Maybe, but what I would like to see is independent IMAP server, LDAP server, ICAL server, webmail application and calendar application from which I can pick one and install it inter-operating with our existing servers. When standard protocols are used, it should be no problem to operate a webcal application independently from the message store. Similar for the calendar.
you're paying for the Service Level Agreement (SLA). What that means is that if your circuit goes down, someone's head usually rolls. In other words, you get a reimbursement for your down time, or at least someone who tries to get you back running as soon as possible. As for your DSL/Cable, it really doesn't matter if you're God, you're down for as long as they feel like ignoring your problem.
In practice this means that all the time you are paying too much for your line, and when it goes down you get that money back. For one or two months. You will not recoup the money spent during all the months it does not go down. You will not get any compensation for lost business because the line was down. So, you are still responsible for arranging your own backup communication path in case the line is down!
With a consumer-grade DSL you get cheaper service, it is faster as long as it works, and when it does not work you use the same backup path that you needed anyway. As long as the DSL does not have unreasonably frequent disturbances, it overall is a much better deal.
We use two bundled consumer-grade DSL lines with ISDN dialup backup for emergency situations. This costs way less than a single business-grade line, is about 10 times faster during regular use, and over the past 5 years has not been down during business hours for any longer than a couple of minutes. And it happens maybe once a year (not counting short interruptions in the middle of the night because of network maintenance). In fact, when looking at the incident reports on the provider's website, there is no indication at all that business-grade users suffer less downtime than we do.
What I don't like about Zimbra is their "we take over your whole shop" approach. Of course, Exchange is not different but we have an existing installation with a working IMAP mailserver, a working web-based calendar (webCalendar http://www.math.utexas.edu/users/mzou/webCal/ ) and existing users using Mozilla and Squirrelmail. I looked into Zimbra at first to use their webmail system, but there seems to be no way of migrating one part of the system without converting everything at once.
In fact, with our current system we are just as locked-in as an Exchange/Outlook user is. About the only thing I could upgrade without lots of difficulty is the IMAP server (UW imapd), and even *that* is tricky because of the naming conventions of folders used by different servers.
The investigative group made a full disassembly of the original software. For a while it could be downloaded from their site. I think the software in itself was not so bad. It makes attempts to store the votes in such a way that votes are unlikely to be lost or changed, and cannot be linked to voters. However, there is no facility at all for securing that the software running on the machine is the untampered software compiled by the manufacturer. Given the age of the hardware, this is excusable. There was no trusted computing platform back then.
Sure, this could and should be improved. However, it is not very likely that a new generation of voting machines will be developed and deployed. The next generation of voting will most likely take place from home, via the Internet. (unnecessary to tell that this opens a whole new can of worms, but fortunately technology has advanced quite a bit in the past decade)
This probably was a Nedap machine similar to what we have in the Netherlands. A field of buttons with overlay on which the names of the candidates are printed, a two-line character-only LCD display that echoes the candidate name selected, and a larger button for confirmation. This is not the machine shown in the picture.
This indeed is a quite simple UI that does not cause problems here. What worries people is that there is a strict sequencing of voters: you present your ID, the clerk notes down your details and gives you the clearance to vote and sometimes even a sequence-ticket that you have to present to the person operating the machine. This at least gives the impression that the entire procedure is carefully recorded and at the end of day all votes can be matched to the voter's names.
In reality, the machine does not record the votes made in sequence, but a voter cannot verify that. Of course, when it would printout votes on a paper trail, it would be much easier to match up votes to voters. For anonymity, paper trails are more risky. They mainly improve the accountability.
There has been very similar discussion in the Netherlands. Here, too, the manufacturer said it was not a computer. An investigative group said "give us one, we will convert it to a chess-playing computer". Impossible, said the manufacturer, but denied them a demo machine. Then, they borrowed one from a municipality, and converted into a chess-playing computer. This, of course, lowered some jaws.
Furthermore, they wrote new firmware for it that manipulated the election results, and showed various different techniques for making sure this was not easily detected. The device widely used in the Netherlands has no precautions at all against manipulation of the firmware by unauthorized parties. The operating lock is a standard C&K lock for which almost all keys are the same. I remembered having such a lock in the junkbox and indeed, its key number is the same as on the voting machines.
But the flaw most easily exploited turned out to be around vote secrecy. The electronics are so badly shielded that someone with a radio receiver within a few tens of meters can detect what vote is being made.
After the usual initial denial, it has been taken up somewhat seriously by authorities. Operational procedures for guarding the firmware have been added (like sealing of the access lid to the electronics). Furthermore, a certain range of one type of machine and the entire series of another brand were declared unfit for use, because the emission problem could not be controlled by the manufacturer.
Not only with those reserved names, but with any extension as well. You cannot name your Word document AUX.DOC or NUL.DOC
The funny thing is that not all applications recognize (like Word) that it is unwise to save your file as NUL.whatever. They just save the file and make you looking for it later.
I am aware of the possibility of using the MAC address. But that is not possible when sitting at the machine to be booted. What I like is to power-on the machine, press F12 for network boot, and then press some key to tell WHAT to boot. When I boot one image, e.g. PXElinux or the 3com Pre-OS, it is not possible to switch to another one (e.g. via a menu) later. The menu would have to be presented by the BIOS. Well, I will to further research, maybe this really can be done within PXE, or a first-stage loader can be found that acts kind of like GRUB.
W.r.t. managing running systems (installing software and patches at night or during weekend): we have already solved that without network boot. But it looks like the enteo product is what I need.
Ok, I hoped that maybe some ready-made PXE image would exist that boots up a small fast-booting OS that can run simple scripts to perform tasks like partitioning, formatting, copying etc. BPbatch exactly fits the bill, but unfortunately I cannot get it to work.
Using Linux for this seems overly complicated, but it could be practical when at least a readily usable configuration would exist. I know about busybox and assembling small Linux systems, but for this purpose (merely getting new machines installed) it is too big a project.
Right now we have a working solution with the 3com stuff, which provides a selection menu and boots DOS from a selected diskette image file. IMHO, an omission in PXE is the lack of parameter passing before the image is loaded. As it is now, you can have only a single PXE image that is served to all the systems that attempt a network boot. You could select on some DHCP parameters in the DHCP server, but there is no parameter that you can control sitting at the keyboard of the booting system. Things would be much more usable when a simple menu was part of the boot sequence, and/or you could press a key which would be passed in the DHCP request.
Now, we can install the 3com image and have a menu that can only select between DOS diskettes, we can have the BPbatch image or we can have a PXElinux image, but there is no way to boot a PC and select between them. And the BPbatch and PXElinux images cannot be loaded from the 3com menu, or vice-versa. So we can install Windows or run a utility from DOS disk (system diagnostics, partition magic etc) but no way to install Linux, for example.
I will see what NetInstall has to offer, and at what cost. We are not a webcafe or education institute, so this whole thing is nothing like important business. Systems are normally installed only once. It is convenient, but difficult to justify a per-system license cost.
What free software are you using to host the "dd" that copies the network image to disk? I have tried using the free version of bpbatch in the past but I have never got it to work correctly. (it seems it loads itself somewhere in memory that later gets overwritten under certain circumstances. I tried to customize some of the examples provided to fdisk the HD and then copy something over from the network, but it always crashes with junk on the screen)
Maybe there are other free PXE "Pre-OS"es?
For now we use a 3com PXE image generator but it runs diskette images. This mostly limits you to use DOS, unnecessarily complicating the "dd" stuff.
I agree. I filter the majority of spam by just doing strict RFC compliance testing in the SMTP engine. It rejects almost everything sent via botnets. What comes through is mostly 419 scamming, because that is sent via bonafide mailservers. But that is easily filtered with SpamAssassin.
What do you mean no manual alarm could cope? How did manual alarms cope before the change?
I am not in the USA but my alarm clock has been following DST for nearly 20 years, actually now I have my second alarm clock that does so. It receives a longwave timestandard transmitter in Germany, and that transmits DST information. The clock neatly steps from 2:00 to 3:00 at the right moment, and when this changes the only people who have to worry are the guys operating the timestandard transmitter.
Those clocks are very inexpensive here. I have several in my house. At the moment of DST change I only need to change the clock in my microwave and my car.
5.0 and 1.5.0 are the same thing! Some marketroid at Sun thought the Java version numbering wasn't increasing as rapid as is customary in the market, and decided that what is internally numbered as 1.5.0 had to be externally marketed as 5.0 That is why the version jumped from 1.4 to 5.0
eps figures work for me, but only when printing to a Postscript printer. this is to be expected, as there is no postcript rendering engine anywhere in the chain except in the printer.
(this can be different in Linux, where there often is a "virtual" postscript printer that uses ghostscript to convert to the printer-specific format before sending it to the printer. but this depends on the spooler configuration which usually depends on the particular distribution and the printer installation wizard they use)
Maybe you should read up a bit on "roaming profiles", which are usually used in an enterprise environment with a Windows domain. Those store your user settings on the server and make them available on any client where you log in. So you don't have to setup your email account or wallpaper every time you use another computer, or re-install it.
Personal experience. Experiments done to see how viable it is to use Linux instead of Windows on our office desktops.
if you go for the memory hogs like OOo or FireFox
A typical Windows user that you want to convert to Linux will want to do the same things in a similar way as what he/she is used to in Windows.
Sure, I can give them a character mode Linux system with mutt, vi, groff and lynx and say "here you can see that you can do with 32MB what Windows needs 512MB for".
But that just isn't realistic. So what is left is that the Linux-counterparts of Windows applications are memory hogs on Linux, while the Windows apps aren't. You are not going to fix that by claiming otherwise or suggesting limited replacements that leave the user wanting things like Office compatability and ease of use. In that case you better install Windows or your users will be very unhappy.
XP is "fat-ass" and needs as much + more resources as Linux does.
This simply isn't true. A system used to run Linux with X, a desktop and some typical end-user applications (say Firefox, Thunderbird, OpenOffice) requires more resources than an XP system with IE, OutlookExpress, Office. Linux advocates like to claim the contrary but they base their claims on old information, limited environments (embedded sytems, limited GUI, crippled apps).
Unless you turn off that feature, of course. You can select if you want to show the login screen immediately, or only after pressing ctrl-alt-del (supposedly more secure, for example when someone could install an application on your system that just looks like a login screen and could use it to snoop passwords)
This often causes practical problems when you want to use two screens, and/or DVI. Even though the newest Intel onboard graphic chips actually support dual screen, the motherboards often have only a single video output connector. And it quite often still is VGA. For example, nearly all Dell systems come with an Intel onboard video chip and have only a single VGA output. When you opt for DVI or dual-monitor, you get an add-in videocard which usually is something like an ATI X300:-( (it used to be nVidia FX5200 in the past)
There are a few systems with DVI-I outputs and often they have wired the analog and digital outputs to two different screens, so a splitter cable enables you to connect two monitors (with different content). However, that is still one digital and one analog.
I wonder why it is so difficult to generate a service pack from all critical fixes on Windows Update every couple of months? Even when it is only called a "rollup package" it would be sooo much more convenient...
A fresh install of Windows XP SP2 now needs about 100 fixes to be installed, of which about 80 are critical security fixes. When you don't turn off system restore during the install it takes ages...
(I know, because we install systems from the network with unattend.txt rather than using imaging)
Slashdot Mirror
Maybe, but what I would like to see is independent IMAP server, LDAP server, ICAL server, webmail application and calendar application from which I can pick one and install it inter-operating with our existing servers.
When standard protocols are used, it should be no problem to operate a webcal application independently from the message store. Similar for the calendar.
you're paying for the Service Level Agreement (SLA). What that means is that if your circuit goes down, someone's head usually rolls. In other words, you get a reimbursement for your down time, or at least someone who tries to get you back running as soon as possible. As for your DSL/Cable, it really doesn't matter if you're God, you're down for as long as they feel like ignoring your problem.
In practice this means that all the time you are paying too much for your line, and when it goes down you get that money back. For one or two months.
You will not recoup the money spent during all the months it does not go down. You will not get any compensation for lost business because the line was down.
So, you are still responsible for arranging your own backup communication path in case the line is down!
With a consumer-grade DSL you get cheaper service, it is faster as long as it works, and when it does not work you use the same backup path that you needed anyway.
As long as the DSL does not have unreasonably frequent disturbances, it overall is a much better deal.
We use two bundled consumer-grade DSL lines with ISDN dialup backup for emergency situations. This costs way less than a single business-grade line, is about 10 times faster during regular use, and over the past 5 years has not been down during business hours for any longer than a couple of minutes. And it happens maybe once a year (not counting short interruptions in the middle of the night because of network maintenance).
In fact, when looking at the incident reports on the provider's website, there is no indication at all that business-grade users suffer less downtime than we do.
What I don't like about Zimbra is their "we take over your whole shop" approach. Of course, Exchange is not different but we have an existing installation with a working IMAP mailserver, a working web-based calendar (webCalendar http://www.math.utexas.edu/users/mzou/webCal/ ) and existing users using Mozilla and Squirrelmail.
I looked into Zimbra at first to use their webmail system, but there seems to be no way of migrating one part of the system without converting everything at once.
In fact, with our current system we are just as locked-in as an Exchange/Outlook user is. About the only thing I could upgrade without lots of difficulty is the IMAP server (UW imapd), and even *that* is tricky because of the naming conventions of folders used by different servers.
The investigative group made a full disassembly of the original software. For a while it could be downloaded from their site.
I think the software in itself was not so bad. It makes attempts to store the votes in such a way that votes are unlikely to be lost or changed, and cannot be linked to voters.
However, there is no facility at all for securing that the software running on the machine is the untampered software compiled by the manufacturer.
Given the age of the hardware, this is excusable. There was no trusted computing platform back then.
Sure, this could and should be improved. However, it is not very likely that a new generation of voting machines will be developed and deployed. The next generation of voting will most likely take place from home, via the Internet.
(unnecessary to tell that this opens a whole new can of worms, but fortunately technology has advanced quite a bit in the past decade)
This probably was a Nedap machine similar to what we have in the Netherlands. A field of buttons with overlay on which the names of the candidates are printed, a two-line character-only LCD display that echoes the candidate name selected, and a larger button for confirmation. This is not the machine shown in the picture.
This indeed is a quite simple UI that does not cause problems here. What worries people is that there is a strict sequencing of voters: you present your ID, the clerk notes down your details and gives you the clearance to vote and sometimes even a sequence-ticket that you have to present to the person operating the machine.
This at least gives the impression that the entire procedure is carefully recorded and at the end of day all votes can be matched to the voter's names.
In reality, the machine does not record the votes made in sequence, but a voter cannot verify that. Of course, when it would printout votes on a paper trail, it would be much easier to match up votes to voters. For anonymity, paper trails are more risky. They mainly improve the accountability.
There has been very similar discussion in the Netherlands.
i sh
Here, too, the manufacturer said it was not a computer. An investigative group said "give us one, we will convert it to a chess-playing computer". Impossible, said the manufacturer, but denied them a demo machine. Then, they borrowed one from a municipality, and converted into a chess-playing computer. This, of course, lowered some jaws.
Furthermore, they wrote new firmware for it that manipulated the election results, and showed various different techniques for making sure this was not easily detected.
The device widely used in the Netherlands has no precautions at all against manipulation of the firmware by unauthorized parties. The operating lock is a standard C&K lock for which almost all keys are the same. I remembered having such a lock in the junkbox and indeed, its key number is the same as on the voting machines.
But the flaw most easily exploited turned out to be around vote secrecy. The electronics are so badly shielded that someone with a radio receiver within a few tens of meters can detect what vote is being made.
After the usual initial denial, it has been taken up somewhat seriously by authorities. Operational procedures for guarding the firmware have been added (like sealing of the access lid to the electronics).
Furthermore, a certain range of one type of machine and the entire series of another brand were declared unfit for use, because the emission problem could not be controlled by the manufacturer.
http://www.wijvertrouwenstemcomputersniet.nl/Engl
Not only with those reserved names, but with any extension as well.
You cannot name your Word document AUX.DOC or NUL.DOC
The funny thing is that not all applications recognize (like Word) that it is unwise to save your file as NUL.whatever. They just save the file and make you looking for it later.
I am aware of the possibility of using the MAC address. But that is not possible when sitting at the machine to be booted.
What I like is to power-on the machine, press F12 for network boot, and then press some key to tell WHAT to boot.
When I boot one image, e.g. PXElinux or the 3com Pre-OS, it is not possible to switch to another one (e.g. via a menu) later.
The menu would have to be presented by the BIOS. Well, I will to further research, maybe this really can be done within PXE, or a first-stage loader can be found that acts kind of like GRUB.
W.r.t. managing running systems (installing software and patches at night or during weekend): we have already solved that without network boot.
But it looks like the enteo product is what I need.
Ok, I hoped that maybe some ready-made PXE image would exist that boots up a small fast-booting OS that can run simple scripts to perform tasks like partitioning, formatting, copying etc.
BPbatch exactly fits the bill, but unfortunately I cannot get it to work.
Using Linux for this seems overly complicated, but it could be practical when at least a readily usable configuration would exist. I know about busybox and assembling small Linux systems, but for this purpose (merely getting new machines installed) it is too big a project.
Right now we have a working solution with the 3com stuff, which provides a selection menu and boots DOS from a selected diskette image file.
IMHO, an omission in PXE is the lack of parameter passing before the image is loaded. As it is now, you can have only a single PXE image that is served to all the systems that attempt a network boot. You could select on some DHCP parameters in the DHCP server, but there is no parameter that you can control sitting at the keyboard of the booting system.
Things would be much more usable when a simple menu was part of the boot sequence, and/or you could press a key which would be passed in the DHCP request.
Now, we can install the 3com image and have a menu that can only select between DOS diskettes, we can have the BPbatch image or we can have a PXElinux image, but there is no way to boot a PC and select between them. And the BPbatch and PXElinux images cannot be loaded from the 3com menu, or vice-versa. So we can install Windows or run a utility from DOS disk (system diagnostics, partition magic etc) but no way to install Linux, for example.
I will see what NetInstall has to offer, and at what cost. We are not a webcafe or education institute, so this whole thing is nothing like important business. Systems are normally installed only once. It is convenient, but difficult to justify a per-system license cost.
What free software are you using to host the "dd" that copies the network image to disk?
I have tried using the free version of bpbatch in the past but I have never got it to work correctly.
(it seems it loads itself somewhere in memory that later gets overwritten under certain circumstances. I tried to customize some of the examples provided to fdisk the HD and then copy something over from the network, but it always crashes with junk on the screen)
Maybe there are other free PXE "Pre-OS"es?
For now we use a 3com PXE image generator but it runs diskette images. This mostly limits you to use DOS, unnecessarily complicating the "dd" stuff.
I agree. I filter the majority of spam by just doing strict RFC compliance testing in the SMTP engine. It rejects almost everything sent via botnets. What comes through is mostly 419 scamming, because that is sent via bonafide mailservers. But that is easily filtered with SpamAssassin.
What do you mean no manual alarm could cope? How did manual alarms cope before the change?
I am not in the USA but my alarm clock has been following DST for nearly 20 years, actually now I have my second alarm clock that does so.
It receives a longwave timestandard transmitter in Germany, and that transmits DST information. The clock neatly steps from 2:00 to 3:00 at the right moment, and when this changes the only people who have to worry are the guys operating the timestandard transmitter.
Those clocks are very inexpensive here. I have several in my house. At the moment of DST change I only need to change the clock in my microwave and my car.
You mean that this feature is only available to IE, and you cannot install a user-written appication that has the same privilege system?
.exe or systemcall from within the .exe, not by "magic mechanism"...
I smell an antitrust lawsuit... such a feature should be determined by ACL on the
It should also be noted that only badly managed systems, where the logged-in user has administrator privileges all the time, are really vulnerable.
5.0 and 1.5.0 are the same thing!
Some marketroid at Sun thought the Java version numbering wasn't increasing as rapid as is customary in the market, and decided that what is internally numbered as 1.5.0 had to be externally marketed as 5.0
That is why the version jumped from 1.4 to 5.0
eps figures work for me, but only when printing to a Postscript printer. this is to be expected, as there is no postcript rendering engine anywhere in the chain except in the printer.
(this can be different in Linux, where there often is a "virtual" postscript printer that uses ghostscript to convert to the printer-specific format before sending it to the printer. but this depends on the spooler configuration which usually depends on the particular distribution and the printer installation wizard they use)
Maybe you should read up a bit on "roaming profiles", which are usually used in an enterprise environment with a Windows domain.
Those store your user settings on the server and make them available on any client where you log in. So you don't have to setup your email account or wallpaper every time you use another computer, or re-install it.
Sorry, what are you sources ?
Personal experience. Experiments done to see how viable it is to use Linux instead of Windows on our office desktops.
if you go for the memory hogs like OOo or FireFox
A typical Windows user that you want to convert to Linux will want to do the same things in a similar way as what he/she is used to in Windows.
Sure, I can give them a character mode Linux system with mutt, vi, groff and lynx and say "here you can see that you can do with 32MB what Windows needs 512MB for".
But that just isn't realistic. So what is left is that the Linux-counterparts of Windows applications are memory hogs on Linux, while the Windows apps aren't.
You are not going to fix that by claiming otherwise or suggesting limited replacements that leave the user wanting things like Office compatability and ease of use. In that case you better install Windows or your users will be very unhappy.
At least that is what happens on my KDE login screen in Linux on my PC at work (which is using a domain with about 300 users for authentication).
XP is "fat-ass" and needs as much + more resources as Linux does.
This simply isn't true. A system used to run Linux with X, a desktop and some typical end-user applications (say Firefox, Thunderbird, OpenOffice) requires more resources than an XP system with IE, OutlookExpress, Office.
Linux advocates like to claim the contrary but they base their claims on old information, limited environments (embedded sytems, limited GUI, crippled apps).
Unless you turn off that feature, of course.
You can select if you want to show the login screen immediately, or only after pressing ctrl-alt-del (supposedly more secure, for example when someone could install an application on your system that just looks like a login screen and could use it to snoop passwords)
It is this kind of wording and this kind of claims that keeps so many hardware manufacturers from supporting Linux.
This often causes practical problems when you want to use two screens, and/or DVI. :-(
Even though the newest Intel onboard graphic chips actually support dual screen, the motherboards often have only a single video output connector. And it quite often still is VGA.
For example, nearly all Dell systems come with an Intel onboard video chip and have only a single VGA output. When you opt for DVI or dual-monitor, you get an add-in videocard which usually is something like an ATI X300
(it used to be nVidia FX5200 in the past)
There are a few systems with DVI-I outputs and often they have wired the analog and digital outputs to two different screens, so a splitter cable enables you to connect two monitors (with different content).
However, that is still one digital and one analog.
Install and configure TrustNoExe and your users cannot run programs they have downloaded, no matter if via webmail, internet, usb sticks, ...
I wonder why it is so difficult to generate a service pack from all critical fixes on Windows Update every couple of months?
Even when it is only called a "rollup package" it would be sooo much more convenient...
A fresh install of Windows XP SP2 now needs about 100 fixes to be installed, of which about 80 are critical security fixes.
When you don't turn off system restore during the install it takes ages...
(I know, because we install systems from the network with unattend.txt rather than using imaging)