Actualy BGP can pick a path based upon bandwith but the first and foremost descision is AS path length and prefix length. You can inject other metrics into BGP as well say local link utilization at the peering point. In general it's a BAD idea push trafic to longer distant peers rather than say get more peering bandwith. And BTW hop count never comes into play ok AS path length BGP dosent know or care about how many intervieing routers are there just how many AS's. Some providers like AbiveNet work in the reverse and try and keep the data on there network for as long as possible the easiest way to do this is to summerize with distance so local/24's are advertised say for NYC to the NYC peers but agrigated up into a/19 in Cali More specific routers are allways prefered.
OK With all that said traffic engineering with BGP can be fun and it's allways a question of how much memory the routers have (BGP tables are BIG and router Memory is was to small think 32 megs ish for the current 110k line BGP tables (and Telstra is responcible for most of those the whole internet would thank them if they thaught of agrigating there routes and buying fargin bigger pipes instead of making the rest of the world backhaul for them)
Most of the cost of a T1 is the line not the bandwith. Normal numbers for bandwith in bulk are down to sub hundreds a month for quality bandwith providers if baught in bulk and properly negotiated. And i'm not talking about cheap bandwith I'm talking nice tier one pretty much everybody but UUNet and AT&T (they are never cheap but still in the low hundred). Oh BTW before you flame me I currently in proccess of quoting a few hundred megs a sec from all the major carriers I can find these are the numbers after 2-3 rounds of negotiations. Granted you still need to account for leased lines from those handoffs throughout your backbone costs of routers etc. T1's are all leased line cost because they are a PITA on real copper or have a lot of up front costs converting them and throwing them on fiber. A DS3 circut that is 30 times faster generaly only runs 5-10 times the cost.
As somebody who was shipped of to work at IBM while in HS by the HS I would have to disagree. Starting early is a great idea especialy in intern and other low level jobs that you need to build up a resume. It can also allow people to WORK through school and not start out life so far back in debt. The partys and socalizing of school may be fun and good networking but if mommy and daddy cant pay for it and your not a jock or a perfect student then I would say taking a 2nd shift help desk position of junior admin is a great way to get a leg up on your beer guzzaling friends (you can do that on the weekend like an adult er 20 something:)
I can see this working very well when combined with a good local after school internship program. It can allow high school students to find a career that they enjoy earn enough money in HS and later to put themselves through school. Yes this is a tough way to do things but I'll teel you this I'll hire a hot teenage geek that loves to do this over some college kid that isn't sure what they want to do with there life.
BTW dont tell me kids dont know what to do go back a few hundred years and people had families by age 18 and a career this is just our society playing one up on itself every generation.
My problem with that analogy (SP?) is the internet is not your house and neither is your internet connection. Houses dont do things by themselves on the most part. Houses dont make money in general. This is a technical problem and thus needs a technical solution. Servers need to come with sencable defaults. But the biggest thing is people need to turn anything off they dont use. There is a difference between leaving your door open in a nice neighborhood is one thing the internet is the worst of the crack neighborghoods at best treat it that way.
It's it more important that MS SQL server shouldent be exposed to the internet directly in the first place. There are no public SQl servers than I can think of and no reason for them besides maybe some open testing and compatability public labs. Port filtering isn't a panecea but it's the second line of defence (after egress filtering by everybody) there is no reason that a SSH port forward or a VPN cant be used hell a GRE tunnel with no encryption instead of having it open and on the internet. This is also the case for many other packages how many MySQL ports I have seen open it's disgusting.
Well apparently you have figured out that a good DOS attack can flood the box with packets effectivly flooding out the valid incomming packets you need something like local QOS to fix this and it has to be set up correctly.
Next step is the switch in front again dependant on the type of switch it may have problems with a DOS attack as far as management goes.
The router above That may hav it's wan links flooded effectivly edging out valid traffic again and dependant on the router type taking up enormous ammounts of CPU time.
Then there are all the upstreams. When you buy bandwith from some local provider like they have they dont have the capacity to deal with the attack nessicarily and at minimum it may be degrading there network.
The only good solution to fix these things is source address verification from ALL ISP's (ok not happening any time soon) and oh that breaks mobil IP and some multi homes sites even the cheasy multiple DSL and cable modem setups for more bandwith and reliability configs.
This is why real VPN software includes a firewall that it's configured by firewall admin not the end user. Granted there are ways around this but they should be monitored and offences that can lead to terminiation.
That depends on what sort of routers they are using. A lot of Cisco gear an others have overhead running ACL's (first Cisco that can do compiled is a 7200 I think) so you want to accept/deny your most common traffic first so your router dosent die when your throughput goes up.
More likly in my experience firewalls get configured to allow al trafic out not just traffic that it will allow responces to (simple port based firewalls with no connection trackings realy hardly considered firewalls anymore) a realy simple port filter might allow all incomming tcp besides initial, dissalow all udp except to servers and ports they know about (Primary DNS boxes, NTP any other UDP service) This is very common setup with older gear especialy with publicaly addressed boxes.
This is why you should never buy works with any form of copy protection on the media. You know what I dont care about the media the media is extranious. Good copy protects scheems do just that make sure only I can use it and that 3 billion people dont copy my copy. Now I dont care if they force me to register via phoe if I'm not net connected it's a reasonable assumption if you can afford a computer you can afford to call a 800 number. No software shouldent tattle but it's resonable for software to imprint itself on the machine somewhat as long as the company allows me to reinstall / migrate software at whim within reason. But realy good software will just avoid copy protection it's bad for there users and says they dont trust them to be responcible because for a game unlike what they thing it will be cracked it's just a question of time.
Well the bloat thing seems to mostly be PS programmers making new high level abstratic GARBAGE that makes the coding easier but generaly less efficient. Granted this is from somebody that dosent have a interactive desktop with less than a gig of main ram generaly because it's cheap and I have some ugly bloatware apps that I use. At the same time I have Linux boxes running on as little as 16 megs or ram and thats not to hard to do (RH dosent like less than32 for an install now though) 32 makes a good firewall or MP3 Player etc etc etc.
I think you would have a difficult time writing 14 channels at once mostly you have buss contention on the PCI side or whatever is connecitng you northbridge to your south bridge remember I2O isn't running on something like this the CPU has to see probably all the data unless they have zero copy buffers for disk data now (not sure I know network wise that was one of the big TUX things) IDE Raid will help with some of this especialy if it has some buffer on it. I would go with 2 drives on 2 channels aka a drive a channel if you want any sort of performance while running IDE still has or untill recently had issues with bus contention and switching from master to slave sorta like OS contex switching.
SCSI or FC would be a better solution dependant on your budget. Old school SGI gear can write SDI data to disk for 4 tracks cheaply enough and you could encode to Divx / whatever your choice is via a HSM like proccess so if your watching it live and using pause were talking nearly Zero loss (those A/D converters for SDI are realy realy high quality as a tab bit expensive) As I think 4 channels was the best the workstation class boxes from a few years back could handel it would work out well as the ingress / egress boxes (go with clustered disk and CXFS and you have a REALY scalable solution just dangle off more encoder / decoder heads) granted this may be a lot more that somebody wants to do for a home project but recording 16 channels of vid in real time sounds like something more than home use.
Here Here lets second that the Newton was the first inovating PDA. The orgigional newton read my HORID hardwritting better than more modern designs in other PDA's or relearing how to write with grafiti. The line drawing abilites were great as well it was easy to draw up a quick deisgn note.
Now thing of what that could do in an age with 512 Meg flash cards running pretty cheap and the general avalibility of wireless (Nextels flat rate service for 60 a month with good coverage per bit is a horid idea for a consumer great for providers though) Roll that up into something that dosent need to fit in my shirt pocket but pretty much needs a screen, a way to get a lot of memory into it, good rechargable battery life and a good OS (maybe Palm emulation if the hardware can deal and old school newton emu as well) I would pay 500 or so for this without the big memory card etc. If they can get it to sub 300 it makes mass market.
Think about it the ability to use Palm and Newton Apps + new apps for the platform. I dont realy need an orginizer I need a portable web browser / pdf reader / email / Terminal emu (OK I'm a network Engineer the serial port on my Newton keeps it in my tool box to this day along with an ugly 8pin to rj45 adapter dongel I made)
There is one problem with this at least Mobile IP it's a spec it's designed to work and it dosent provide a return path like a VPN. Now granted most people would be better served by a VPN than using mobile IP. Next is ou have to remember there are a LOT of things that assume packets can originate from any address.
Geo Load ballancers, especialy the ones that use DNS timed race now this is somehting with a fixed Ip address but it's hard to keep up all sorts of exceptions. BTW this works pretty well.
Dual headed users (say DSL and Cable) that LB there outgoing requests. Now granted this is minor and isn't supposed to work but it's nice and gets you a good deal of redundancy.
Multihomed servers, they can send packets from any of there interface IP's via any of it's interface IP's some mail servers and DNS servers do this alot. Remember GigE is a new thing and servers have seen 100bt as a limitation for awhile.
And there are a lot of other issues. Now I have setup IPP's to do this and it works well enough at say a up to a tier 3 provider. This will give every multihomed carrier problems with traffic management especialy ones with old routers not taking more than a BGP default, not having the same carriers at every router, running disconnected AS's (Ok this is a no no but they do it)
Basic egress filtering should be in place for everybody all the martians should be blocked perferably at the ingress routers doing DSL, Cable and Modem agrigation. Yea this has some maintnence associated with it as the martians list changes every now and again as new/8 are given out. It's easier to tell an aggrigation router to only allow packets say from the/32 of a modem user than make all the edge routers filter things.
Yea 5 9's can be tough to get the execs to pay for but it's a large part of my engineering life. The big floating thing with reliability now is that 5 9's period or 5 9's scedualed there is a BIG difference. Work for data centers ISP IPP etc and they start to get it after there 2 9's gear starts breaking down about now and they have to shutdown to fix it. It's all about life span redundant power supplies does very little to help you at a year old assuming you burned it in well. It's when things get 3 years old (a lot of the dot com gear now) and it fails that you have issues. Acheaving 5 9's isn't done in hardware or people or resources alone it's a combination of all of it and a work methodology that everybody has to adhear to. I can claim 90 9's of uptime and be right if my only production time is 2 minutes a day it's a lot harder when status affecting work windows are included in those numbers because there is no "late night" anymore people access things aroud the world all the time yea traffic might be reduced but reduced does not mean off.
I can agree with that assessment look at the bsic install on windows everything whats to make you reboot. Tech support wants you to reboot. I did my time at IBM in tech support and if you told somebody to just reboot a rs6k even that wouldent be accepted forget a as400 or 390 gear. Reset a process or peice of hardware yea but never the system granted those are all a lot more painfull to reboot because of it it starts at the installation on windows and goes from there.
On a side note my local telco wants me to reboot anytime something happems with the DSL now I have a 99.999 SLA with them (business class 6 in 1.5 out) and they own up to the DSL bridge. I dont see this as conducive to 5 9's of reliability I have never in 7 years of working with real rotuers had to reboot one for anything but to reload an IOS or recover a password.
Most programers need to be retaught how to program anything reliable. OK yea thats flame bait but in reality at least for whats comming out of programming shops nowadays it's mostly garbage this isn't nessicarly the programmers fault as much as get it out the door management and patch it later additude. I think the net with easy patch delivery has caused a whole new revilution in shoddy workmanship. But I am biased I left programming and moved to engineering a long time ago. Unfortunenatly under windows it's hard to tell if the application messed up or the operating system brew chucks or if the hardware is just junk.
Actualy in these cases I'm generly activly trying to get backup. I'm not the on call person like some small mom and pop but if you have ever built anything thats business critical and something comes up that the 2 tiers below me cant deal with then they will give me a call. It's not my ego it's the reality of midsized business. The person that built a system is responcible for it period untill somebody else takes over that responcibility. Now what you seem to fail to understand is most management and especialy technical management is expected to be reachable 24/7/365 is something particulary bad happens. This might mean fielding a call every month or so. Personaly I think of it this way the better I do my job the less calls I get.
Now if I got paid to sit at home by the phone I would but untill that happens I'm going to go out and have fun and if something happens I feel the phone go off find a secluded spot and answer it/check messages etc. Now when I was oncall 24/7/365 yes it was critial and no it couldent wait and hour when you loose a few hundred k an hour in revenue not it's not something that can wait unless it can wait untill you find yourself another job. Sometimes you get called in and it's fixed by the time you get there and if having to run into work happens once a quarter it happens and it's part of the job (thats my ego:)
As somebody that has needed to be reachable 24/7/365 and actualy enjoys the true theater, broadway, opera etc I see nothing wrong with cell on vibrate. If I get a call (it's happened many times) generaly its for something trivial I need to chew out the noc manager for letting reach me or on the odd occation it's something important you take it out on the lobby. It's no more disturing than somebody that needs to leave to use the bathroom before intermission. I also wait at the back of the theater untill intermission or a quick set change to return to my seat if I'm not on the aisle (I try to get seats there just in case)
Now this being said I think it's more of an issue in the movie theaters when there are piles of people that dont NEED to be in contact but still feel the need to answer there phone. Oh and for some of us this is what SMS if for if it's realy a problem they leave me a text message that I can read right on the phone and decide wether or not it's important.
Well it's nice to see the bait and switch methds being used:) Realy this would be a boon to my travaling at least. Funny I wonder what the latency will be like and why they need such a big router to do the job. A 265x can run things wire speed to 100bt easly unless they have all sorts of things going on. Now I winder will they be handing out public IP space and what the SWIP will look like for that one:)
That was the Policy at the IBM facility I worked at in the early 90's. I tossed piles of computers into this big ugly compacting trailor once that was done with it I doubt you could recover anything. Funny thing about that is employies took piles of "compacted" parts home with them well I guess if they wanted the data in the first place they could have gotten it anyway in building security was light network wise untill you hit big iron.
I understand this about patents but this would seem to be a special case if my previous premis is true. SCO was part of a company that released a free version of windows along with a commercial verion per the GPL. Now my question is how SCO can claim infringment on something they gave away for free per the GPL. It dosent seem to matter wether or not they wrote it but rather that they distributed it for free under the GPL code that infringed on there patent that to me and I'm not a lawyer would seem to be giving it implicit permission effectivly giving up there rights tot he patented material at least as it pertains to Linux now if somebody else uses that material outside of linux then it would seem to be questionable. pretty much my line of logic is they gave a free liscence in perpituity to the Linux movement when they released for free the infringing code.
Hold the phone SCO was part of a company the distributed Linux for free correct? If this is true how can they claim pattent infringement on something they gave out for free under the GPL? It's one thing to say thats an infinging work pay me but it's another to have one part of the company releasing it for free under the GPL then get spun off and claim it's yours again.
Now granted they can claim infringement on anything but it's another thing alltogether to distribute infinging work then claim it's infringing. I shouldent be able to pattent something release it for free under a liscence then go and sue for infringement as I did give out that code.
Slashdot Mirror
Actualy BGP can pick a path based upon bandwith but the first and foremost descision is AS path length and prefix length. You can inject other metrics into BGP as well say local link utilization at the peering point. In general it's a BAD idea push trafic to longer distant peers rather than say get more peering bandwith. And BTW hop count never comes into play ok AS path length BGP dosent know or care about how many intervieing routers are there just how many AS's. Some providers like AbiveNet work in the reverse and try and keep the data on there network for as long as possible the easiest way to do this is to summerize with distance so local /24's are advertised say for NYC to the NYC peers but agrigated up into a /19 in Cali More specific routers are allways prefered.
OK With all that said traffic engineering with BGP can be fun and it's allways a question of how much memory the routers have (BGP tables are BIG and router Memory is was to small think 32 megs ish for the current 110k line BGP tables (and Telstra is responcible for most of those the whole internet would thank them if they thaught of agrigating there routes and buying fargin bigger pipes instead of making the rest of the world backhaul for them)
Most of the cost of a T1 is the line not the bandwith. Normal numbers for bandwith in bulk are down to sub hundreds a month for quality bandwith providers if baught in bulk and properly negotiated. And i'm not talking about cheap bandwith I'm talking nice tier one pretty much everybody but UUNet and AT&T (they are never cheap but still in the low hundred). Oh BTW before you flame me I currently in proccess of quoting a few hundred megs a sec from all the major carriers I can find these are the numbers after 2-3 rounds of negotiations. Granted you still need to account for leased lines from those handoffs throughout your backbone costs of routers etc. T1's are all leased line cost because they are a PITA on real copper or have a lot of up front costs converting them and throwing them on fiber. A DS3 circut that is 30 times faster generaly only runs 5-10 times the cost.
As somebody who was shipped of to work at IBM while in HS by the HS I would have to disagree. Starting early is a great idea especialy in intern and other low level jobs that you need to build up a resume. It can also allow people to WORK through school and not start out life so far back in debt. The partys and socalizing of school may be fun and good networking but if mommy and daddy cant pay for it and your not a jock or a perfect student then I would say taking a 2nd shift help desk position of junior admin is a great way to get a leg up on your beer guzzaling friends (you can do that on the weekend like an adult er 20 something :)
I can see this working very well when combined with a good local after school internship program. It can allow high school students to find a career that they enjoy earn enough money in HS and later to put themselves through school. Yes this is a tough way to do things but I'll teel you this I'll hire a hot teenage geek that loves to do this over some college kid that isn't sure what they want to do with there life.
BTW dont tell me kids dont know what to do go back a few hundred years and people had families by age 18 and a career this is just our society playing one up on itself every generation.
My problem with that analogy (SP?) is the internet is not your house and neither is your internet connection. Houses dont do things by themselves on the most part. Houses dont make money in general. This is a technical problem and thus needs a technical solution. Servers need to come with sencable defaults. But the biggest thing is people need to turn anything off they dont use. There is a difference between leaving your door open in a nice neighborhood is one thing the internet is the worst of the crack neighborghoods at best treat it that way.
It's it more important that MS SQL server shouldent be exposed to the internet directly in the first place. There are no public SQl servers than I can think of and no reason for them besides maybe some open testing and compatability public labs. Port filtering isn't a panecea but it's the second line of defence (after egress filtering by everybody) there is no reason that a SSH port forward or a VPN cant be used hell a GRE tunnel with no encryption instead of having it open and on the internet. This is also the case for many other packages how many MySQL ports I have seen open it's disgusting.
Well apparently you have figured out that a good DOS attack can flood the box with packets effectivly flooding out the valid incomming packets you need something like local QOS to fix this and it has to be set up correctly.
Next step is the switch in front again dependant on the type of switch it may have problems with a DOS attack as far as management goes.
The router above That may hav it's wan links flooded effectivly edging out valid traffic again and dependant on the router type taking up enormous ammounts of CPU time.
Then there are all the upstreams. When you buy bandwith from some local provider like they have they dont have the capacity to deal with the attack nessicarily and at minimum it may be degrading there network.
The only good solution to fix these things is source address verification from ALL ISP's (ok not happening any time soon) and oh that breaks mobil IP and some multi homes sites even the cheasy multiple DSL and cable modem setups for more bandwith and reliability configs.
This is why real VPN software includes a firewall that it's configured by firewall admin not the end user. Granted there are ways around this but they should be monitored and offences that can lead to terminiation.
So your saying you haddent patched MS servers on the internet with a patch thats is over 6 months old? And this is for a business?
That depends on what sort of routers they are using. A lot of Cisco gear an others have overhead running ACL's (first Cisco that can do compiled is a 7200 I think) so you want to accept/deny your most common traffic first so your router dosent die when your throughput goes up.
More likly in my experience firewalls get configured to allow al trafic out not just traffic that it will allow responces to (simple port based firewalls with no connection trackings realy hardly considered firewalls anymore) a realy simple port filter might allow all incomming tcp besides initial, dissalow all udp except to servers and ports they know about (Primary DNS boxes, NTP any other UDP service) This is very common setup with older gear especialy with publicaly addressed boxes.
This is why you should never buy works with any form of copy protection on the media. You know what I dont care about the media the media is extranious. Good copy protects scheems do just that make sure only I can use it and that 3 billion people dont copy my copy. Now I dont care if they force me to register via phoe if I'm not net connected it's a reasonable assumption if you can afford a computer you can afford to call a 800 number. No software shouldent tattle but it's resonable for software to imprint itself on the machine somewhat as long as the company allows me to reinstall / migrate software at whim within reason. But realy good software will just avoid copy protection it's bad for there users and says they dont trust them to be responcible because for a game unlike what they thing it will be cracked it's just a question of time.
Well the bloat thing seems to mostly be PS programmers making new high level abstratic GARBAGE that makes the coding easier but generaly less efficient. Granted this is from somebody that dosent have a interactive desktop with less than a gig of main ram generaly because it's cheap and I have some ugly bloatware apps that I use. At the same time I have Linux boxes running on as little as 16 megs or ram and thats not to hard to do (RH dosent like less than32 for an install now though) 32 makes a good firewall or MP3 Player etc etc etc.
I think you would have a difficult time writing 14 channels at once mostly you have buss contention on the PCI side or whatever is connecitng you northbridge to your south bridge remember I2O isn't running on something like this the CPU has to see probably all the data unless they have zero copy buffers for disk data now (not sure I know network wise that was one of the big TUX things) IDE Raid will help with some of this especialy if it has some buffer on it. I would go with 2 drives on 2 channels aka a drive a channel if you want any sort of performance while running IDE still has or untill recently had issues with bus contention and switching from master to slave sorta like OS contex switching.
SCSI or FC would be a better solution dependant on your budget. Old school SGI gear can write SDI data to disk for 4 tracks cheaply enough and you could encode to Divx / whatever your choice is via a HSM like proccess so if your watching it live and using pause were talking nearly Zero loss (those A/D converters for SDI are realy realy high quality as a tab bit expensive) As I think 4 channels was the best the workstation class boxes from a few years back could handel it would work out well as the ingress / egress boxes (go with clustered disk and CXFS and you have a REALY scalable solution just dangle off more encoder / decoder heads) granted this may be a lot more that somebody wants to do for a home project but recording 16 channels of vid in real time sounds like something more than home use.
Here Here lets second that the Newton was the first inovating PDA. The orgigional newton read my HORID hardwritting better than more modern designs in other PDA's or relearing how to write with grafiti. The line drawing abilites were great as well it was easy to draw up a quick deisgn note.
Now thing of what that could do in an age with 512 Meg flash cards running pretty cheap and the general avalibility of wireless (Nextels flat rate service for 60 a month with good coverage per bit is a horid idea for a consumer great for providers though) Roll that up into something that dosent need to fit in my shirt pocket but pretty much needs a screen, a way to get a lot of memory into it, good rechargable battery life and a good OS (maybe Palm emulation if the hardware can deal and old school newton emu as well) I would pay 500 or so for this without the big memory card etc. If they can get it to sub 300 it makes mass market.
Think about it the ability to use Palm and Newton Apps + new apps for the platform. I dont realy need an orginizer I need a portable web browser / pdf reader / email / Terminal emu (OK I'm a network Engineer the serial port on my Newton keeps it in my tool box to this day along with an ugly 8pin to rj45 adapter dongel I made)
There is one problem with this at least Mobile IP it's a spec it's designed to work and it dosent provide a return path like a VPN. Now granted most people would be better served by a VPN than using mobile IP. Next is ou have to remember there are a LOT of things that assume packets can originate from any address.
/8 are given out. It's easier to tell an aggrigation router to only allow packets say from the /32 of a modem user than make all the edge routers filter things.
Geo Load ballancers, especialy the ones that use DNS timed race now this is somehting with a fixed Ip address but it's hard to keep up all sorts of exceptions. BTW this works pretty well.
Dual headed users (say DSL and Cable) that LB there outgoing requests. Now granted this is minor and isn't supposed to work but it's nice and gets you a good deal of redundancy.
Multihomed servers, they can send packets from any of there interface IP's via any of it's interface IP's some mail servers and DNS servers do this alot. Remember GigE is a new thing and servers have seen 100bt as a limitation for awhile.
And there are a lot of other issues. Now I have setup IPP's to do this and it works well enough at say a up to a tier 3 provider. This will give every multihomed carrier problems with traffic management especialy ones with old routers not taking more than a BGP default, not having the same carriers at every router, running disconnected AS's (Ok this is a no no but they do it)
Basic egress filtering should be in place for everybody all the martians should be blocked perferably at the ingress routers doing DSL, Cable and Modem agrigation. Yea this has some maintnence associated with it as the martians list changes every now and again as new
Yea 5 9's can be tough to get the execs to pay for but it's a large part of my engineering life. The big floating thing with reliability now is that 5 9's period or 5 9's scedualed there is a BIG difference. Work for data centers ISP IPP etc and they start to get it after there 2 9's gear starts breaking down about now and they have to shutdown to fix it. It's all about life span redundant power supplies does very little to help you at a year old assuming you burned it in well. It's when things get 3 years old (a lot of the dot com gear now) and it fails that you have issues. Acheaving 5 9's isn't done in hardware or people or resources alone it's a combination of all of it and a work methodology that everybody has to adhear to. I can claim 90 9's of uptime and be right if my only production time is 2 minutes a day it's a lot harder when status affecting work windows are included in those numbers because there is no "late night" anymore people access things aroud the world all the time yea traffic might be reduced but reduced does not mean off.
I can agree with that assessment look at the bsic install on windows everything whats to make you reboot. Tech support wants you to reboot. I did my time at IBM in tech support and if you told somebody to just reboot a rs6k even that wouldent be accepted forget a as400 or 390 gear. Reset a process or peice of hardware yea but never the system granted those are all a lot more painfull to reboot because of it it starts at the installation on windows and goes from there.
On a side note my local telco wants me to reboot anytime something happems with the DSL now I have a 99.999 SLA with them (business class 6 in 1.5 out) and they own up to the DSL bridge. I dont see this as conducive to 5 9's of reliability I have never in 7 years of working with real rotuers had to reboot one for anything but to reload an IOS or recover a password.
Most programers need to be retaught how to program anything reliable. OK yea thats flame bait but in reality at least for whats comming out of programming shops nowadays it's mostly garbage this isn't nessicarly the programmers fault as much as get it out the door management and patch it later additude. I think the net with easy patch delivery has caused a whole new revilution in shoddy workmanship. But I am biased I left programming and moved to engineering a long time ago. Unfortunenatly under windows it's hard to tell if the application messed up or the operating system brew chucks or if the hardware is just junk.
Actualy in these cases I'm generly activly trying to get backup. I'm not the on call person like some small mom and pop but if you have ever built anything thats business critical and something comes up that the 2 tiers below me cant deal with then they will give me a call. It's not my ego it's the reality of midsized business. The person that built a system is responcible for it period untill somebody else takes over that responcibility. Now what you seem to fail to understand is most management and especialy technical management is expected to be reachable 24/7/365 is something particulary bad happens. This might mean fielding a call every month or so. Personaly I think of it this way the better I do my job the less calls I get.
:)
Now if I got paid to sit at home by the phone I would but untill that happens I'm going to go out and have fun and if something happens I feel the phone go off find a secluded spot and answer it/check messages etc. Now when I was oncall 24/7/365 yes it was critial and no it couldent wait and hour when you loose a few hundred k an hour in revenue not it's not something that can wait unless it can wait untill you find yourself another job. Sometimes you get called in and it's fixed by the time you get there and if having to run into work happens once a quarter it happens and it's part of the job (thats my ego
As somebody that has needed to be reachable 24/7/365 and actualy enjoys the true theater, broadway, opera etc I see nothing wrong with cell on vibrate. If I get a call (it's happened many times) generaly its for something trivial I need to chew out the noc manager for letting reach me or on the odd occation it's something important you take it out on the lobby. It's no more disturing than somebody that needs to leave to use the bathroom before intermission. I also wait at the back of the theater untill intermission or a quick set change to return to my seat if I'm not on the aisle (I try to get seats there just in case)
Now this being said I think it's more of an issue in the movie theaters when there are piles of people that dont NEED to be in contact but still feel the need to answer there phone. Oh and for some of us this is what SMS if for if it's realy a problem they leave me a text message that I can read right on the phone and decide wether or not it's important.
This is because MS has a monopoly position and they are calling something that realy isn't fully Java Java.
Well it's nice to see the bait and switch methds being used :) Realy this would be a boon to my travaling at least. Funny I wonder what the latency will be like and why they need such a big router to do the job. A 265x can run things wire speed to 100bt easly unless they have all sorts of things going on. Now I winder will they be handing out public IP space and what the SWIP will look like for that one :)
That was the Policy at the IBM facility I worked at in the early 90's. I tossed piles of computers into this big ugly compacting trailor once that was done with it I doubt you could recover anything. Funny thing about that is employies took piles of "compacted" parts home with them well I guess if they wanted the data in the first place they could have gotten it anyway in building security was light network wise untill you hit big iron.
I understand this about patents but this would seem to be a special case if my previous premis is true. SCO was part of a company that released a free version of windows along with a commercial verion per the GPL. Now my question is how SCO can claim infringment on something they gave away for free per the GPL. It dosent seem to matter wether or not they wrote it but rather that they distributed it for free under the GPL code that infringed on there patent that to me and I'm not a lawyer would seem to be giving it implicit permission effectivly giving up there rights tot he patented material at least as it pertains to Linux now if somebody else uses that material outside of linux then it would seem to be questionable. pretty much my line of logic is they gave a free liscence in perpituity to the Linux movement when they released for free the infringing code.
Hold the phone SCO was part of a company the distributed Linux for free correct? If this is true how can they claim pattent infringement on something they gave out for free under the GPL? It's one thing to say thats an infinging work pay me but it's another to have one part of the company releasing it for free under the GPL then get spun off and claim it's yours again.
Now granted they can claim infringement on anything but it's another thing alltogether to distribute infinging work then claim it's infringing. I shouldent be able to pattent something release it for free under a liscence then go and sue for infringement as I did give out that code.