This is not something to do as a "newb", as you describe it. Such a large project requires attention to mirroring, high availability, backup, load handling, security, and API's for accessing the data that are beyond a weekend "just set it up and run" project. It's a good time to contact your local DBA's for their guidance, and their preferences, and let them help you save time addressing the concerns they will raise later, and which may not be on the original plan.
Stability for a server class project would suggest a commercially supported product, like SuSE or RHEL. Bleeding edge utility requirements would suggest a development version, such as Ubuntu or Fedora or Mandriva. Debian and gentoo are not good choices for new admins who are not already very experienced and comfortable hand-editing their own configurations, ore turning dozens of badly advised Google referenced solutions into one supportable choice.
Much of this depends on your local resources: will you deploy in a data center with remote support, which many ISP's will happily provide with backup and failover configurations built-in? Or do you need to build your own purely as a proof of concept?
None of this is new. The clear desire to control the ability to access hardware, storage media, to boot an OS, andn to authorized applications to run or access to data, was buiilt into the "Palladium" project and was renamed "Trusted Computing". While much of its glamour has been lost, and the difficulty of enforcing its controls has been shown to be hackable with virtualization, it emains a technology designed to prevent access to hardware and data based on commercial licenses, rather than any security or defense of user data.
This is another attempt at the same goals, to foster and enforce Microsoft monopolies by controlling the ability to use the hardware, itself.
I suspect that the training to manipulate a normal skin resistance based lie detector, such as a Scientology "e-meter", would apply equally well to this technology. It would be fascinating to run the test against someone trained to such hypnotic responses to questions and see if this tool is as easily manipulated by the same mental and physical techniques.
But the IOS comes with the hardware, and that could easily be a modified version. I'm not _assuming_ that the IOS is actually lacks Cisco's unpublished backdoors, but it's another potential motive for Cisco and the US government, to be harsh about counterfeit equipment. And with IOS being closed source, and the backdoors being concealed for so long, it's difficult for a typical network administrator to test.
I'm afraid that NT also included a great deal of VMS, illegally brought over by David Cutler and his colleagues when their project at DEC was cancelled. I'm afraid that the Wikipedia pages and easily available references on Google have lost too much detail, over time, to reflect how much of NT was really VMS. But it still includes this reference to David Cutler's behavior.
> DEC dropped the project he brought the expertise and around 20 engineers with him to Microsoft. DEC also believed he brought Mica's code to Microsoft and sued.[4] Microsoft eventually paid US$150 million and agreed to support DEC's Alpha CPU chip in NT..
ssh-agent leaves an active and unlocked ssh-key available on the relevant server. Many of my colleagues refuse to run their ssh-agent as part of their own persoanl host's working environment, and prefer to host them on their most used server, despite my warnings. Others find ssh-agent burdensome and simply use passphrase-free keys, and there is not yet any graceful way to prevent that except to audit for them on the client hosts, which is awkward and intrusive.
"Patents in secret" are quite popular practice in software and in manufacturing. The source code is in your source control system, and the ultimate product betrays little evidence of its origins. This is one of the big problems with software patents: the end product is difficult for the purchaser to review for how it was made.
There was a famous video at http://www.youtube.com/watch?v=TCbFEgFajGU, of a law firm explaining to employers that they can deliberately run large numbers of fake job ads with too many qualifications to avoid hiring US citizens and creatively select less expensive foreign workers. I'm afraid that the practice is common, not merely for hiring cheaper foreign workers but to allow the bureaucrats to use any criteria they wish to select candidates for traits where are very carefully _not_ listed, such as age, gender, religion, physical health, lack of management experience that might cause a worker to question decisions, union membership, skin color, or nationality. And even where all these are directly against federal law and the published standards of your potential employer, they _will_ affect the employer's willingness to accept partial or equivalent credentials to the listed job requirements. It's also used in salary negotiation, to get the candidate to accept less money on the grounds that they are not ideal for the role.
Tha'ts odd: I'm getting recruiting calls from both UK recruiters, and UK partners trying to get me to come work for them and looking for senior references, and I am _not_ cheaper than their local personnel. I suspect there are some shortages in particular skill sets with few senior people who've studied and mastered multiple technologies. Another potential issue is that the skills of someone who's "built a few machines" often have to be unlearned before they can do robust, professional grade work. But for a home user, they're considered quite expert and quite able to help their friends out with minor issues.
There are limits to their reliability and lifespan: but they are _cheap_ compared to duplicaton costs of a lot of paper documents. The format for CD's and DVD's has, fortunately, been quite stable, and PDF format for the documents is also reliable for being able to read them. It's far safer in my experience to have an organized CD of a year's paperwork and put a spare CD in an offsite location than to try to preserve the paper trail, keep it searchable and legible in local file cabinets. And replicating the paperwork becomes labor intensive and expensive in simple printing costs.
Local backup is useful, especially for data you don't care to publish or have anyone overwrite. Fiscal data and GPG keys, for example, can be usefully stored on permanent media.
I know you're kidding a bit, but I'll answer your point anyway.
I still have several USB external R/W DVD drives for situations where people were too cheap to invest in them for laptops, desktops, or servers. DVD and CD drives are still handy for accessing media you've not bothered transferring to ISO images, or burning CD's or DVD's for others when they don't want to deal with the security risks of a USB memory stick that can be too easily rewritten or used to walk away with data. So for me, at least, I don't need such a drive because it's an unnecessary expense.
With the above racist propagand set aside, the Israelies take their data security very, very seriously. I've met several gifted Israeli mathematicians whose publications were halted in the US for security reasons, due to the US laws on publishing encryption technologies internaitonally.
If you pay me enough money, I will support it. Of course, for the same amount of money and integration work with other more modern software, such as FireFox and Konqueror, you could buy 100 newer, supported, and more powerful machines, or recover dozens of machines by refurbishing and recovering discarded hardwrae that _still_ has newer and supported components.
Of all surprising authors, Terry Pratchett and his collaborators in "The Science of DiscWorld" books covered this very well. Ideas and beliefs can evolve, as part of what Pratchett and his collaborators called "extelligence", and the ability to define members of one's own group and separate them from other groups can create its own very powerful evolutionary effects on both cultures and bodies.
They're fascinating books, especially if you've already read enough Discworld stories to catch the literary references, but the alternating Discworld/real science chapters are very illustrating of the history of Earth in the first book, and of human culture in the second book.
There are numerous US presidents who've had profound medical issues. The range is described well at http://www.doctorzebra.com/prez/t_roster.htm, and it's enlightening. Considering that they're middle aged or older, and the stress of the kind of active life necessary to campaign for president, it's not surprising that they had illnesses or traumas that were at the limit of their era's abilities to treat.
Expecting a presidential candidate to weigh every decision in the light of "what will the voters think" is, itself, political suicide at the highet levels. It will mark a candidate as incapable of making a policy.
All of the blind people who use public restrooms, perhaps? Have you ever tried to find the toilet paper in a public restroom when the bulb has failed, or when the light switch is difficult to find?
This is not only true of software development: it's true in IT as well. Far too often visiting with corporate partners, I see racks of equipment wired like a cat's cradle, where the engineer on site says "oh, we're not very neat" and I see core servers at risk from unlabeled wires, power and cabling that are out of spec and likely to fail under load, and where expensive system redundancy is wasted because the matched pair of servers are plugged into the same switch and the same power supply and the same storage array.
I'm afraid to say that I've become known as an uholy terror about this, because I or a colleague I can trust photograph the racks, bring boxes of cables and appropriate ties and labelmakers, and demand access to their inventory systems and network diagrams to record _all_ the systems. The result is, far too often, political furor as people discover that the much vaunted "high availability" systems were absolutely _not_ high availability. Far too often, fiber optic cables are bent too sharply and unprotected from tangled cables, expensive smart UPS's and switches have never had their configurations recorded, and redundant systems and systems have never been configured for proper failover behavior. And often we discore absolutely mission critical systems that are not on any network diagram or inventory that were ignored in the project plan.
Really, we're as prone to the problems in IT as the devlopers are.
There are problems with our original poster's original claims. Wireless, for example, is very useful for traveling devices and remote smartphones. But safe such access is dependent on VPN or other encryption technologies. But the amount of electromagnetic noise from all the wireless connections is rapidly approaching saturation in crowded locations, and there aren't enough big frequency ranges that the FCC might release to resolve this in the USA. Wired will remain critical for bandwidth and security in the actual working network and systems environments in which we, as engineers, spend so much of our time.
Similarly, much of what is outsourced to the could and _should_ be outsourced to a competent shop. A 20 person company has no business wasting their resources hiring, and keeping trained, someone able to manage core DNS, backup, email, and calendar systems when such cheap and reliable "cloud" systems are available. The services are too important, and too vulnerable to error, for everybody in even "a company full of geeks" to be trusted to run it reliably. I'm afraid I've seen far too many environments where every developer thought themselves capable of running the core IT services, and the results were so uneven and patchy that it was quite unstable.
I _do_ see the Linux transformation of core servers as an ongoing process. The purchase of Sun by Oracle is the death knell for Solaris, and HP is abandoning HP-UX. While Apple has effectively taken over FreeBSD, they've also effectively taken it closed source, and far more storage devices, plug-in email appliances such as database servers and mini-computer based video boxes and smartphones are being Linux based. Even most MS Exchange based shops use a Linux system in front to handle spam filtering. If Samba 4 can ever get out of alpha testing, I expect a huge array of Windows based network storage systems to be replaced nearly overnight by more stable and less expensive Samba based systems.
MS Word will continue to dominate document writing: Powerpoint will continue to dominate presentations. While the open source tools will continue to improve, Microsoft will _again_ manipulate their undocumented and inconsistent API's to break compatibility.
Virtualization is going to get even more interesting. I can't quite picture which way it's going to go, but I do anticipate that a lot of legacy hardware preservation, keeping old systems alive in a rack so we can fall back to recover data or just in case the new system fails, will be replaced by virtualization. And _those_ go to the cloud very well.
Last, I strongly suspect that the migration to IPv6 will still be less than 50% of all devices, probably less than 30%. The benefits of IPv6 simply do not matter to most environments, who are better off in security and network management terms using NAT and thus have no internal need for IPv6. The result will be growth of "mixed-stack" solutions, but until there is reason to leave IPv4 for systems already configured, the "hysteresis" of remaining consistent with existing infrastructure will preclude wholesale migration.
Oh, my I must disagree. Carly started a set of corporate disasters that are still dragging down HP, ranging from their purchase of Compaq (which was an amazing mismatch with HP's former reputation for quality hardware in many fields, and the loss of manufacturing quality by preserving any products or personnal from Compaq is still hurting HP).
There is a good article at http://www.businesspundit.com/10-reasons-people-hate-carly-fiorina/. It points out, correctly, that HP made money _despite_ Carly's misguided decisions, not because of them: HP's printer business continued to be quite profitable for reasons that were in place before Carly's reign.
Currrently, yes. There is no punishment, and in general only modest engineering cost to setting up a new spam net. This encourages new "entrepreneurs" to enter the field, even if they make no overall profit doing so. Spam services are being _sold_ to legitimate and illegitimate clients, and the claims of profit are overblown. But since no one publishes good numbers on its success rates, they can continue lying and drumming up business to fools and criminals.
The return on investment need not be real: it only needs to be portrayed as real to get customers of the wholesale spamming services, which are such a large proportion of modern spam.
Good point. I do think that booting from SAN is a better defined problem, and coreboot is working well with "Etherboot", which hints that network based booting is working well with it. I've not personally tried doing fiber channel booting with it, but it is in use on a lot of small customized network applicances.
It's a great solution long, long before Linux takes over the server and desktop market. It's already in use on the OLPC project, very successfully: the fear of moving off the old and very painful existing BIOS's is more one of familiarity and a maze of intellectual property rights for existing firmware, rather than a technological or performance justification for using the existing systems.
The RAM checks on a new server are going to take quite some time, and should really be allowed to run to completion on a newly installed server. Memory can be jostled or twisted, and some installers will buy cheap third party memory and fail to seat it properly or even buy the wrong models of RAM. Some of them will even buy slower, cheaper RAM and not tell the purchaser, which takes checking to discover.
Actually, they need not take so long. The entire BIOS configuration toolkit has become a nightmare of workarounds and scans for old components that don't even exist, and it's a proprietary software and licensing nightmare to try to tune. The "coreboot" toolkit, formerly known as "linuxbios", has turned a lot of that into genuine GPL freeware. The code is legible, easier to update, much cleaner, and vastly faster. It's awkard to get the initial compatible BIOS loaded, but once it's available for a motherboard, the performance, speed, and ability to _report_ the state of the hardware and BIOS to normal system utilities is well worth the invested effort.
I do wish the motherboard manufacturers would complete their investigations of this technology and switfch wholesale. The code has matured and documentation is now very useful to those building custom applications, and it is in effective commercial use.
Slashdot Mirror
This is not something to do as a "newb", as you describe it. Such a large project requires attention to mirroring, high availability, backup, load handling, security, and API's for accessing the data that are beyond a weekend "just set it up and run" project. It's a good time to contact your local DBA's for their guidance, and their preferences, and let them help you save time addressing the concerns they will raise later, and which may not be on the original plan.
Stability for a server class project would suggest a commercially supported product, like SuSE or RHEL. Bleeding edge utility requirements would suggest a development version, such as Ubuntu or Fedora or Mandriva. Debian and gentoo are not good choices for new admins who are not already very experienced and comfortable hand-editing their own configurations, ore turning dozens of badly advised Google referenced solutions into one supportable choice.
Much of this depends on your local resources: will you deploy in a data center with remote support, which many ISP's will happily provide with backup and failover configurations built-in? Or do you need to build your own purely as a proof of concept?
None of this is new. The clear desire to control the ability to access hardware, storage media, to boot an OS, andn to authorized applications to run or access to data, was buiilt into the "Palladium" project and was renamed "Trusted Computing". While much of its glamour has been lost, and the difficulty of enforcing its controls has been shown to be hackable with virtualization, it emains a technology designed to prevent access to hardware and data based on commercial licenses, rather than any security or defense of user data.
This is another attempt at the same goals, to foster and enforce Microsoft monopolies by controlling the ability to use the hardware, itself.
I suspect that the training to manipulate a normal skin resistance based lie detector, such as a Scientology "e-meter", would apply equally well to this technology. It would be fascinating to run the test against someone trained to such hypnotic responses to questions and see if this tool is as easily manipulated by the same mental and physical techniques.
The ones that have been publicly acknowledged are software based, yes. I see the bug report at http://www.cisco.com/en/US/products/products_security_advisory09186a00805e3234.shtm as a typical example of Cisco vulnerability.
But the IOS comes with the hardware, and that could easily be a modified version. I'm not _assuming_ that the IOS is actually lacks Cisco's unpublished backdoors, but it's another potential motive for Cisco and the US government, to be harsh about counterfeit equipment. And with IOS being closed source, and the backdoors being concealed for so long, it's difficult for a typical network administrator to test.
I'm afraid that NT also included a great deal of VMS, illegally brought over by David Cutler and his colleagues when their project at DEC was cancelled. I'm afraid that the Wikipedia pages and easily available references on Google have lost too much detail, over time, to reflect how much of NT was really VMS. But it still includes this reference to David Cutler's behavior.
> DEC dropped the project he brought the expertise and around 20 engineers with him to Microsoft. DEC also believed he brought Mica's code to Microsoft and sued.[4] Microsoft eventually paid US$150 million and agreed to support DEC's Alpha CPU chip in NT..
ssh-agent leaves an active and unlocked ssh-key available on the relevant server. Many of my colleagues refuse to run their ssh-agent as part of their own persoanl host's working environment, and prefer to host them on their most used server, despite my warnings. Others find ssh-agent burdensome and simply use passphrase-free keys, and there is not yet any graceful way to prevent that except to audit for them on the client hosts, which is awkward and intrusive.
Given Cisco's history of federal cooperation, I do have to wonder if the deeper concern was that such equipment _lacked_ the backdoors Cisco provides for "legal" monitoring. Some examples are described at http://www.forbes.com/2010/02/03/hackers-networking-equipment-technology-security-cisco.html.
"Patents in secret" are quite popular practice in software and in manufacturing. The source code is in your source control system, and the ultimate product betrays little evidence of its origins. This is one of the big problems with software patents: the end product is difficult for the purchaser to review for how it was made.
There was a famous video at http://www.youtube.com/watch?v=TCbFEgFajGU, of a law firm explaining to employers that they can deliberately run large numbers of fake job ads with too many qualifications to avoid hiring US citizens and creatively select less expensive foreign workers. I'm afraid that the practice is common, not merely for hiring cheaper foreign workers but to allow the bureaucrats to use any criteria they wish to select candidates for traits where are very carefully _not_ listed, such as age, gender, religion, physical health, lack of management experience that might cause a worker to question decisions, union membership, skin color, or nationality. And even where all these are directly against federal law and the published standards of your potential employer, they _will_ affect the employer's willingness to accept partial or equivalent credentials to the listed job requirements. It's also used in salary negotiation, to get the candidate to accept less money on the grounds that they are not ideal for the role.
Tha'ts odd: I'm getting recruiting calls from both UK recruiters, and UK partners trying to get me to come work for them and looking for senior references, and I am _not_ cheaper than their local personnel. I suspect there are some shortages in particular skill sets with few senior people who've studied and mastered multiple technologies. Another potential issue is that the skills of someone who's "built a few machines" often have to be unlearned before they can do robust, professional grade work. But for a home user, they're considered quite expert and quite able to help their friends out with minor issues.
There are limits to their reliability and lifespan: but they are _cheap_ compared to duplicaton costs of a lot of paper documents. The format for CD's and DVD's has, fortunately, been quite stable, and PDF format for the documents is also reliable for being able to read them. It's far safer in my experience to have an organized CD of a year's paperwork and put a spare CD in an offsite location than to try to preserve the paper trail, keep it searchable and legible in local file cabinets. And replicating the paperwork becomes labor intensive and expensive in simple printing costs.
Local backup is useful, especially for data you don't care to publish or have anyone overwrite. Fiscal data and GPG keys, for example, can be usefully stored on permanent media.
I know you're kidding a bit, but I'll answer your point anyway.
I still have several USB external R/W DVD drives for situations where people were too cheap to invest in them for laptops, desktops, or servers. DVD and CD drives are still handy for accessing media you've not bothered transferring to ISO images, or burning CD's or DVD's for others when they don't want to deal with the security risks of a USB memory stick that can be too easily rewritten or used to walk away with data. So for me, at least, I don't need such a drive because it's an unnecessary expense.
With the above racist propagand set aside, the Israelies take their data security very, very seriously. I've met several gifted Israeli mathematicians whose publications were halted in the US for security reasons, due to the US laws on publishing encryption technologies internaitonally.
If you pay me enough money, I will support it. Of course, for the same amount of money and integration work with other more modern software, such as FireFox and Konqueror, you could buy 100 newer, supported, and more powerful machines, or recover dozens of machines by refurbishing and recovering discarded hardwrae that _still_ has newer and supported components.
Which makes more economic and engineering sense?
Of all surprising authors, Terry Pratchett and his collaborators in "The Science of DiscWorld" books covered this very well. Ideas and beliefs can evolve, as part of what Pratchett and his collaborators called "extelligence", and the ability to define members of one's own group and separate them from other groups can create its own very powerful evolutionary effects on both cultures and bodies.
They're fascinating books, especially if you've already read enough Discworld stories to catch the literary references, but the alternating Discworld/real science chapters are very illustrating of the history of Earth in the first book, and of human culture in the second book.
There are numerous US presidents who've had profound medical issues. The range is described well at http://www.doctorzebra.com/prez/t_roster.htm, and it's enlightening. Considering that they're middle aged or older, and the stress of the kind of active life necessary to campaign for president, it's not surprising that they had illnesses or traumas that were at the limit of their era's abilities to treat.
Expecting a presidential candidate to weigh every decision in the light of "what will the voters think" is, itself, political suicide at the highet levels. It will mark a candidate as incapable of making a policy.
All of the blind people who use public restrooms, perhaps? Have you ever tried to find the toilet paper in a public restroom when the bulb has failed, or when the light switch is difficult to find?
This is not only true of software development: it's true in IT as well. Far too often visiting with corporate partners, I see racks of equipment wired like a cat's cradle, where the engineer on site says "oh, we're not very neat" and I see core servers at risk from unlabeled wires, power and cabling that are out of spec and likely to fail under load, and where expensive system redundancy is wasted because the matched pair of servers are plugged into the same switch and the same power supply and the same storage array.
I'm afraid to say that I've become known as an uholy terror about this, because I or a colleague I can trust photograph the racks, bring boxes of cables and appropriate ties and labelmakers, and demand access to their inventory systems and network diagrams to record _all_ the systems. The result is, far too often, political furor as people discover that the much vaunted "high availability" systems were absolutely _not_ high availability. Far too often, fiber optic cables are bent too sharply and unprotected from tangled cables, expensive smart UPS's and switches have never had their configurations recorded, and redundant systems and systems have never been configured for proper failover behavior. And often we discore absolutely mission critical systems that are not on any network diagram or inventory that were ignored in the project plan.
Really, we're as prone to the problems in IT as the devlopers are.
There are problems with our original poster's original claims. Wireless, for example, is very useful for traveling devices and remote smartphones. But safe such access is dependent on VPN or other encryption technologies. But the amount of electromagnetic noise from all the wireless connections is rapidly approaching saturation in crowded locations, and there aren't enough big frequency ranges that the FCC might release to resolve this in the USA. Wired will remain critical for bandwidth and security in the actual working network and systems environments in which we, as engineers, spend so much of our time.
Similarly, much of what is outsourced to the could and _should_ be outsourced to a competent shop. A 20 person company has no business wasting their resources hiring, and keeping trained, someone able to manage core DNS, backup, email, and calendar systems when such cheap and reliable "cloud" systems are available. The services are too important, and too vulnerable to error, for everybody in even "a company full of geeks" to be trusted to run it reliably. I'm afraid I've seen far too many environments where every developer thought themselves capable of running the core IT services, and the results were so uneven and patchy that it was quite unstable.
I _do_ see the Linux transformation of core servers as an ongoing process. The purchase of Sun by Oracle is the death knell for Solaris, and HP is abandoning HP-UX. While Apple has effectively taken over FreeBSD, they've also effectively taken it closed source, and far more storage devices, plug-in email appliances such as database servers and mini-computer based video boxes and smartphones are being Linux based. Even most MS Exchange based shops use a Linux system in front to handle spam filtering. If Samba 4 can ever get out of alpha testing, I expect a huge array of Windows based network storage systems to be replaced nearly overnight by more stable and less expensive Samba based systems.
MS Word will continue to dominate document writing: Powerpoint will continue to dominate presentations. While the open source tools will continue to improve, Microsoft will _again_ manipulate their undocumented and inconsistent API's to break compatibility.
Virtualization is going to get even more interesting. I can't quite picture which way it's going to go, but I do anticipate that a lot of legacy hardware preservation, keeping old systems alive in a rack so we can fall back to recover data or just in case the new system fails, will be replaced by virtualization. And _those_ go to the cloud very well.
Last, I strongly suspect that the migration to IPv6 will still be less than 50% of all devices, probably less than 30%. The benefits of IPv6 simply do not matter to most environments, who are better off in security and network management terms using NAT and thus have no internal need for IPv6. The result will be growth of "mixed-stack" solutions, but until there is reason to leave IPv4 for systems already configured, the "hysteresis" of remaining consistent with existing infrastructure will preclude wholesale migration.
Oh, my I must disagree. Carly started a set of corporate disasters that are still dragging down HP, ranging from their purchase of Compaq (which was an amazing mismatch with HP's former reputation for quality hardware in many fields, and the loss of manufacturing quality by preserving any products or personnal from Compaq is still hurting HP).
There is a good article at http://www.businesspundit.com/10-reasons-people-hate-carly-fiorina/. It points out, correctly, that HP made money _despite_ Carly's misguided decisions, not because of them: HP's printer business continued to be quite profitable for reasons that were in place before Carly's reign.
Currrently, yes. There is no punishment, and in general only modest engineering cost to setting up a new spam net. This encourages new "entrepreneurs" to enter the field, even if they make no overall profit doing so. Spam services are being _sold_ to legitimate and illegitimate clients, and the claims of profit are overblown. But since no one publishes good numbers on its success rates, they can continue lying and drumming up business to fools and criminals.
The return on investment need not be real: it only needs to be portrayed as real to get customers of the wholesale spamming services, which are such a large proportion of modern spam.
Good point. I do think that booting from SAN is a better defined problem, and coreboot is working well with "Etherboot", which hints that network based booting is working well with it. I've not personally tried doing fiber channel booting with it, but it is in use on a lot of small customized network applicances.
It's a great solution long, long before Linux takes over the server and desktop market. It's already in use on the OLPC project, very successfully: the fear of moving off the old and very painful existing BIOS's is more one of familiarity and a maze of intellectual property rights for existing firmware, rather than a technological or performance justification for using the existing systems.
The RAM checks on a new server are going to take quite some time, and should really be allowed to run to completion on a newly installed server. Memory can be jostled or twisted, and some installers will buy cheap third party memory and fail to seat it properly or even buy the wrong models of RAM. Some of them will even buy slower, cheaper RAM and not tell the purchaser, which takes checking to discover.
Actually, they need not take so long. The entire BIOS configuration toolkit has become a nightmare of workarounds and scans for old components that don't even exist, and it's a proprietary software and licensing nightmare to try to tune. The "coreboot" toolkit, formerly known as "linuxbios", has turned a lot of that into genuine GPL freeware. The code is legible, easier to update, much cleaner, and vastly faster. It's awkard to get the initial compatible BIOS loaded, but once it's available for a motherboard, the performance, speed, and ability to _report_ the state of the hardware and BIOS to normal system utilities is well worth the invested effort.
I do wish the motherboard manufacturers would complete their investigations of this technology and switfch wholesale. The code has matured and documentation is now very useful to those building custom applications, and it is in effective commercial use.