1. Wake up at required time 2. Take care of personal hygiene 3. Be suitable clothed 4. Eat/drink as required 5. Leave house and follow Go to Work/School/Beach/Pub Process
A process like this is basically a checklist for what needs to be done or checked to accomplish any task. Then a process generally consists of procedures. One procedure for Morning Process pt 1. could be how to set the alarm clock, another for pt 2. can be "sniff your arm pit to determine of you need to wash today".
The general problem is that the Process is usually micromanaged. This is bad because it steals resources from all involved in the process. Cut processes, minimize them, empower employees.
As a side effect, it will be easier to get and maintain ISO-900x certification with a minimal processe philosophy.:)
>How many people have set up something reasonable complex and not >documented it.
I'm not saying that it shouldn't be documented. Documentation is important, even for a one server environment. But many processes are not. Process != Documentation. Documentations should include server inventory (hardware and software), dependencies and anything special (SLA for the server, non-standard configuration etc).
If you take the configuration database. This is one thing that can usually be 100% automated. A daily job that generates a list of filesets installed, compares it to yesterdays and checks differences. Gets hardware installed, checks network cards and IP addresses. Basically does the config/inventory for you. This can be a shell script, perl+mysql or CA Unicenter or Tivoli or whatever. No way this needs to be a three binder Process.
The problem with a Process is that is often assumes the user of that process is stupid. Too many work hours are spend on making complex processes that noone use. Sysadmin isn't rocket science, it's just infrastructure, computers and programs.
Yep. Simplified; on a mainframe batch jobs/programs are submitted to a job queue and the JCL describes the job. What execution class is should run in (execution class = priority, memory size available etc), what files or devices (datasets) it should read and write to (DD = Data Definition).
This is all under control of HASP/JES in different versions and various operating systems.
So basically: programmer compiles programs, submits JCL and program to JES3 and waits for it to run and get output. First hurdle is to get pass the "JCL Error" step:)
o Change control? o Incident management? o Problem management? o Change window? o Service level negotiations? o Capacity management? o Security management?
All of these points are needed for sysadmin, but you don't really need a process for them. use common sense and knowledge of the big picture to manage systems. Real good sysadmins doesn't need processes for how to insert a floppy or how to eject a tape from a DLT drive.
Processes are for McDonalds employees (remember section A.6.2 and say "Do you want fries with that?" if customer has only ordered a burger).
A good mentoring system with experienced sysadmins is what you need. Then the IT systems/infrastructure can be blackboxed from a management perspective.
What a collection of self-evident drivel... Basically he's saying that overclocking can make your system unstable, you should check your BIOS version, driver version and not run too much beta stuff. And maybe overclocking can increase performance, unless it makes your computer crash....
> To refresh your memory: Back before we had PDA's, we called such software PIM's.
Hehe, yes... Personal Information Managers. Though at the early start SideKick said it all. T'was the first widespread TSR app, made by Frank Borland himself from his little cabin up in the Santa Cruz mountains. Phillipe Kahn went to Starfish, last SideKick release was in 1999...
If there's one thing DOS isn't, it's Obese... The DOS kernel (IO.SYS and MSDOS.SYS) could even be called a microkernel (apart from the fact that DOS isn't really an OS).
Still running PC-DOS 2000 (I've actually never run MS-DOS, PC-DOS was what came with my PC, and later on PC-DOS had more and better utilities) here, needed for some games like Privateer2 and other VCPI games.
My first PC-DOS version was 2.0. It supported fixed disks (harddisks) and directories. Also it was the first version that supported file handles (stdout, in and err, handle 0, 1 and 2 respectively). Saw another one here claim that DOS didn't support stderr, but that's wrong. Before 2.0 DOS used FCBs (like CP/M) to open files. DOS internals really showed off it's CP/M heritage.
What was good about DOS was that you had 100% hardware control... what was bad about DOS was that you had 100% hardware control. By many definitions, it wasn't even an OS, as it didn't do everything an OS should do. But it was a single-user system from the start, and as such it was good enough, with low overhead -- important in an age where 16K was the entry PC memory size... and 64K was a lot.
That reminds me... 2.0 also was the first version that supported 180K/360K floppies, with the new support for 9 sectors pr track (up from 8).
Lotus 123, MultiMate word proc (actually, and OEM version kalled WriteIT; and integrated package with CalcIT, KeepIT and several other apps), TurboPascal compiler, SideKick "PDA", Norton Utilities... later Norton Commander (still one of the best file managers).
> The french TGV (the fastest - 515 km/h that's 320 miles per hour) > is a souped-up ordinary train.
Pretty "ordinary" yes... but there's one striking feature that differs from ordinary trains:
On a normal carriage, you have two boogie wheel pairs, one on each end of the carriage. On the TGV two carriages shares the same boogie in the intersection. Picture here: TGV boogie
This picture is actually from a tilting prototype of the TGV.
You can read more about the modifications to the TGV (Train Grande Vitesse) here, and some history here.
make menuconfig is an easy to use console app. I dread the day when we must use X to configure a kernel.
One of the most important aspects of Linux (esp. as a server) is that you can configure anything though a telnet/ssh session and vi (or whatever editor you like).
Try to use X to reconfigure a server in Ulan Bator.... good luck.
But reading on about lkc... it's an extensible system where you could write your own interface on top of the framework and methology of lkc, so you could have one spiffy gui and one down to basics text mode kernel config tool.
Just don't underestimate the value of running a server though ssh alone.
> It's odd that RH, SuSe and Mandrake compete with > each other more than their common enemy.
I don't think so. I think this is a good thing, because it will give us better Linux distros. And better Linux distros will someday jam Linux into Microsofts monopolistic gears.
The day every hardware gadget and game in stores comes with a anonymous note on the back that says "Supports Windows YX and Linux", the goal is reached.
Through competition (and only competition) will Linux improve to that point.
True, and for most low end stuff USB (esp. 2.0) works very well.
But isn't firewire still better for media streams due to built in sound and picture sync? Maybe FireWire is still a cheaper option for digital video cameras as the sync has to be implemented on top of an USB link. Unless USB2.0 can sync "good enough"...
Anyone seen USB2.0 video cameras and the quality of the transfer?
I would prefer a single protocol, cable and electric, maybe with optional stuff on top, much like TCP/IP does it -- or SCSI. Then we would have standard connectors for all gadgets, just different interface chips if the gadged needed high-speed, datastream sync, high latency or whatever.
They we will still be able to connect that new camera to the new PC we buys in 2014...
>Objects are perceived as the same distance away >when light takes the same amount of time to >traverse from each of the objects.
BZZT! Wrong. While the brain is a fantastic piece of biology, variation in lightspeed from objects that surrounds you in a room or outside are way too small for you to register.
The brain calculates differences in angles from our two eyes to find out how far away an object is, as well as references to other objects for far away objects.
Stare at an object and open and close your right then left eye. You will notice that objects shift from left to right as you do this. Objects closer to your eyes shifts relatively more than objects further away. For objects further away, a linear approched is also used, as in object B is behind A, therefore B is further away.
Of course you realize that todays technology cannot provide what you want, and that's probably your point:)
There's just no way that we can store enough energy (that can be tapped as electricity)... not to mention recharge time.
The Think can store 11.5kWh, 100 amps. There's no way you can get that kind of energy out of solar panels in any reasonable amount of time. And a measly 11.5kWh is not enough, a few MWh would help. If anyone can find a way to store that amount of energy, a Nobel prize wouldn't be out of the question.
The battery tech we are using today is practically the same as it has been for the last 50 years. Sure, we have some new materials (like LiIon), but for a high capacity, high output (as a car battery would be)... nothing new.
GigaFahrad caps that can are stable and can deliver lots of amps over a long period... or still SF stuff like microfusion, cold fusion, taps into other dimensions...
We'll see how the fuel cells turns out. Would make good hybrid alternatives. Problem today with low emission hybrid cars is that they are more expensive to make (you have two engines and battery arrays). That will be a non issue with higher gasoline prices (I.e $10 a gallon) as fuel gets sparser.
Re:Another article stolen from Kuro5hin.
on
Electric Armor
·
· Score: 1
> HEAT is a shaped charge, it has a 2 kilogram > warhead that fires its explosive in jet stream > directly in front of the round.
HEAT rounds can be larger than 2KG. TOW, Hellfire, Panserfaust etc all have larger warheads. But you might have been thinking of RPG-7 and M-72, that both have warheads in the 2KG range... with the actual explosive making up less than 0.5KG.
One thing that I haven't seen discussed here is how fast the electrical "shield" can recharge. I doubt that it can recharge fast enough (or store enough juice in teh caps) for a dual warhead HEAT round. It could even have problems with multiple simultaneous fire, i.e in an ambush with 3-5 soldiers firing RPG-7, M-72 or M-136 RFC. That's the only way that I would even consider taking out a M1 with light AA-rockets (from behind, above (i.e in a city) in a coordinated attack.
I wonder why APFSDS type ammo is not available in smaller RFC style weapons.....
In the medical area, artificial eyes and hearing implants has made gigant leaps forward in just the last couple of years. From being a SciFi fantasy, we can now project images directly into the optic nerver. Sure it's very low quality, black&white and fuzzy and poor contrast and out of focus and generally lousy.... but that's just an engineering task:)
Imagine having no displays at all, just sound and video overlayed directly into neural interface chips!
This might be a reality 15-30 years down the road...
>As a software tester, I take your assertion that testers are monkeys as an offence.
It's more like "infinity number of monkeys, with infinity number of PC's..." etc...
With the quality level of todays software, we could need them... the monkeys that is.
And come on, I'm not saying that testers are monkeys. Testers are not the sole reasons for bugs in software. Design, programmers, architecture... *deadlines* (otoh, without deadlines programs would never reach version 1.0)
Brooks' Law: "Adding manpower to a late software project makes it later."
This can be simplified:
"Adding manpower to a software project makes it later."
There's rarely that many programmers needed for a given task anyway. You need a project leader and lots of monkeys to test it... very few projects should have more than 10 programmers (if any).
Disabling software that is in production like this is a big no-no. What about a web site that goes down (maybe causing a big player many thousands of $$$ or more) would effectively remove Open Source as an option for business use.
A syslog entry, a kernel message, a mail sent would all be ok. I.e like "This software is 12 months old, there are probably a newer version out with more features, bug fixes and plugged security holes out. Please check at http://..."
But no matter how you look at it, it is the system admins duty to maintain a system. When new updates comes out, they should be put into the development->test->pilot->acceptance test->production cyclus commont for RealProduction(tm) systems.
I agree with you. It is a dated paragraf that distributed.net should have removed. It was a valid argument back in the days when 42-bit keys was the maximum allowed for exported systems.
And 42 bits are clearly too veak. Today when 128-bits ar common and allowed to be used by almost the entire world, it's not an issue anymore.
Slashdot Mirror
I know that.
:)
Example:
Morning Process
1. Wake up at required time
2. Take care of personal hygiene
3. Be suitable clothed
4. Eat/drink as required
5. Leave house and follow Go to Work/School/Beach/Pub Process
A process like this is basically a checklist for what needs to be done or checked to accomplish any task. Then a process generally consists of procedures. One procedure for Morning Process pt 1. could be how to set the alarm clock, another for pt 2. can be "sniff your arm pit to determine of you need to wash today".
The general problem is that the Process is usually micromanaged. This is bad because it steals resources from all involved in the process. Cut processes, minimize them, empower employees.
As a side effect, it will be easier to get and maintain ISO-900x certification with a minimal processe philosophy.
>How many people have set up something reasonable complex and not
>documented it.
I'm not saying that it shouldn't be documented. Documentation is important, even for a one server environment. But many processes are not. Process != Documentation. Documentations should include server inventory (hardware and software), dependencies and anything special (SLA for the server, non-standard configuration etc).
If you take the configuration database. This is one thing that can usually be 100% automated. A daily job that generates a list of filesets installed, compares it to yesterdays and checks differences. Gets hardware installed, checks network cards and IP addresses. Basically does the config/inventory for you. This can be a shell script, perl+mysql or CA Unicenter or Tivoli or whatever. No way this needs to be a three binder Process.
The problem with a Process is that is often assumes the user of that process is stupid. Too many work hours are spend on making complex processes that noone use. Sysadmin isn't rocket science, it's just infrastructure, computers and programs.
Yep. Simplified; on a mainframe batch jobs/programs are submitted to a job queue and the JCL describes the job. What execution class is should run in (execution class = priority, memory size available etc), what files or devices (datasets) it should read and write to (DD = Data Definition).
:)
This is all under control of HASP/JES in different versions and various operating systems.
So basically: programmer compiles programs, submits JCL and program to JES3 and waits for it to run and get output. First hurdle is to get pass the "JCL Error" step
hehe, I bet most ppl here wouldn't know a JCL if it jumped up and bit them in the... nose. :)
;)
But DD != dd
Processes?
o Change control?
o Incident management?
o Problem management?
o Change window?
o Service level negotiations?
o Capacity management?
o Security management?
All of these points are needed for sysadmin, but you don't really need a process for them. use common sense and knowledge of the big picture to manage systems. Real good sysadmins doesn't need processes for how to insert a floppy or how to eject a tape from a DLT drive.
Processes are for McDonalds employees (remember section A.6.2 and say "Do you want fries with that?" if customer has only ordered a burger).
A good mentoring system with experienced sysadmins is what you need. Then the IT systems/infrastructure can be blackboxed from a management perspective.
What a collection of self-evident drivel... Basically he's saying that overclocking can make your system unstable, you should check your BIOS version, driver version and not run too much beta stuff. And maybe overclocking can increase performance, unless it makes your computer crash....
Yeah... helpful.
> To refresh your memory: Back before we had PDA's, we called such software PIM's.
Hehe, yes... Personal Information Managers. Though at the early start SideKick said it all. T'was the first widespread TSR app, made by Frank Borland himself from his little cabin up in the Santa Cruz mountains. Phillipe Kahn went to Starfish, last SideKick release was in 1999...
> DOS -- Dumb Obese System
If there's one thing DOS isn't, it's Obese... The DOS kernel (IO.SYS and MSDOS.SYS) could even be called a microkernel (apart from the fact that DOS isn't really an OS).
Still running PC-DOS 2000 (I've actually never run MS-DOS, PC-DOS was what came with my PC, and later on PC-DOS had more and better utilities) here, needed for some games like Privateer2 and other VCPI games.
My first PC-DOS version was 2.0. It supported fixed disks (harddisks) and directories. Also it was the first version that supported file handles (stdout, in and err, handle 0, 1 and 2 respectively). Saw another one here claim that DOS didn't support stderr, but that's wrong. Before 2.0 DOS used FCBs (like CP/M) to open files. DOS internals really showed off it's CP/M heritage.
What was good about DOS was that you had 100% hardware control... what was bad about DOS was that you had 100% hardware control. By many definitions, it wasn't even an OS, as it didn't do everything an OS should do. But it was a single-user system from the start, and as such it was good enough, with low overhead -- important in an age where 16K was the entry PC memory size... and 64K was a lot.
That reminds me... 2.0 also was the first version that supported 180K/360K floppies, with the new support for 9 sectors pr track (up from 8).
Lotus 123, MultiMate word proc (actually, and OEM version kalled WriteIT; and integrated package with CalcIT, KeepIT and several other apps), TurboPascal compiler, SideKick "PDA", Norton Utilities... later Norton Commander (still one of the best file managers).
Ah, the memories....
> The french TGV (the fastest - 515 km/h that's 320 miles per hour)
> is a souped-up ordinary train.
Pretty "ordinary" yes... but there's one striking feature that differs from ordinary trains:
On a normal carriage, you have two boogie wheel pairs, one on each end of the carriage. On the TGV two carriages shares the same boogie in the intersection. Picture here: TGV boogie
This picture is actually from a tilting prototype of the TGV.
You can read more about the modifications to the TGV (Train Grande Vitesse) here, and some history here.
make menuconfig is an easy to use console app. I dread the day when we must use X to configure a kernel.
One of the most important aspects of Linux (esp. as a server) is that you can configure anything though a telnet/ssh session and vi (or whatever editor you like).
Try to use X to reconfigure a server in Ulan Bator.... good luck.
But reading on about lkc... it's an extensible system where you could write your own interface on top of the framework and methology of lkc, so you could have one spiffy gui and one down to basics text mode kernel config tool.
Just don't underestimate the value of running a server though ssh alone.
> It's odd that RH, SuSe and Mandrake compete with
:)
> each other more than their common enemy.
I don't think so. I think this is a good thing, because it will give us better Linux distros. And better Linux distros will someday jam Linux into Microsofts monopolistic gears.
The day every hardware gadget and game in stores comes with a anonymous note on the back that says "Supports Windows YX and Linux", the goal is reached.
Through competition (and only competition) will Linux improve to that point.
IMHO and YMMV
And as a meaningless nitpick:
OS/2 supports double drive letters, so there you can have AA:...
Maybe that's where this DOS/Windows double letter teories comes from?
>I think someone should come out with a standard called WhiteJaw
No no... not WhiteJaw. It's a bit racist to imply that only white people can use it. And Jaws was a terrifying movie that will scare the kids.
BlueTooth is a much more PC name...
> but USB 2.0 is still cheaper to implement.
True, and for most low end stuff USB (esp. 2.0) works very well.
But isn't firewire still better for media streams due to built in sound and picture sync? Maybe FireWire is still a cheaper option for digital video cameras as the sync has to be implemented on top of an USB link. Unless USB2.0 can sync "good enough"...
Anyone seen USB2.0 video cameras and the quality of the transfer?
I would prefer a single protocol, cable and electric, maybe with optional stuff on top, much like TCP/IP does it -- or SCSI. Then we would have standard connectors for all gadgets, just different interface chips if the gadged needed high-speed, datastream sync, high latency or whatever.
They we will still be able to connect that new camera to the new PC we buys in 2014...
hah! dreamer....
>Objects are perceived as the same distance away
>when light takes the same amount of time to
>traverse from each of the objects.
BZZT! Wrong. While the brain is a fantastic piece of biology, variation in lightspeed from objects that surrounds you in a room or outside are way too small for you to register.
The brain calculates differences in angles from our two eyes to find out how far away an object is, as well as references to other objects for far away objects.
Stare at an object and open and close your right then left eye. You will notice that objects shift from left to right as you do this. Objects closer to your eyes shifts relatively more than objects further away. For objects further away, a linear approched is also used, as in object B is behind A, therefore B is further away.
Of course you realize that todays technology cannot provide what you want, and that's probably your point :)
There's just no way that we can store enough energy (that can be tapped as electricity)... not to mention recharge time.
The Think can store 11.5kWh, 100 amps. There's no way you can get that kind of energy out of solar panels in any reasonable amount of time. And a measly 11.5kWh is not enough, a few MWh would help. If anyone can find a way to store that amount of energy, a Nobel prize wouldn't be out of the question.
The battery tech we are using today is practically the same as it has been for the last 50 years. Sure, we have some new materials (like LiIon), but for a high capacity, high output (as a car battery would be)... nothing new.
GigaFahrad caps that can are stable and can deliver lots of amps over a long period... or still SF stuff like microfusion, cold fusion, taps into other dimensions...
We'll see how the fuel cells turns out. Would make good hybrid alternatives. Problem today with low emission hybrid cars is that they are more expensive to make (you have two engines and battery arrays). That will be a non issue with higher gasoline prices (I.e $10 a gallon) as fuel gets sparser.
> HEAT is a shaped charge, it has a 2 kilogram
> warhead that fires its explosive in jet stream
> directly in front of the round.
HEAT rounds can be larger than 2KG. TOW, Hellfire, Panserfaust etc all have larger warheads. But you might have been thinking of RPG-7 and M-72, that both have warheads in the 2KG range... with the actual explosive making up less than 0.5KG.
One thing that I haven't seen discussed here is how fast the electrical "shield" can recharge. I doubt that it can recharge fast enough (or store enough juice in teh caps) for a dual warhead HEAT round. It could even have problems with multiple simultaneous fire, i.e in an ambush with 3-5 soldiers firing RPG-7, M-72 or M-136 RFC. That's the only way that I would even consider taking out a M1 with light AA-rockets (from behind, above (i.e in a city) in a coordinated attack.
I wonder why APFSDS type ammo is not available in smaller RFC style weapons.....
In the medical area, artificial eyes and hearing implants has made gigant leaps forward in just the last couple of years. From being a SciFi fantasy, we can now project images directly into the optic nerver. Sure it's very low quality, black&white and fuzzy and poor contrast and out of focus and generally lousy.... but that's just an engineering task :)
Imagine having no displays at all, just sound and video overlayed directly into neural interface chips!
This might be a reality 15-30 years down the road...
>your simplification is somewhat faulty
Of course it is! It is a simplification.
:)
>As a software tester, I take your assertion that testers are monkeys as an offence.
It's more like "infinity number of monkeys, with infinity number of PC's..." etc...
With the quality level of todays software, we could need them... the monkeys that is.
And come on, I'm not saying that testers are monkeys. Testers are not the sole reasons for bugs in software. Design, programmers, architecture... *deadlines* (otoh, without deadlines programs would never reach version 1.0)
Uhm.... yes... and I even did a preview.
:)
Maybe there is a deeper truth in it. Maybe many software projects would be better off without programmers?
This can be simplified: "Adding manpower to a software project makes it later."
There's rarely that many programmers needed for a given task anyway. You need a project leader and lots of monkeys to test it... very few projects should have more than 10 programmers (if any).
Baaad Idea!
Disabling software that is in production like this is a big no-no. What about a web site that goes down (maybe causing a big player many thousands of $$$ or more) would effectively remove Open Source as an option for business use.
A syslog entry, a kernel message, a mail sent would all be ok. I.e like "This software is 12 months old, there are probably a newer version out with more features, bug fixes and plugged security holes out. Please check at http://..."
But no matter how you look at it, it is the system admins duty to maintain a system. When new updates comes out, they should be put into the development->test->pilot->acceptance test->production cyclus commont for RealProduction(tm) systems.
There is no way you can automate this!
I agree with you. It is a dated paragraf that distributed.net should have removed. It was a valid argument back in the days when 42-bit keys was the maximum allowed for exported systems.
And 42 bits are clearly too veak. Today when 128-bits ar common and allowed to be used by almost the entire world, it's not an issue anymore.