In general it a good thing to keep developers off production boxes and kitty corner them into test and devel boxes so that everything is well documented. You dont want to force a devel to document every change they make to there development server just every change they had to make to get it working in preproduction so that the admins can do the same in production. Besides production should never have x running and putty etc gives a better envirnement than local text console.
Nope we can get OTA digital straight into the PC but direct TV does not provide any digital outs. There is a hack to get firewire out of there HD receivers but it's a lot more expensive than just buying HD Tivo's. It should be possible to script a connector to have myth control the rocording on an HD tivo and automate the transfer over to a proper myth backend and vice vera since there is a web interface for sceduling recordings and CLI for moving content (not sure if it can stream a currently recording show I know it can playback a streaming show) but those tow are the only digital options I can speek of.
Cable cards are avalible from some cable companies it's not widespread and not all the channels are digital to start with so you may need to go analog for some channels. I dont know if pc cards capable of receiving digital cable with cable cards are currently on the market and supported by linux / myth generaly they suffer from excessive pricing as I know here is $12 per additional card or nealerly 20 more per month than directv is charging for more tuners.
As somebody that had a Myth setup and Tivo's, you can do the same thing to any tivo that has been hacked (no harder than a mythtv setup). You can also do all the transcoding you would like, play your avi's etc via the tivo. Tivo's are realy good at watching things, myth has mame and thus piles of games and does a good job of recording things.
For me the quality loss of myth isn't aceptable (hr10-250 HD tivo's) and that is not going to change untill firewire recording from digital cable becomes possible. Myth is a great project and could do what I want (HD Directv receivers can be hacked to provide firewire out) but the HD tivo's do a better job at less cost myth front end gets used for games and weather.
Loggin of IM's is required for some companies legal dept, depending on how you read HIPPA anything where your sending data out of the envirnment where it's not logged for review would qualify as not having accpetable safegaurds on data.
Your looking for iSCSI. Google for iscsi target drivers for linux and you can export files or drives to any other system. That takes care of the block device. Multiple reader and writer file systems can be had if you look and feel like spending.
You can duplicate the functionality of the device with a pda on any decent wireless data network except that for email etc it's push technology like a pager and without sacrificing battery life. At ti's core it's a two way pager more than an cell phone. This means it's not activly checking your email all the time it's just waiting for a packet with it's name on it so it does not have to transmit except to ack the data being received. If your not in range you will get all your new emails etc once you are.
It does all this with good encryption (meets hippa) and battery life measured in days.
Ok some simple things like running gycol AC that do not have to run a compressor when it's cold out (circulation pumps and fans are a lot cheaper)
Ducting the return air to the outside when the outside (or basement:) air is cooler and does nt need significant humidity adjustments. You will go through a lot more air filters but it's cheaper. Depending on the building the basement is actualy a pretty massive heat sink to the ground this works great if it's mostly open bulk storage etc. It also has the added advantage of keeping the basement air circulating.
Get power monitoring per circut to fild your bigest issues so you put time and effort into the right places. If your mostly an internet facing colo investigate any trends the differ from your outbound MRTG. For example I once found that the data backup systems were causing a lot of power usage during the mid day peek and could be shifted.
15k SCSI is a power hog, if you need IO's look at solid state anything else can run on slower drives with good savings. As an added bonus anything that needs the IO's like DB's will scream on solid state.
Insulate interior walls aroun the DC th ceiling above etc it will cut down on noise and heat tranfer from your normaly much warmer spaces around. Ancilary rooms like power block rooms can be ducted to outside air all the transformers panels etc are designed to be outside and can generate a lot of heat load, UPS's not as much they need to avoid getting to hot but batterys dont like being cold. You would be surprised it was taking 10kw of cooling to take care of the power room at one DC I was at.
Check power in front of and after your UPS some models can cost you 20% or more due to there design. This is a trade off in switchover time though so it is something to be looked at vs your SLA's and desired public image. A lot of real server class gear does nto mind increased switchover time but cheap PC PSU's are not as tollerant (it's all about the Cap's in the PSU)
Fuel cells can be a good alteritive or primary power soruce depending on pricing of natural gas vs the power grid. They have to be one of the most reliable sourced of power generaly because the gas companies bury most of there infrastructure. On site backup can be via propane.
I will second that. All small sysadmin type jobs have been through referal for small companies they are mostly looking for a single face they get to know and trust.
Matched drives give you better performance but they are not technicaly required. Some raid cards might have checked for this but none that I have worked with. 3ware specificaly does not use a chunk of a drive so that different drive sized can be accomodated I have a 4 drive 2 maxtor 2 WD raid 5 on a 3ware 95xx and it works fine. Cheep windows mirror and stripe software "raid" controlers probably have this issue but it should work fine putting a larger drive in to replace the failling unit, as there logic is a simple write every block twice and say your as big as the first drive or write ever other block to each drive and say yours 2x as big. If your using a "raid" card that cheap you might as well use software raid and get a better feature set (expansion, raid level migration, raid 5 support and sub drive arrays for starters)
Actualy if you can track everybody in the area with this thing you dont have to blow up the houses with people just get a loudspeaker and go ok everybody out of there house we know where you are please exit the buildings with your hands up. Local populace comes out and gets hussled into a holding area. If there is one group left blow them up. If somebody comes out shooting you can see people coming to the door before tehy get there and it would seem to have the resolution to larger weapons as in anything that can take out an Bradley. Yes there will be cases when people dont come out are innocent etc but it's war not police, war accepts collateral dammage as a cost of war police should not be given that (they are more and more but thats another topic).
Well the internet has 3 things that need any sort of central control IP's, ASN's and the DNS root. IP's are a finite resource and have to be given out based upon need. DNS could be made rootless at the server level but you still need some authoritative group to arbitrate domain names etc. mDNS along with some crypto like DNSsec could make a distributed root while maintaining some sence of stability. AS to ASN's it's just a number that acts as a unique identifer in the world wide BGP mesh that makes it all work:) again it's just a question of some group taking responcibility for it publishing a list and being reachable but like the DNS issue it's all about getting EVERYBODY to change over to it. mDSN has the best chance of moving as it's easy to be backwards compatable IP's and ASN's realy would require handoff from the current people that control them.
I guess you didn't read the huge just buy a happauge card they work in the HowTo's??
Realy the only huge issue that stops me from using MythTV is quality loss as compared to my HD direct tivo's, and there appears to be a rather unsightly and expensive fix for that one as well (hacked HD direct tivo's that output HD via firewire) I say unsightly as it requires anther external PC to run there code and isn't very living room friendly at least untill I get to putting in an equipment closet. I would love to see the day when we get a real interconnect standard in place firewire works very well.
Actualy your incorrect, the/64 can be the public IP's on your cable modem etc (remember it's a L2 bridge) same goes for your cell phones etc. Your descibing one possible addressing scheme that is not required. Per http://www.ietf.org/rfc/rfc2462.txt providers are free to use a statefull addressing scheme. A statefull scheme could be seen by the provider to provide accounting. Nowhere in the IPv6 spec does it require that ISP's assume a router at the customer end rather they are free to assume a single end station and have there gear enforce that assumption.
And the ISP's are asking where the profit in doing it is? There are a lot of downfalls to providers Multicast being the big one along with a whole lot of training. I do love all the people that think all of a sudden there toasters can have real IP's and NAT will go away, nothing in IPv6 says they have to give you more than one IP without paying more for it just like today.
Memory coruption happens it's a function of radiation and ECC does not fix/catch all of it. Routers have uptimes counted in years unlike your average PC or windows server so yes it does make a bit of sence.
Well most cisco routers have socketed and/or slot based flash. The slot based ones have these realy cute write protect switches on the end.
OK you have general peering a bit messed up here is a normal agreement for UUNet http://global.mci.com/uunet/peering/ this is for tier 1 peering. It's expected that tier 1's will have a mix of content and users to acheive a fairly ballanced interconnect. Assuming you have the traffic you can per with UUNet for nothing but your cost to get to the common interconnect points.
OK as somebody that has MS on there laptop because of visio. Visio is not a drawing program for most it's a macro / layout program. Try walking into an unknown undocumented network there are no good network mapping tools for Linux and only two for Windows and those need visio to display. Yea you can put pretty stencils in visio but for my business it's all about it's macro and API as a lot of software builds ontop on visio.
I'm fairly sure 3ware has allready taken care of this, when running with a BBU it keeps a copy of all data in the sata drives write cache and writes it to disk again if the power is lost.
OK first things first figure out the IO's you need to do and how they need to scale. If your looking for just bulk storage look into some nice big SATA drives. 4RU cases can get you 24 500 gig drives with 22 usable in raid 5 on a pair of 3ware 12 port or similar raid controlers or 11TB's per Unit. Serve these rater large chunks up with iSCSI. Take a HA cluster and merge those chunks together with software raid. The end servers just need to be fast enough to handle your interconnect speeds (gig or better I would hope) the HA pair needs a good deal of computaional ability to do raid calcs. All of them can use as much ram as you can shove in them if performance is a goal.
This isn't the fastest config by far but it's cheap and reliable.
Now with this being said there generaly isn't any good reason to make a disk that big, split things up if at all possible you do not want to deal with a PB of data in on huge volume.
Funny building a car isn't that hard the power plant is rather complicated but the rest can be made of some nice tube stock and sheet metal for the most part. Granted it wont ride as nice as most commercial cars but it will stand up in an accident better than any of them. Granted I'm talking about good old fashion dune buggy with sheet metal attached. Never had any federal guide line issues just one state inspector made sure nothing would fall off and the wheels were covered.
Better yet ask for service credits. Have a few clients with L3 and / or Cogent they all have problem tickets in for the lack of reachability to the other and have allready gone into some pretty stiff penalties under there contracts. Yes L3 and Cogent will try and weasel there way out of the service credits but it's costing each of them money to do so and as that number goes up they will have to figure out a way to make it work.
OK I do this all day long so let me try and expain how teir 1's work in general yes there are exceptions. Teir 1's peer in a fully meshed network meaning all tier 1's have to have connections to all other teir 1's generaly in a multitude of locations. Teir ones only advertise the routes of themselves and there clients not routes learned from there peers. If you want a full set of routes then you need to pay for your connection. This actualy helps stability on the day to day as all teir 1's connect to all the other teir 1's thus nobody is transiting traffic from one to the other meaning L3 could go off the map but that only affects them and there single holmed clients (single holming is BAD)
Cogent is not a bonified teir 1 as they still pay for some of there transit.
1 OK sounds like a good idea realy thats just a devel issue on UI Design.
2 Ick windows installs are not standardized most just use the windows easy to deal with one. The whole registry nastyness is ugly. As to Linux requiring a graphical installer is pure BS there are plenty of people in text mode because it works and works well. If Linux needs anything it's a standardized robust packaging system something that can deal with custom compiling a kernel. Once you have that you can make functional GUI and command line tools to deal with instalation and upgrades. Working like Windows is broken if Linux is going to do it it should do it better.
3 Read #2 realy your talking about one issue and a couple different aspects.
4 Go out and write GUI tools to do so if you like, it's not something that developers want to take care of in general I would think Linspire or the like has that functionality. I would allso assume KDE or Gnome is working on a bit to do just that though I remember a network places thing in one of the two the last time I logged into a Linux GUI (blame mythtv). BTW I do mount windows network shares in XP with the command line net program it's much faster than the GUI tools.
Been there done that, AFS http://www.faqs.org/faqs/afs-faq/ works wonders. Pretty much it's a nice fault tolerant file sharing system that supports direconnected ops meaning you can work with everything in disk cache and checkout / checkin things as needed.
Slashdot Mirror
In general it a good thing to keep developers off production boxes and kitty corner them into test and devel boxes so that everything is well documented. You dont want to force a devel to document every change they make to there development server just every change they had to make to get it working in preproduction so that the admins can do the same in production. Besides production should never have x running and putty etc gives a better envirnement than local text console.
Nope we can get OTA digital straight into the PC but direct TV does not provide any digital outs. There is a hack to get firewire out of there HD receivers but it's a lot more expensive than just buying HD Tivo's. It should be possible to script a connector to have myth control the rocording on an HD tivo and automate the transfer over to a proper myth backend and vice vera since there is a web interface for sceduling recordings and CLI for moving content (not sure if it can stream a currently recording show I know it can playback a streaming show) but those tow are the only digital options I can speek of.
Cable cards are avalible from some cable companies it's not widespread and not all the channels are digital to start with so you may need to go analog for some channels. I dont know if pc cards capable of receiving digital cable with cable cards are currently on the market and supported by linux / myth generaly they suffer from excessive pricing as I know here is $12 per additional card or nealerly 20 more per month than directv is charging for more tuners.
As somebody that had a Myth setup and Tivo's, you can do the same thing to any tivo that has been hacked (no harder than a mythtv setup). You can also do all the transcoding you would like, play your avi's etc via the tivo. Tivo's are realy good at watching things, myth has mame and thus piles of games and does a good job of recording things.
For me the quality loss of myth isn't aceptable (hr10-250 HD tivo's) and that is not going to change untill firewire recording from digital cable becomes possible. Myth is a great project and could do what I want (HD Directv receivers can be hacked to provide firewire out) but the HD tivo's do a better job at less cost myth front end gets used for games and weather.
Loggin of IM's is required for some companies legal dept, depending on how you read HIPPA anything where your sending data out of the envirnment where it's not logged for review would qualify as not having accpetable safegaurds on data.
Your looking for iSCSI. Google for iscsi target drivers for linux and you can export files or drives to any other system. That takes care of the block device. Multiple reader and writer file systems can be had if you look and feel like spending.
You can duplicate the functionality of the device with a pda on any decent wireless data network except that for email etc it's push technology like a pager and without sacrificing battery life. At ti's core it's a two way pager more than an cell phone. This means it's not activly checking your email all the time it's just waiting for a packet with it's name on it so it does not have to transmit except to ack the data being received. If your not in range you will get all your new emails etc once you are.
It does all this with good encryption (meets hippa) and battery life measured in days.
Ok some simple things like running gycol AC that do not have to run a compressor when it's cold out (circulation pumps and fans are a lot cheaper)
:) air is cooler and does nt need significant humidity adjustments. You will go through a lot more air filters but it's cheaper. Depending on the building the basement is actualy a pretty massive heat sink to the ground this works great if it's mostly open bulk storage etc. It also has the added advantage of keeping the basement air circulating.
Ducting the return air to the outside when the outside (or basement
Get power monitoring per circut to fild your bigest issues so you put time and effort into the right places. If your mostly an internet facing colo investigate any trends the differ from your outbound MRTG. For example I once found that the data backup systems were causing a lot of power usage during the mid day peek and could be shifted.
15k SCSI is a power hog, if you need IO's look at solid state anything else can run on slower drives with good savings. As an added bonus anything that needs the IO's like DB's will scream on solid state.
Insulate interior walls aroun the DC th ceiling above etc it will cut down on noise and heat tranfer from your normaly much warmer spaces around. Ancilary rooms like power block rooms can be ducted to outside air all the transformers panels etc are designed to be outside and can generate a lot of heat load, UPS's not as much they need to avoid getting to hot but batterys dont like being cold. You would be surprised it was taking 10kw of cooling to take care of the power room at one DC I was at.
Check power in front of and after your UPS some models can cost you 20% or more due to there design. This is a trade off in switchover time though so it is something to be looked at vs your SLA's and desired public image. A lot of real server class gear does nto mind increased switchover time but cheap PC PSU's are not as tollerant (it's all about the Cap's in the PSU)
Fuel cells can be a good alteritive or primary power soruce depending on pricing of natural gas vs the power grid. They have to be one of the most reliable sourced of power generaly because the gas companies bury most of there infrastructure. On site backup can be via propane.
I will second that. All small sysadmin type jobs have been through referal for small companies they are mostly looking for a single face they get to know and trust.
Matched drives give you better performance but they are not technicaly required. Some raid cards might have checked for this but none that I have worked with. 3ware specificaly does not use a chunk of a drive so that different drive sized can be accomodated I have a 4 drive 2 maxtor 2 WD raid 5 on a 3ware 95xx and it works fine. Cheep windows mirror and stripe software "raid" controlers probably have this issue but it should work fine putting a larger drive in to replace the failling unit, as there logic is a simple write every block twice and say your as big as the first drive or write ever other block to each drive and say yours 2x as big. If your using a "raid" card that cheap you might as well use software raid and get a better feature set (expansion, raid level migration, raid 5 support and sub drive arrays for starters)
Actualy if you can track everybody in the area with this thing you dont have to blow up the houses with people just get a loudspeaker and go ok everybody out of there house we know where you are please exit the buildings with your hands up. Local populace comes out and gets hussled into a holding area. If there is one group left blow them up. If somebody comes out shooting you can see people coming to the door before tehy get there and it would seem to have the resolution to larger weapons as in anything that can take out an Bradley. Yes there will be cases when people dont come out are innocent etc but it's war not police, war accepts collateral dammage as a cost of war police should not be given that (they are more and more but thats another topic).
Well the internet has 3 things that need any sort of central control IP's, ASN's and the DNS root. IP's are a finite resource and have to be given out based upon need. DNS could be made rootless at the server level but you still need some authoritative group to arbitrate domain names etc. mDNS along with some crypto like DNSsec could make a distributed root while maintaining some sence of stability. AS to ASN's it's just a number that acts as a unique identifer in the world wide BGP mesh that makes it all work :) again it's just a question of some group taking responcibility for it publishing a list and being reachable but like the DNS issue it's all about getting EVERYBODY to change over to it. mDSN has the best chance of moving as it's easy to be backwards compatable IP's and ASN's realy would require handoff from the current people that control them.
I guess you didn't read the huge just buy a happauge card they work in the HowTo's??
Realy the only huge issue that stops me from using MythTV is quality loss as compared to my HD direct tivo's, and there appears to be a rather unsightly and expensive fix for that one as well (hacked HD direct tivo's that output HD via firewire) I say unsightly as it requires anther external PC to run there code and isn't very living room friendly at least untill I get to putting in an equipment closet. I would love to see the day when we get a real interconnect standard in place firewire works very well.
Actualy your incorrect, the /64 can be the public IP's on your cable modem etc (remember it's a L2 bridge) same goes for your cell phones etc. Your descibing one possible addressing scheme that is not required. Per http://www.ietf.org/rfc/rfc2462.txt providers are free to use a statefull addressing scheme. A statefull scheme could be seen by the provider to provide accounting. Nowhere in the IPv6 spec does it require that ISP's assume a router at the customer end rather they are free to assume a single end station and have there gear enforce that assumption.
And the ISP's are asking where the profit in doing it is? There are a lot of downfalls to providers Multicast being the big one along with a whole lot of training. I do love all the people that think all of a sudden there toasters can have real IP's and NAT will go away, nothing in IPv6 says they have to give you more than one IP without paying more for it just like today.
Last hop is Amazon owned IP space they might have been looking to hide the whois on the domain.
whois 207.171.166.182
[Querying whois.arin.net]
[whois.arin.net]
OrgName: Amazon.com, Inc.
OrgID: AMAZON-4
Address: 605 5th Ave S
City: SEATTLE
StateProv: WA
PostalCode: 98104
Country: US
NetRange: 207.171.160.0 - 207.171.191.255
CIDR: 207.171.160.0/19
NetName: AMAZON-01
NetHandle: NET-207-171-160-0-1
Parent: NET-207-0-0-0-0
NetType: Direct Assignment
NameServer: NS-1.AMAZON.COM
NameServer: NS-2.AMAZON.COM
NameServer: NS-3.AMAZON.COM
NameServer: AUTH00.NS.UU.NET
Comment:
RegDate: 1999-09-23
Updated: 2002-03-19
RTechHandle: AC6-ORG-ARIN
RTechName: Amazon.com, Inc.
RTechPhone: +1-206-266-2187
RTechEmail: NOC@amazon.com
OrgTechHandle: ROLEA19-ARIN
OrgTechName: Role Account
OrgTechPhone: +1-206-266-2187
OrgTechEmail: noc@amazon.com
Memory coruption happens it's a function of radiation and ECC does not fix/catch all of it. Routers have uptimes counted in years unlike your average PC or windows server so yes it does make a bit of sence.
Well most cisco routers have socketed and/or slot based flash. The slot based ones have these realy cute write protect switches on the end.
OK you have general peering a bit messed up here is a normal agreement for UUNet http://global.mci.com/uunet/peering/ this is for tier 1 peering. It's expected that tier 1's will have a mix of content and users to acheive a fairly ballanced interconnect. Assuming you have the traffic you can per with UUNet for nothing but your cost to get to the common interconnect points.
OK as somebody that has MS on there laptop because of visio. Visio is not a drawing program for most it's a macro / layout program. Try walking into an unknown undocumented network there are no good network mapping tools for Linux and only two for Windows and those need visio to display. Yea you can put pretty stencils in visio but for my business it's all about it's macro and API as a lot of software builds ontop on visio.
I'm fairly sure 3ware has allready taken care of this, when running with a BBU it keeps a copy of all data in the sata drives write cache and writes it to disk again if the power is lost.
OK first things first figure out the IO's you need to do and how they need to scale. If your looking for just bulk storage look into some nice big SATA drives. 4RU cases can get you 24 500 gig drives with 22 usable in raid 5 on a pair of 3ware 12 port or similar raid controlers or 11TB's per Unit. Serve these rater large chunks up with iSCSI. Take a HA cluster and merge those chunks together with software raid. The end servers just need to be fast enough to handle your interconnect speeds (gig or better I would hope) the HA pair needs a good deal of computaional ability to do raid calcs. All of them can use as much ram as you can shove in them if performance is a goal.
This isn't the fastest config by far but it's cheap and reliable.
Now with this being said there generaly isn't any good reason to make a disk that big, split things up if at all possible you do not want to deal with a PB of data in on huge volume.
Funny building a car isn't that hard the power plant is rather complicated but the rest can be made of some nice tube stock and sheet metal for the most part. Granted it wont ride as nice as most commercial cars but it will stand up in an accident better than any of them. Granted I'm talking about good old fashion dune buggy with sheet metal attached. Never had any federal guide line issues just one state inspector made sure nothing would fall off and the wheels were covered.
Better yet ask for service credits. Have a few clients with L3 and / or Cogent they all have problem tickets in for the lack of reachability to the other and have allready gone into some pretty stiff penalties under there contracts. Yes L3 and Cogent will try and weasel there way out of the service credits but it's costing each of them money to do so and as that number goes up they will have to figure out a way to make it work.
OK I do this all day long so let me try and expain how teir 1's work in general yes there are exceptions. Teir 1's peer in a fully meshed network meaning all tier 1's have to have connections to all other teir 1's generaly in a multitude of locations. Teir ones only advertise the routes of themselves and there clients not routes learned from there peers. If you want a full set of routes then you need to pay for your connection. This actualy helps stability on the day to day as all teir 1's connect to all the other teir 1's thus nobody is transiting traffic from one to the other meaning L3 could go off the map but that only affects them and there single holmed clients (single holming is BAD)
Cogent is not a bonified teir 1 as they still pay for some of there transit.
1 OK sounds like a good idea realy thats just a devel issue on UI Design.
2 Ick windows installs are not standardized most just use the windows easy to deal with one. The whole registry nastyness is ugly. As to Linux requiring a graphical installer is pure BS there are plenty of people in text mode because it works and works well. If Linux needs anything it's a standardized robust packaging system something that can deal with custom compiling a kernel. Once you have that you can make functional GUI and command line tools to deal with instalation and upgrades. Working like Windows is broken if Linux is going to do it it should do it better.
3 Read #2 realy your talking about one issue and a couple different aspects.
4 Go out and write GUI tools to do so if you like, it's not something that developers want to take care of in general I would think Linspire or the like has that functionality. I would allso assume KDE or Gnome is working on a bit to do just that though I remember a network places thing in one of the two the last time I logged into a Linux GUI (blame mythtv). BTW I do mount windows network shares in XP with the command line net program it's much faster than the GUI tools.
Been there done that, AFS http://www.faqs.org/faqs/afs-faq/ works wonders. Pretty much it's a nice fault tolerant file sharing system that supports direconnected ops meaning you can work with everything in disk cache and checkout / checkin things as needed.