the code morphing technology was open, you would essentially end up with an open source processor. What could be cooler than that!
It would be a real shame to lose the most important piece of Transmeta's technology - the code morphing. Lower power was just a side benefit.
Image being able to design a totally new architecture unique to your specific application. Utilizing Transmeta's technology, you could design a specialized interface to the hardware, unique to your application, and then build a software platform around it.
Sure, that doesn't seem useful to someone running Windows applications, but think about how easy it would be to create specialized embedded devices. If you needed a processor with only 30 instructions, instead of the 4 billion provided by present-day CISC technology, you could create a pseudo-RISC layer on top of the chip and write software optimized for those procedures.
I'll be very disappointed if, in 30 years I find myself thinking how it should've revolutionized the industry, but was instead forgotten about.
Simple answer: Find something non computer science related to do.
What aspect of CS do you dislike? Programming? There are tons of non-programming jobs out there. As a sysadmin, other than the occasional Perl script, I don't write any code.
Really, just because you graduate with a CS degree does NOT mean that you need to go out and become a programmer, or even need to find a CS-related job. Ironically, I know a few English majors who are now brilliant sysadmins.
I don't mean to sound like a Troll, but if you're that close to graduation, then finish college and look for something which you like. A college degree is just proof that you can learn.
Because we're talking about consumer electronics. Because consumer electronics means consumers can just plug it in and go. Let's face it, computers are NOT consumer electronics. This box is supposed to be a stereo component, and as such, you need an easy way to sit on your sofa and turn it on, skip tracks, and at least manage basic CD-style functions.
IMO, a web browser is good for creating playlists, but it's not practical for a component system. Heck, you can't even *assume* everyone who will purchase this will have a computer or a network configured.
6. Members of the public may submit written comments about the proposed Final Judgment to a designated official of the Antitrust Division of the United States Department of Justice for a period of 60 days after publication of the proposed Final Judgment and Competitive Impact Statement in the Federal Register.
7. Within 30 days after the close of the 60-day public comment period, the United States will file with the Court and publish in the Federal Register any comments it receives and its response to those comments.
Here's your opportunity to submit intelligent comments about their anti-competitive actions which non/less-technical people will read. (no offense to /., but the Post has a much larger and more diverse readership).