The number one killer of hard drives is heat. IBM did a study where for each 1 degree C over ambient, the MTBF dropped in half. If you want to make sure your drives last, make sure they are well cooled. Most modern drives have built-in temp sensors that you can query with various tools.
I'm convinced that Redhat will not hold this dominant position for much longer due to their insane pricing model.
I looked into 5 year costs for 70 servers and compared Windows and Redhat. Total cost of redhat was about 3 times more. Considering that my project really needed Linux, I went with Centos for nothing.
The problem RedHat has, is that the cost of supporting 70 servers with their OS is not 65 times the cost of one server (the price break is miniscule.) If they offered an alternate model of a site license with X number of incidents, I would have used them in an instant. The redhat model only works for companies with just a few servers and inexperienced staff.
Some other company with half a brain will take CentOS with a per-incident / per-hour pricing model and eat redhat alive. No, I don't think this company is Oracle.
The problem is that when it comes down to voting on important issues, it doesn't matter if you are a moderate republican or democrat - you are going to vote down party lines in *most* cases. If you don't, the party leadership comes to you and tells you that you are not going to get that $200M highway project your state needs, you won't be given a seat on a committee, etc. You have to play ball. That's what each and every politician has found who went in with a big set of ideals and promises of change.
To be honest, most AV software sucks anyway. It's bloated, slows the machine down, and doesn't handle zero-day viruses. AV software is great at closing the barn door after the cows already got out. Considering that so many attacks immediatly disable AV software, just what good is it?
AV software works fairly well on email servers to stop that vector, but a decent (non-windows) firewall and keeping up with patches is really all that is needed.
how do you guys get rid of these nasty rootkit and evolved spywares which can hide very well without reformatting
You don't. It is not worth the time and effort unless your personal / professional time has zero value. Get your data off and reinstall / restore from image.
Otherwise (if you are getting paid well for it) you can boot off a live CD or install the drive as a second in another system (one that has all the autorun crap disabled), Run AV/AS(pyware) on the drive, edit the registry removing all the startup items that you know isn't needed, run md5 comparisons on all the system files, and go from there. Dumping the registry and comparing with a known good registry is helpful at spotting crap.
It was probaby just a small handful of viruses (at most) that infected 7500 files. I would seriously doubt that the poster had 7500 DIFFERENT viruses. I can also believe that, at most, dozens of spyware programs were installed, but not hundreds. Spyware programs frequently claim that each cookie and registry entry is spyware.
Right now, the telco's own the lines and own the services. CLEC's can lease space at a huge price that does not allow true competition.
What is needed is to split the ILEC's into services, and physical infrastructure. Require the PHY telco's to build-out the rest of the network so that at least everyone can get SOME kind of DSL, and have a long term plan to migrate to fiber. The services side (telephone, ISP) would have to compete with everyone else (CLEC's) and pay the exact same amount for "colo" space at the CO.
Oracle, like Cisco and Sun, and other enterprise class companies, have support teams around the world that will work with you 24 hours a day until the problem is resolved. I don't buy the Redhat supported linux because: 1) I can support all of linux myself just fine, and 2) it is WAY overpriced for what you get when dealing with volume licenses. The volume license model is f*cked. It doesn't cost 73 times more to support 75 servers that are all clones of each other on identicle hardware than it does 1 (the price break is virtually nil.) Maybe 3 times, but not 73. I would much prefer to buy chunks of time or incidents than license per server, but they don't have that option. Windows is WAY less expensive. So I donate to and use Centos.
If Digium hardware products were as solid as Sangoma, you may have had a valid point. That is not the case however. Their existance also pushes Digium to improve, much like Firefox forces MS to improve IE. I hear what you are saying, which is why I encourage ABE, but I'll still buy Sangoma hardware. Sangoma's support is very good (better than digiums) on interface issues anyway. Paying Digium for ABE should be enough. If they won't support you for non-interface issues just because you don't have digium boards, then their paid product is worthless.
As another poster said, UPS. Yes, you will STILL probably have at least one analog phone anyway, more for faxing than anything else however (faxing and PBX's, especially soft PBX's, don't mix. In asterisk specifically, success rates vary widely.)
As for 911: In an emergency situation, you may not have someone in the office that knows what is going on, or even available. You want as much detail as possible going to emergency responders. If I have MY child in the school, I would sure as hell want this. Furthermore, the office may not even KNOW about a 911 situation. Could be a teacher alone in a classroom after hours who has a heart attack.
Large office buildings have the same issue, and it is even MANDATED in some places that detailed location information be provided. Since it costs close to nothing to do (other than time) why not do it?
I agree with going with ABE (Asterisk Business Edition) but I would get Sangoma cards. They work in more (all) systems. You can go analog for "common" extensions, and something like Polycom 601's for "office" extensions where you generally need more functionality. Add a sidecar for the "operator" stations. Most likely, he is not going to be buying equipement for this off of ebay, so good channel banks can get expensive. The cost of a good analog phone plus the cost of the channel bank is about the same as a polycom 301. The cost of cabling is the killer. Most schools I've seen have horrible wiring.
Polycom is my personal favorite, having tried a number of alternatives. The only thing I don't like is the lack of a backlight for the display, but it hasn't been a major issue in practice. Cisco is nice but be VERY aware of all the licensing issues - it makes them MUCH more expensive than polycoms. Avoid cheap SIP phones - they just suck and will give you nothing but trouble.
I agree with "don't use cheap hardware," but I would go with Sangoma cards. Digium cards are too finiky with the PCI bus. Sangoma has been doing network cards for MUCH longer, and "just work." Furthermore, they use Octastic echo cancellers which are top in the industry. Sangoma supports Asterisk Very Very well. I've used both, and won't buy digium T1 cards again - been burned too many times.
Second, he will use T1 lines (PRI) and not a "vonage" like VoIP service. No need for the analog line. One analog line and E911 won't work well for schools anyway. You need to be able to map the exact location in the building for 911. You do this by making sure you have a DID for each phone, and setting the callerID info correctly on outgoing calls. The telco provider will work with you to ensure that the 911 database has the location in the school of each DID number. That way, when the 911 dispacher gets your number, it shows not only the school address, but the exact location in the school of the caller. This system doesn't depend on the phone system at all. What you CAN do is set callerID to the generic main number for all calls EXCEPT 911 (where you send the full DID number) if you normally don't want extensions shown to callee's.
My read on this was that the airport was claiming that the free wifi was interfearing with LICENSED frequencies (non-wifi communications,) yet with no evidence. Of course they ALSO claimed that it was interfearing with their own paid wifi service - a claim without merit.
Windows is an operating system that any boner on the planet can install and use.
I dispute that.
Windows is easy to install if you have it pre-installed on a machine for you (oem install).
Give the average Joe on the street a windows install disk and they won't be able to do it because the drivers for his video card/ sound / network are not included with windows. Windows fails the "grandpa test" - something that Ubuntu is wonderful at.
On the other hand, trying to get an all-in-one printer/fax/scanner working on Linux will make even a seasoned kernel developer go catatonic (which isn't the fault of Linux, but rather the device manufacturer - but that's a whole other story...)
Let's put it this way. You will never see XP SP3, just like you will never see Win2Ksp5 despite the fact that most enterprise customers are still on 2000.
XP and Vista don't provide any compelling benefit over 2000 for most companies. Gamers are different a little different. The only thing that will force corporate customers to upgrade is lack of updates.
The 2.4ghz band, when used for unlicensed communications, is fully and internationally acknowledged as free and open for legal communications.
Subject to Part 15:
(b) Operation of an intentional, unintentional, or incidental radiator is subject to the conditions that no harmful interference is caused and that interference must be accepted that may be caused by the operation of an authorized radio station, by another intentional or unintentional radiator, by industrial, scientific and medical (ISM) equipment, or by an incidental radiator.
The airport was basically saying that the free wifi was interfearing with an "authorized radio station." The FCC said No, and they are the final authority on the subject. That is what this is all about.
1) Without critical mass, it doesn't. 2) No shit. 3) Ban all you want. Domains are cheap so spammers will create more...
End result - no change at all in spam volumes. If all the big ISP's got together and said that in January 1, 2008, they would no longer accept mail from anyone without an SPF record / SenderID, you MAY get 70% compliance. But I doubt it. In order to be truely effective, you need 90%+ compliance. Even at 100% compliance, you won't fix #3.
Slashdot Mirror
The number one killer of hard drives is heat. IBM did a study where for each 1 degree C over ambient, the MTBF dropped in half. If you want to make sure your drives last, make sure they are well cooled. Most modern drives have built-in temp sensors that you can query with various tools.
that Java on Debian will be as easy as "apt-get install java"
:-)
Close... More like: "apt-get install java java-doc java-common java-dev java-examples java-gnome libjava libjava-doc libjava-dev"
I'm convinced that Redhat will not hold this dominant position for much longer due to their insane pricing model.
I looked into 5 year costs for 70 servers and compared Windows and Redhat. Total cost of redhat was about 3 times more. Considering that my project really needed Linux, I went with Centos for nothing.
The problem RedHat has, is that the cost of supporting 70 servers with their OS is not 65 times the cost of one server (the price break is miniscule.) If they offered an alternate model of a site license with X number of incidents, I would have used them in an instant. The redhat model only works for companies with just a few servers and inexperienced staff.
Some other company with half a brain will take CentOS with a per-incident / per-hour pricing model and eat redhat alive. No, I don't think this company is Oracle.
The problem is that when it comes down to voting on important issues, it doesn't matter if you are a moderate republican or democrat - you are going to vote down party lines in *most* cases. If you don't, the party leadership comes to you and tells you that you are not going to get that $200M highway project your state needs, you won't be given a seat on a committee, etc. You have to play ball. That's what each and every politician has found who went in with a big set of ideals and promises of change.
To be honest, most AV software sucks anyway. It's bloated, slows the machine down, and doesn't handle zero-day viruses. AV software is great at closing the barn door after the cows already got out. Considering that so many attacks immediatly disable AV software, just what good is it?
AV software works fairly well on email servers to stop that vector, but a decent (non-windows) firewall and keeping up with patches is really all that is needed.
Yeah, but you didn't patch the damn thing and now it's infected with different crap that is just as bad as the old crap.
how do you guys get rid of these nasty rootkit and evolved spywares which can hide very well without reformatting
You don't. It is not worth the time and effort unless your personal / professional time has zero value. Get your data off and reinstall / restore from image.
Otherwise (if you are getting paid well for it) you can boot off a live CD or install the drive as a second in another system (one that has all the autorun crap disabled), Run AV/AS(pyware) on the drive, edit the registry removing all the startup items that you know isn't needed, run md5 comparisons on all the system files, and go from there. Dumping the registry and comparing with a known good registry is helpful at spotting crap.
It was probaby just a small handful of viruses (at most) that infected 7500 files. I would seriously doubt that the poster had 7500 DIFFERENT viruses. I can also believe that, at most, dozens of spyware programs were installed, but not hundreds. Spyware programs frequently claim that each cookie and registry entry is spyware.
Right now, the telco's own the lines and own the services. CLEC's can lease space at a huge price that does not allow true competition.
What is needed is to split the ILEC's into services, and physical infrastructure. Require the PHY telco's to build-out the rest of the network so that at least everyone can get SOME kind of DSL, and have a long term plan to migrate to fiber. The services side (telephone, ISP) would have to compete with everyone else (CLEC's) and pay the exact same amount for "colo" space at the CO.
It depends how you have it configured. Read the docs to understand it.
Oracle, like Cisco and Sun, and other enterprise class companies, have support teams around the world that will work with you 24 hours a day until the problem is resolved. I don't buy the Redhat supported linux because: 1) I can support all of linux myself just fine, and 2) it is WAY overpriced for what you get when dealing with volume licenses. The volume license model is f*cked. It doesn't cost 73 times more to support 75 servers that are all clones of each other on identicle hardware than it does 1 (the price break is virtually nil.) Maybe 3 times, but not 73. I would much prefer to buy chunks of time or incidents than license per server, but they don't have that option. Windows is WAY less expensive. So I donate to and use Centos.
I've read that Internet Explorer 7 has been decoupled from the shell
Yeah, but it still has hActive-X which is worse anyway.
If Digium hardware products were as solid as Sangoma, you may have had a valid point. That is not the case however. Their existance also pushes Digium to improve, much like Firefox forces MS to improve IE. I hear what you are saying, which is why I encourage ABE, but I'll still buy Sangoma hardware. Sangoma's support is very good (better than digiums) on interface issues anyway. Paying Digium for ABE should be enough. If they won't support you for non-interface issues just because you don't have digium boards, then their paid product is worthless.
FYI: there are better solutions. Nomachine / freenx for one. Looks like there is an Xrdp project too, which I haven't tried (but intend to.)
As another poster said, UPS. Yes, you will STILL probably have at least one analog phone anyway, more for faxing than anything else however (faxing and PBX's, especially soft PBX's, don't mix. In asterisk specifically, success rates vary widely.)
As for 911: In an emergency situation, you may not have someone in the office that knows what is going on, or even available. You want as much detail as possible going to emergency responders. If I have MY child in the school, I would sure as hell want this. Furthermore, the office may not even KNOW about a 911 situation. Could be a teacher alone in a classroom after hours who has a heart attack.
Large office buildings have the same issue, and it is even MANDATED in some places that detailed location information be provided. Since it costs close to nothing to do (other than time) why not do it?
I agree with going with ABE (Asterisk Business Edition) but I would get Sangoma cards. They work in more (all) systems. You can go analog for "common" extensions, and something like Polycom 601's for "office" extensions where you generally need more functionality. Add a sidecar for the "operator" stations. Most likely, he is not going to be buying equipement for this off of ebay, so good channel banks can get expensive. The cost of a good analog phone plus the cost of the channel bank is about the same as a polycom 301. The cost of cabling is the killer. Most schools I've seen have horrible wiring.
Polycom is my personal favorite, having tried a number of alternatives. The only thing I don't like is the lack of a backlight for the display, but it hasn't been a major issue in practice. Cisco is nice but be VERY aware of all the licensing issues - it makes them MUCH more expensive than polycoms. Avoid cheap SIP phones - they just suck and will give you nothing but trouble.
I agree with "don't use cheap hardware," but I would go with Sangoma cards. Digium cards are too finiky with the PCI bus. Sangoma has been doing network cards for MUCH longer, and "just work." Furthermore, they use Octastic echo cancellers which are top in the industry. Sangoma supports Asterisk Very Very well. I've used both, and won't buy digium T1 cards again - been burned too many times.
Second, he will use T1 lines (PRI) and not a "vonage" like VoIP service. No need for the analog line. One analog line and E911 won't work well for schools anyway. You need to be able to map the exact location in the building for 911. You do this by making sure you have a DID for each phone, and setting the callerID info correctly on outgoing calls. The telco provider will work with you to ensure that the 911 database has the location in the school of each DID number. That way, when the 911 dispacher gets your number, it shows not only the school address, but the exact location in the school of the caller. This system doesn't depend on the phone system at all. What you CAN do is set callerID to the generic main number for all calls EXCEPT 911 (where you send the full DID number) if you normally don't want extensions shown to callee's.
My read on this was that the airport was claiming that the free wifi was interfearing with LICENSED frequencies (non-wifi communications,) yet with no evidence. Of course they ALSO claimed that it was interfearing with their own paid wifi service - a claim without merit.
But whatever. The point is moot.
Microsoft is into full-on "all your base" mode now
You mean that at one point they wern't??? Had to be 20+ years ago...
Windows is an operating system that any boner on the planet can install and use.
I dispute that.
Windows is easy to install if you have it pre-installed on a machine for you (oem install).
Give the average Joe on the street a windows install disk and they won't be able to do it because the drivers for his video card/ sound / network are not included with windows. Windows fails the "grandpa test" - something that Ubuntu is wonderful at.
On the other hand, trying to get an all-in-one printer/fax/scanner working on Linux will make even a seasoned kernel developer go catatonic (which isn't the fault of Linux, but rather the device manufacturer - but that's a whole other story...)
WinXP with eyecandy disabled - Activation = Win2K for most practical purposes.
Let's put it this way. You will never see XP SP3, just like you will never see Win2Ksp5 despite the fact that most enterprise customers are still on 2000.
XP and Vista don't provide any compelling benefit over 2000 for most companies. Gamers are different a little different. The only thing that will force corporate customers to upgrade is lack of updates.
The 2.4ghz band, when used for unlicensed communications, is fully and internationally acknowledged as free and open for legal communications.
Subject to Part 15:
(b) Operation of an intentional, unintentional, or incidental radiator is subject to the conditions that no harmful interference is caused and that interference must be accepted that may be caused by the operation of an authorized radio station, by another intentional or unintentional radiator, by industrial, scientific and medical (ISM) equipment, or by an incidental radiator.
The airport was basically saying that the free wifi was interfearing with an "authorized radio station." The FCC said No, and they are the final authority on the subject. That is what this is all about.
1) Without critical mass, it doesn't.
2) No shit.
3) Ban all you want. Domains are cheap so spammers will create more...
End result - no change at all in spam volumes. If all the big ISP's got together and said that in January 1, 2008, they would no longer accept mail from anyone without an SPF record / SenderID, you MAY get 70% compliance. But I doubt it. In order to be truely effective, you need 90%+ compliance. Even at 100% compliance, you won't fix #3.
Spambayes, spamassassin - it doesn't matter. Both are CPU heavy. Netresult is the same. Use a whitelist to avoid the CPU intensive scanning.