I was about to make a joke about you messing up the measurement systems and then I realized that was the entire purpose of your post. *smack self* Evidentally the caffiene and sugar have not taken affect yet; must drink more!
Both Oracle 8i and IBM's DB2 use Java extensively both for their DB administration GUIs as well as for middleware code. If you didn't know, these are the number 1 and number 2 Enterprise database systems in the world.
Have you actually used Oracle's tools much? There are SEVERAL problems with them and IMHO they demonstrate exactly how NOT to use Java.
1. The administration tool is SLOOOOOW.
2. The installation tool is slow and you have to have the proper version of the install tool in order to install specific versions of their (freely avaliable) drivers. The install tool is not (last time I checked) freely downloadable.
Of course the install tool issues (besides speed) are really examples of poor design/distribution on Oracle's part but the speed issue still exists. I sincerely question the wisdom of Oracle wasting time re-inventing the wheel by writing their own install tool.
I have seen a few Java applications that functioned correctly but I have yet to see a Java application that approached anything I would call acceptable speed (and I'm talking about P3-450 and P3-600 systems). This is both with MS's JVM and Sun's.
Interpreted languages will ALWAYS be slower than natively compiled languages (Given the same quality of code)
It is the responsiblity of the developers/architects to make wise decisions as to when the loss of performance is worth the time saved (if in fact any time would be saved). Most of the Java projects I've seen have all come about due to the buzz factor of Java, not because it was the best solution for the problem at hand.
On a side note, I resent the implication that just because Oracle and IBM make two of the most popular products in the market that they are inherently writing 'cool' or better applications. I believe MS is #1 is several markets but there are at least one or two/. members that would agree that they don't have the best or coolest products.
I'm sure we all know this but it's a well documented fact that companies fix publicly known security holes MUCH faster than publicly undiscovered (but known by the company) security holes.
I suppose the question is which method does more damage?
With publicly known vulnerabilities you get a lot more script kiddie attacks but I would bet there are a lot fewer serious attacks due to the fact that sysadmins should know what to look for. On the other hand if it's not public knowledge the amount of serious attacks are probably much greater. If the public doesn't know about it then sysadmins don't know how to prepare for it or monitor it (if they can) untill a fix is released.
-Zane
Re:review of Katz's review (summary: katz != siske
on
Slashdot Meets X-Men
·
· Score: 1
you idiot, we AREN'T supposed to hate Magneto. Even Senator Kelly's initial position is a REASONABLE one. This is why X-Men is so powerful as a story. The issues and moral stands are ambiguous. The reason for showing us Poland 1944 was supposed to give you the context if you had been paying attention. This was not a movie about black and white good vs. evil. Perhaps you're just incapable of seeing that however. A simple flick with neatly structured plot of Good Guys vs Bad Guys is what you'd prefer?
LOL! This is exactly what I was thinking when I read that line. You're not supposed to hate him, that's the whole point. I love it when movies present these kind of issues to the audience. Not all "bad guys" are hate-mongering evil-loving monsters. In this case we see a pretty rational, intelligent person take a course of action he believes is right. In fact, he thinks he's even giving the normal humans a gift by making them mutants.
I only wish they hadn't made Magneto so ready to take life. I suppose they put that in so that people like Katz could figure out he was supposed to be the bad guy.:)
Are there any Secure SMTP systems out there? Basically an architecture somewhat like SSH in that the protocol basically remains the same but operates on a different port and encrypts all incoming/outgoing communctications and possibly encrypts the message store. If we could come up with a standard system for this (with the encryption portion developed outside the US in a country with lax encryption export control) it would solve a lot of the wiretapping/sniffing issues with e-mail. It's not difficult to do and since user's wouldn't have to interact with it directly it shouldn't be hard to implement seamlessly.
As for encrypting to the client that's a little more difficult since you'd have to have client-side support but still easily doable. Eventually things like this will become standard, so why not do it now?:)
You wouldn't want to dimple a car, it would probably produce too much drag over smooth panels.
I read several months in (in Popular Science IIRC) that airplane manufactures were finding that less drag was produced from dimpled surfaces on their airplanes in wind-tunnel tests. They cited the golf-ball's dimpled surface in the explanation. If this could be applied to airplanes where wind-resistance is a little more of a factor, could not the same thing be applied to cars?
Frankly even if it gave me 10 miles/gallon more I wouldn't drive it if it looked as bad as I think it would look.:)
Nonsense: I'm talking about the HTML active content enabled preview pane Microsoft came up with. Even if they have entirely disabled this, the fact that they did it and shipped it makes the email virus a reality.
Yes, but this pane does not run anything that can't be run in a standard browser. It uses standard security settings which for the most part are OK. Even with "secure" ActiveX controls the standard setting is to ask the user if they want to install. I get asked all the time "Do you want to run this Microsoft Corp control?". I'm not saying it's 100% secure, I'm sure there are exploits that haven't been found, I'm just saying there are no current major exploits. Neither mellisa or ILOVEYOU took advantage of the preview pane. In fact, I have yet to see an e-mail virus that could take advantage of that pane.
Yes, if you're an idiot and you set yourself to low security then you could be in trouble, but then you deserve it for screwing with something you know nothing about.
A lot of people fear that preview pane but I have yet to see anything truly evil come from it. I am not a MS fan but I'm tired of hearing people come up with religious claims about the security of their products. If you can demonstrate one virus that uses the preview pane then I'll eat my works. My e-mail address is listed above.
That approach is what gave us the email virus- which was once a _myth_.
That myth is still very much a myth. The original hoax was to the effect of "Open any email with subject and your computer is toast!" The most recent e-mail virii have all been attachments that require a user to open them, just like any other trojan (or attachment with a virus). The only difference with this virus was that it used a non-standard executable type.
Good Lord, they charge enough for certifcation as it is. You really want to make a business that would ONLY profit off of certifications? You'd end up paying $200/test that expire in 6 months.:)
Java virtual machines? Every major OS has now successfully integrated Java, but now Windows can't? Does that mean I have to run my Java apps within IE, or application providers have to include a VM with their package?
Doesn't Windows allow OS integration for non-MS apps? Yes, it does. For instance, look at all the shell extensions, drivers, services, third-party DLLs, etc... There is a HUGE difference between techinically integrating something with the OS and integrating something from a marketing/sales standpoint. This decision basically forces MS's OS company to deal with it's applications company as it would deal with any third-party company. What this will lead to is a more open and flexible Windows.
Metaframe is the biggest example of an OS Integrated 3rd party app.
Netscape distributes a JVM with it's browser and it's not "Integrated with the OS" in the manner you seem to be describing yet it functions just fine (well it functions how it was programmed to function anyway:) so I fail to see why having MS's JVM seperate from the OS affects anything.
One thing I haven't seen mentioned very much yet is the benefit of Delphi/C++Builders debugger. Their debugger is very similar to MSVC's and *shudder* VB. You can compile and run your apllication and step through the code line by line watching all your variables values as they change or add conditional breakpoints or just wait untill the app crashes and debug from there. From what I've seen on the linux side (and admittedly it's been a while since I looked for a graphical IDE with a well integrated debugger for Linux) nothing else come close to that level of debugging under Linux.
And please don't say that xgdb is any comparison. Unless you've used Delhpi, C++Builder, MSVC or VB for debugging then you can't quite compare. I have used all of those (includding xgdb).
Because there is little to no defense once it gets to your router. Sure you can block ICMP traffic or traffic from specific subnets but that blocking only happens at a specific router (be it yours or a router at the ISP's ISP). Somewhere that bandwidth is being swamped by the DOS (usually at your router or inside your network if your router isn't filtering ICMP). As a few people have pointed out in the past the only really good way (currently) of stopping DDOS attacks (the main culprits, normal DOS/flood attacks that aren't distributed usually are much much lower bandwidth) is having EVERY router on the internet (or at least at major ISPs like MCI/Sprint/etc...) setup not to allow packets from inside their network to be sent out if the source address on the packet is not from their network.
Routers that don't properly validate outgoing traffic are the main culprits in these type of attacks.
What is the minimum size per minute of footage for decent visual quality video, for MP4 and for its predecessor?
Well the major issue with this question is what/who defines "decent visual quality"? This is subjective from person to person. You have to take into account frame rate (most noticable below 30 for most people), color distortion/banding, resolution, etc..
Once again I've commited nothing usefull to this discussion.:)
The company 'Infinity' makes those speakers. I've got a full set of Infinity speakers for my surround sound setup at home and in my car (replaced my factory speakers with them).
4000 PCs cannot be a viable economic replacement. That amount of hardware would require as highly a specialised environment as that of a mainframe (cooling and electricity), and certainly much more real estate.
The specialized hardware required is no more than any standard server room. Most decent server rooms have a minimum of a raised floor, several rack cabinets and climate control. The space required for these systems is not as much as it may seem.
Say you're using a 34U 19" rack to hold these servers. If they're using 1U servers for each node (very probably) then each rack holds 34 nodes. That's 118 cabinets. Each cabinet is approx 2'x3' and if we assume 2x's the cabinet size for total space (walkways and such) then that's only ~1400 square feet or the size of a large apartment. That's not abysmally huge for a server room. If you consider that they might be using 1Ux4 servlets (4 nodes per U) then we're down to 25% of that estimate.
Things could be much better or worse than I've depicted but the point is it's very doable in a reasonable amount of space with standard equipment.
As for your suggestion that they move to a high capacity single server or mainframe that's rather ludicrous. Assume they're paying $2500/node (a very large assumption) that's $10,000,000. From what I've read about mainframes that won't buy you much.
Have you bothered to check the list of the top 500 most powerfull super-computers? Several Linux clusters are on that list (and very high up the list at that) and these clusters were built for FAR less money than most (if not all) of their peers.
A well designed cluster system is very easy to maintain. You maintain one (or several) NFS root servers and every system remote boots. To change the configuration of one node you change it at one point. To install a new node you simply install the new hardware, make a couple of simple configuration changes and you're set. When you've done 4000 of them already another node probably takes less than 15-30 minutes to install.
Ack, that was actually 3 paragraphs, wish this damn thing defaulted to text. Here's the properly formatted readable version:
Finally when we isolate one of those nasty "That's the way our product handles it" kind of performance bugs we can fix it!! Oracle has a VERY nasty one where for every row in a parent table all children tables are checked for data that would prevent the delete (in a FK parent/child relationship) regardless of wether or not those tables were empty. We had a delete on a 35,000 row table that took 25 minutes on a SUN E4500 w/ 6GB RAM and an optical disk array with a HUGE cache (something like 16GB) even though we deleted the children table right before deleting the parent table. Placing indexes on the FK's fixed the problem but still, it should only have to check those tables once and see 'It's not empty' then stop checking.
Maybe we'll actually have an enterprise-level database that allows us to fix ugly crap like that ourselves instead of waiting for some stodgy company to decide your particular issue is high enough priority.
I just don't understand why more companies don't go Open Source. Yes, I know it's a scary concept letting everyone see your source but legally no-one has any rights to that source that you don't give them. Your competition cannot legally steal your work (though wether or not they could get away with it quietly is yet to be seen). Really it does nothing but make your product a better product and attract people like myself who love having the code handy.
Finally when we isolate one of those nasty "That's the way our product handles it" kind of performance bugs we can fix it!! Oracle has a VERY nasty one where for every row in a parent table all children tables are checked for data that would prevent the delete (in a FK parent/child relationship) regardless of wether or not those tables were empty. We had a delete on a 35,000 row table that took 25 minutes on a SUN E4500 w/ 6GB RAM and an optical disk array with a HUGE cache (something like 16GB) even though we deleted the children table right before deleting the parent table. Placing indexes on the FK's fixed the problem but still, it should only have to check those tables once and see 'It's not empty' then stop checking. Maybe we'll actually have an enterprise-level database that allows us to fix ugly crap like that ourselves instead of waiting for some stodgy company to decide your particular issue is high enough priority. I just don't understand why more companies don't go Open Source. Yes, I know it's a scary concept letting everyone see your source but legally no-one has any rights to that source that you don't give them. Your competition cannot legally steal your work (though wether or not they could get away with it quietly is yet to be seen). Really it does nothing but make your product a better product and attract people like myself who love having the code handy. -Zane
As I told a friend the other day, if I could write an ILOVEYOU variant that killed everyone who executed it I would solve 90% of the world's problems.:)
Slashdot Mirror
Dorf on Writing is MUCH MUCH better.
I was about to make a joke about you messing up the measurement systems and then I realized that was the entire purpose of your post. *smack self* Evidentally the caffiene and sugar have not taken affect yet; must drink more!
-Zane
Have you actually used Oracle's tools much? There are SEVERAL problems with them and IMHO they demonstrate exactly how NOT to use Java.
1. The administration tool is SLOOOOOW. 2. The installation tool is slow and you have to have the proper version of the install tool in order to install specific versions of their (freely avaliable) drivers. The install tool is not (last time I checked) freely downloadable.
Of course the install tool issues (besides speed) are really examples of poor design/distribution on Oracle's part but the speed issue still exists. I sincerely question the wisdom of Oracle wasting time re-inventing the wheel by writing their own install tool.
I have seen a few Java applications that functioned correctly but I have yet to see a Java application that approached anything I would call acceptable speed (and I'm talking about P3-450 and P3-600 systems). This is both with MS's JVM and Sun's.
Interpreted languages will ALWAYS be slower than natively compiled languages (Given the same quality of code)
It is the responsiblity of the developers/architects to make wise decisions as to when the loss of performance is worth the time saved (if in fact any time would be saved). Most of the Java projects I've seen have all come about due to the buzz factor of Java, not because it was the best solution for the problem at hand.
On a side note, I resent the implication that just because Oracle and IBM make two of the most popular products in the market that they are inherently writing 'cool' or better applications. I believe MS is #1 is several markets but there are at least one or two /. members that would agree that they don't have the best or coolest products.
-Zane
I'm sure we all know this but it's a well documented fact that companies fix publicly known security holes MUCH faster than publicly undiscovered (but known by the company) security holes.
I suppose the question is which method does more damage?
With publicly known vulnerabilities you get a lot more script kiddie attacks but I would bet there are a lot fewer serious attacks due to the fact that sysadmins should know what to look for. On the other hand if it's not public knowledge the amount of serious attacks are probably much greater. If the public doesn't know about it then sysadmins don't know how to prepare for it or monitor it (if they can) untill a fix is released.
-Zane
LOL! This is exactly what I was thinking when I read that line. You're not supposed to hate him, that's the whole point. I love it when movies present these kind of issues to the audience. Not all "bad guys" are hate-mongering evil-loving monsters. In this case we see a pretty rational, intelligent person take a course of action he believes is right. In fact, he thinks he's even giving the normal humans a gift by making them mutants.
I only wish they hadn't made Magneto so ready to take life. I suppose they put that in so that people like Katz could figure out he was supposed to be the bad guy. :)
-Zane
Are there any Secure SMTP systems out there? Basically an architecture somewhat like SSH in that the protocol basically remains the same but operates on a different port and encrypts all incoming/outgoing communctications and possibly encrypts the message store. If we could come up with a standard system for this (with the encryption portion developed outside the US in a country with lax encryption export control) it would solve a lot of the wiretapping/sniffing issues with e-mail. It's not difficult to do and since user's wouldn't have to interact with it directly it shouldn't be hard to implement seamlessly.
:)
As for encrypting to the client that's a little more difficult since you'd have to have client-side support but still easily doable. Eventually things like this will become standard, so why not do it now?
-Zane
I read several months in (in Popular Science IIRC) that airplane manufactures were finding that less drag was produced from dimpled surfaces on their airplanes in wind-tunnel tests. They cited the golf-ball's dimpled surface in the explanation. If this could be applied to airplanes where wind-resistance is a little more of a factor, could not the same thing be applied to cars?
Frankly even if it gave me 10 miles/gallon more I wouldn't drive it if it looked as bad as I think it would look. :)
-Zane
Yes, but this pane does not run anything that can't be run in a standard browser. It uses standard security settings which for the most part are OK. Even with "secure" ActiveX controls the standard setting is to ask the user if they want to install. I get asked all the time "Do you want to run this Microsoft Corp control?". I'm not saying it's 100% secure, I'm sure there are exploits that haven't been found, I'm just saying there are no current major exploits. Neither mellisa or ILOVEYOU took advantage of the preview pane. In fact, I have yet to see an e-mail virus that could take advantage of that pane.
Yes, if you're an idiot and you set yourself to low security then you could be in trouble, but then you deserve it for screwing with something you know nothing about.
A lot of people fear that preview pane but I have yet to see anything truly evil come from it. I am not a MS fan but I'm tired of hearing people come up with religious claims about the security of their products. If you can demonstrate one virus that uses the preview pane then I'll eat my works. My e-mail address is listed above.
-Zane
That myth is still very much a myth. The original hoax was to the effect of "Open any email with subject and your computer is toast!" The most recent e-mail virii have all been attachments that require a user to open them, just like any other trojan (or attachment with a virus). The only difference with this virus was that it used a non-standard executable type.
-Zane
Good Lord, they charge enough for certifcation as it is. You really want to make a business that would ONLY profit off of certifications? You'd end up paying $200/test that expire in 6 months. :)
-Zane
Doesn't Windows allow OS integration for non-MS apps? Yes, it does. For instance, look at all the shell extensions, drivers, services, third-party DLLs, etc... There is a HUGE difference between techinically integrating something with the OS and integrating something from a marketing/sales standpoint. This decision basically forces MS's OS company to deal with it's applications company as it would deal with any third-party company. What this will lead to is a more open and flexible Windows.
Metaframe is the biggest example of an OS Integrated 3rd party app.
Netscape distributes a JVM with it's browser and it's not "Integrated with the OS" in the manner you seem to be describing yet it functions just fine (well it functions how it was programmed to function anyway :) so I fail to see why having MS's JVM seperate from the OS affects anything.
-Zane
One thing I haven't seen mentioned very much yet is the benefit of Delphi/C++Builders debugger. Their debugger is very similar to MSVC's and *shudder* VB. You can compile and run your apllication and step through the code line by line watching all your variables values as they change or add conditional breakpoints or just wait untill the app crashes and debug from there. From what I've seen on the linux side (and admittedly it's been a while since I looked for a graphical IDE with a well integrated debugger for Linux) nothing else come close to that level of debugging under Linux.
And please don't say that xgdb is any comparison. Unless you've used Delhpi, C++Builder, MSVC or VB for debugging then you can't quite compare. I have used all of those (includding xgdb).
-Zane
Because there is little to no defense once it gets to your router. Sure you can block ICMP traffic or traffic from specific subnets but that blocking only happens at a specific router (be it yours or a router at the ISP's ISP). Somewhere that bandwidth is being swamped by the DOS (usually at your router or inside your network if your router isn't filtering ICMP). As a few people have pointed out in the past the only really good way (currently) of stopping DDOS attacks (the main culprits, normal DOS/flood attacks that aren't distributed usually are much much lower bandwidth) is having EVERY router on the internet (or at least at major ISPs like MCI/Sprint/etc...) setup not to allow packets from inside their network to be sent out if the source address on the packet is not from their network.
Routers that don't properly validate outgoing traffic are the main culprits in these type of attacks.
-Zane
Way too much Haiku :)
Someone please stop this madness
FIRST HAIKU FLAME POST!!
-Zane
Well the major issue with this question is what/who defines "decent visual quality"? This is subjective from person to person. You have to take into account frame rate (most noticable below 30 for most people), color distortion/banding, resolution, etc..
Once again I've commited nothing usefull to this discussion. :)
-Zane
The company 'Infinity' makes those speakers. I've got a full set of Infinity speakers for my surround sound setup at home and in my car (replaced my factory speakers with them).
:)
Yes, they are awesome speakers.
-Zane
Great, now we'll have Panty Raider 4D.
-Zane
The specialized hardware required is no more than any standard server room. Most decent server rooms have a minimum of a raised floor, several rack cabinets and climate control. The space required for these systems is not as much as it may seem.
Say you're using a 34U 19" rack to hold these servers. If they're using 1U servers for each node (very probably) then each rack holds 34 nodes. That's 118 cabinets. Each cabinet is approx 2'x3' and if we assume 2x's the cabinet size for total space (walkways and such) then that's only ~1400 square feet or the size of a large apartment. That's not abysmally huge for a server room. If you consider that they might be using 1Ux4 servlets (4 nodes per U) then we're down to 25% of that estimate.
Things could be much better or worse than I've depicted but the point is it's very doable in a reasonable amount of space with standard equipment.
As for your suggestion that they move to a high capacity single server or mainframe that's rather ludicrous. Assume they're paying $2500/node (a very large assumption) that's $10,000,000. From what I've read about mainframes that won't buy you much.
Have you bothered to check the list of the top 500 most powerfull super-computers? Several Linux clusters are on that list (and very high up the list at that) and these clusters were built for FAR less money than most (if not all) of their peers.
A well designed cluster system is very easy to maintain. You maintain one (or several) NFS root servers and every system remote boots. To change the configuration of one node you change it at one point. To install a new node you simply install the new hardware, make a couple of simple configuration changes and you're set. When you've done 4000 of them already another node probably takes less than 15-30 minutes to install.
-Zane
Nice moderation. This wasn't an attempt at flaming anyone or pissing anyone off, this was meant to be mildly funny.
:)
Maybe I have too much time on my hands. Maybe some of the moderators need to be a little less serious. Maybe we all could use a vacation.
-Zane
Grr.. Stupid plain old text post filtered out my:
(Brain implant activated by office assistant via visual cues...)
Was embedded in greater-than/less-than characters. Anyone know how to literally print those characters when post via plain old text?
-Zane
If they can just delay the sentancing for another two years they'll OWN EVERYTHING! MUHAHAHAHA!!
Please ignore the above garbage, there is nothing to see here, move along.
-Zane
Ack, that was actually 3 paragraphs, wish this damn thing defaulted to text. Here's the properly formatted readable version:
Finally when we isolate one of those nasty "That's the way our product handles it" kind of performance bugs we can fix it!! Oracle has a VERY nasty one where for every row in a parent table all children tables are checked for data that would prevent the delete (in a FK parent/child relationship) regardless of wether or not those tables were empty. We had a delete on a 35,000 row table that took 25 minutes on a SUN E4500 w/ 6GB RAM and an optical disk array with a HUGE cache (something like 16GB) even though we deleted the children table right before deleting the parent table. Placing indexes on the FK's fixed the problem but still, it should only have to check those tables once and see 'It's not empty' then stop checking.
Maybe we'll actually have an enterprise-level database that allows us to fix ugly crap like that ourselves instead of waiting for some stodgy company to decide your particular issue is high enough priority.
I just don't understand why more companies don't go Open Source. Yes, I know it's a scary concept letting everyone see your source but legally no-one has any rights to that source that you don't give them. Your competition cannot legally steal your work (though wether or not they could get away with it quietly is yet to be seen). Really it does nothing but make your product a better product and attract people like myself who love having the code handy.
-Zane
Finally when we isolate one of those nasty "That's the way our product handles it" kind of performance bugs we can fix it!! Oracle has a VERY nasty one where for every row in a parent table all children tables are checked for data that would prevent the delete (in a FK parent/child relationship) regardless of wether or not those tables were empty. We had a delete on a 35,000 row table that took 25 minutes on a SUN E4500 w/ 6GB RAM and an optical disk array with a HUGE cache (something like 16GB) even though we deleted the children table right before deleting the parent table. Placing indexes on the FK's fixed the problem but still, it should only have to check those tables once and see 'It's not empty' then stop checking. Maybe we'll actually have an enterprise-level database that allows us to fix ugly crap like that ourselves instead of waiting for some stodgy company to decide your particular issue is high enough priority. I just don't understand why more companies don't go Open Source. Yes, I know it's a scary concept letting everyone see your source but legally no-one has any rights to that source that you don't give them. Your competition cannot legally steal your work (though wether or not they could get away with it quietly is yet to be seen). Really it does nothing but make your product a better product and attract people like myself who love having the code handy. -Zane
As I told a friend the other day, if I could write an ILOVEYOU variant that killed everyone who executed it I would solve 90% of the world's problems. :)
-Zane