FTP is fine for suitable purposes, which limits it to:
Anonymous upload directory (where downloads are prohibited), and anonymous downloads (public dissemination of information). Being a lowest common denominator, your clients don't need special software to push files to the uploads directory or to download files.
But beyond that, you should be using HTTPS pages or SCP or something else...
Well, my observation is that older Pentium 4 systems are kind of slow, even for simple web browsing. The issue is not with the CPU, but that a lot of PCs from that era come with little memory, like 250-500MB. I have seen a few friends neighbors upgrade because of that. But PCs from the time when Core 2 was around normally came with more memory, enough for office productivity or web browser.
It's a little of both really. I had a single-core P4 laptop (2GHz?) with 1GB of RAM and Windows XP on it. It was still slow to respond when doing stuff that hammered the CPU, because the CPU was just too busy and XP was lousy at giving the UI priority.
Yet I still use my Thinkpad T61s on a daily basis for work (Core 2 Duo, 2.2GHz, 8GB RAM, Win7 Pro). That second core makes a big difference in how responsive the machine feels. And for multi-taskers who like to have multiple things running at once, a quad-core CPU is not an extravagance.
The "fast enough for 99% of office work" came when the dual-core CPUs dropped to the $100 point. That was around 2007 and it was easy to get a XP machine with 2GB.
Once you have dual (or quad) core CPUs and 4-8GB of RAM, the next natural upgrade is SSDs. Which is the other magic ingredient that I installed into my 2007 era laptop to keep it going. The SSD makes it so that the CPU is the bottleneck and I can multi-task to my heart's content rather then twiddling my thumbs for minutes at a time as some slow program starts up. Or if I start something disk-intensive, I can still get other tasks on the machine done while that process runs.
The most important thing for drives is proper cooling and avoiding manufacturing defects (bad batch).
I'd put "power-supply quality" to the list as #2. There's not much control over bad batches unless you spread your drive purchases for an array out over a period of weeks and source from multiple vendors. Or have a vendor who will pick/pack from multiple lots when filling your order.
The big issue, as you discovered, is that a RAID-5 rebuild puts a lot of stress on the remaining disks (RAID-6 has similar issues).
I much prefer RAID-10 (which is RAID-0 over RAID-1 pairs) because when you have to replace a drive, the rebuild only impacts the drives in that RAID-1 pairing. The rebuild time is related to the size of an individual drive, not the overall size of an array. A 12-disk RAID-10 array of 600GB drives takes just as long to rebuild as a 24-disk RAID-10 array of 600GB drives. Whereas a 24-disk RAID-5 array of 600GB drives takes at least 2x as long to rebuild as a 12-disk RAID-5 array of 600GB drives.
(Sometimes, there's no choice but to use RAID-6. You just need to understand the failure modes and that it is not as forgiving.)
The bigger issue is that EVE has (over the past 2-3 years) become "bring titans / supercaps or go home". A budding organization who wants to try and compete in low-sec? They'll get blobbed by super-caps and titans.
CCP really needs to exclude super-caps and titans from low-sec. Which would make low-sec the playground of capitals and smaller. (Titans, super-caps, and caps are already excluded from hi-sec, so this would be a natural progression.)
As beings raised in a mostly 2 dimensional plane, it's natural for a truly 3-dimensional no-gravity-bias large-scale interaction to bewilder us. I think this might be one of the things EVE got right.
Nope. EVE gets 3-d wrong. There is a gravity-bias in that all ships will by default orient "upwards" (like ships on the ocean surface) unless they are moving or pointing at something.
If you turn off your engines in EVE, you stop moving.
The downside of that is really nasty passwords, with changes of case, numbers and symbols end up written on Post-it notes and stuck on the fronts of computers.
That's a social problem - one best solved via social means. Like disciplinary action via management.
On the training side, we tell our users that it's fine to write passwords down in case they forget, but they need to be kept secure. Which means keeping them in your wallet next to your money, or in a safe-deposit box, or a lockbox.
The important flip-side of the issue is that we don't force users to change passwords every 30 days. Which means that once they learn it (takes 1-2 weeks on average if they use it daily), they don't have to worry that it will change 30 days later. So they get validation for memorizing it, their life gets easier and is not made arbitrarily more difficult every 30 days.
If I find the service useful, Firefox will remember my email and password for login and/or I'll be able to recover the password using their system.
If you are going to let Firefox remember the password for it anyway -- then why not come up with some random gibberish for your password in the first place?
tr -dc [:alnum:]
Which has the advantage of 62 possibles per digit instead of only 16 possibles per password digit. For an 8-character password, using 62 possibles per digit instead of only 16 is about 50,000x more secure.
Five platter drives have typically run hotter then normal and consume more power then a four platter drives. Heat is probably the top killer of hard-drives, closely followed by power issues.
It's best not to use 5-platter drives in a situation where the room can't be kept below 70-75F and/or where the drives have iffy airflow.
This is why you just buy whatever is cheap and rig up a RAID 5. A drive craps out and you throw another one in and keep on going.
RAID-5 on a system where you don't think you have high quality drives, a high quality power supply, battery backup (for the RAID card) plus a high quality UPS unit (preferably multiple UPS units hooked up to a set of redundant PSUs inside your case) -- is simply a bad idea. Sooner or later, you *will* lose the array due to a double-drive failure. Oh and make sure that you have plans to swap out drives on a regular basis and have a working backup plan.
RAID-6 is better, but not by much. It can at least deal with a double drive failure. But performance still goes in the gutter while it's degraded and/or rebuilding.
One of the more fault-tolerant setups is a 3-way RAID-1 mirror where you can lose 2 of 3 drives without losing data. The downside is that it is only 33% efficient while RAID-6 (1 spare, 2 parity, 5 data) is 62% efficient. A well configured RAID-10 setup also works well, but never gets much above 40-45% space efficiency if you set aside a hot spare for it.
Main reason why I prefer RAID-10 for larger arrays is that the time to rebuild a failed disk is linear to the size of a single disk within the array (because you have mirror pairs). With RAID-5 / RAID-6 the rebuild time scales with the total size of the array. For a 15-20 drive array, that means RAID-10 could rebuild the failed drive in 1/5 to 1/10 the time of the RAID-5 or RAID-6 array with the same number of spindles.
It's not Moore's Law -- but the problem that there is only so much performance you can get out of a single CPU core (due to heat / energy loss) and most programs are constrained to a single thread of execution.
Single-core performance has gone from doubling every 15-18 months to taking 8-12 years to double. That means machines from the mid-2000s are only 30-40% slower at single-threaded tasks then modern machines.
The only change-up has been the introduction of inexpensive dual/quad core CPUs which allow the O/S to spread the workload over more CPUs. This *does* have an effect that the O/S seems more responsive, but once you get into multi-core land, very few users are going to notice the difference between dual/quad/hex/octo core setups.
Multi-threaded programming is slowly becoming more common, but there's just a lot of programming issues where it's not easy to split your workload over multiple threads.
Moore's Law lets us put more CPU cores on the chip, but it's the heat/energy loss issues that are putting a limit on per-core performance.
Except for RAM, the vast majority of PC users will never fully max out their machine. They won't even get close to what the CPU can do. Even 10 years ago when someone asked me what kind of PC they should buy, I would tell them to buy the oldest machine they can find with twice as much memory as they think they need -- because in my experience, lots of RAM contributes more to the longevity of a machine than loads of CPU.
That was not very good advice 10 years ago. While true that OEMs typically sold you a minimum of RAM, hobbling your customers by telling them to buy the oldest machine in the shop was poor advice.
Modern advice is that you need at least a dual-core CPU (makes the O/S much more responsive) along with lots of RAM. For XP, my recommendation was a minimum of 2GB and once RAM got cheaper, 4GB. For Win7, a realistic minimum is 4GB, but 8GB is not that expensive and will work better long-term.
The drop in SSD prices also means that consumer SSDs are a strong recommendation for the primary O/S drive. Even on an older Vista/Win7 machine (that is at least dual-core), dropping in a SSD can breath new life into a machine that seems too slow to be useful. My Thinkpad T61p from 2007 is still a useful machine because it has Win7, 8GB RAM, dual-core CPU and a SSD.
Yeah, sure, OK, but that's why the time to start all this lengthy work was like 4 years ago, right?
Right, which industry do you work in again? Because here in the real world, we're struggling with recessions, budget cuts and trying to make payroll on a month to month basis. And that's after laying off 1/4 of our workers due to a drop in business.
With Microsoft giving in even a little bit, I think the big corps are going to put enough pressure on them to extend XP's patches another year.
Really? That 3.5GB RAM limit was a bit of a nuisance for some specific things, but realistically how many office computers ever run up against it?
Look around for your power users / multi-taskers in the office. They probably have their email open all day (and if it's Thunderbird, it's not light on the memory footprint for packrats). Plus web browser, plus word, excel, plus some sort of instant message tool for internal communication, plus some job-specific stuff.
They are quite likely using 2.0-2.5GB of RAM under XP. That will only increase over time. Putting 8GB of RAM and Win7 64bit on their desks is not really overkill any more. If it's DDR3, that 8GB RAM is probably the same cost as 4GB of DDR2. And you won't have to worry about bumping up their RAM if something changes down the road.
Developers? Even the most basic developers should be bumping up the RAM. And that's even if they aren't running VMs, multiple IDEs, some background servers or what have you.
As a long time AMD fan (if we whitebox build, it's always an AMD chip), I have to say "it depends".
For a lot of applications, per-core performance is what matters. And for the last few years, Intel beats AMD hands-down on per-core performance. As in 30-50% faster. That i3 for $200 is going to run rings around the AMD for $200. For a lot of single-threaded programs (many games are CPU-bound by a single thread), that 30-50% faster speed matters.
However, if your application is multi-threaded and the problem you are trying to solve (media transcoding) is easily done in parallel, then the AMD chips are a better fit.
The "Bulldozer" architecture was a dud. Lots of cores for cheap, but low performance per core under a lot of workloads. The Piledriver architecture is better and AMD is at least somewhat competitive again.
I'm very curious to see how well the new Steamroller (Kaveri) series chips perform.
I wouldn't lump C# in with VB.NET. C# is a lot closer to Java and they both hit about the same market segment. Both are equally capable for large projects.
The main advantage of Java over C# is cross-platform compatibility for your business logic / business objects. It's the COBOL of the next 20-30 years. A good read is "Java: The Good Parts" (published in 2010 by O'Reilly).
Use C# and you're stuck running on Microsoft software stacks (or *maybe* Mono, if you are brave).
We spent 2 years looking at various languages as we moved off VB6/VBScript/ASP. Locking ourselves into C# was not an option. PHP and a lot of the other languages fail to scale past 100k lines of code. Ultimately settling in on Java / Spring for anything that was going to be more then 5k lines of code.
Having had to write my own SysV init scripts... I can't say that I'll be sorry to see them go. There's a lot of corner cases that they don't handle well (dependencies on another service are hard) and you constantly have to roll-your-own solution to common problems.
It's the difference between building a modern website using Java servlets (JSP) files vs a modern framework like Spring.
Yes, you can do very quick creations and tweaks of existing SysV init files, but as soon as you need to do anything more complex then start a standalone service, it gets hairy.
Still not sure systemd is the solution... but will try it in RHEL 7.
From the (mistaken? wise?) use of a.300 in an IPv4 address
Wise. It's like using phone numbers that contain "555". Prevents your fans from DDoS'ing some innocent IPv4 address that actually is in use (or someone's phone number).
In general, when the IPv4 or phone numbers are valid, it's probably a PR/Marketing thing to drive traffic to a website or to get you to call for a special offer.
I started programming in the late 80s, went professional / full-time after about 15 years of hobbyist level programming (COBOL, Fortran77, Pascal, CA-Clipper / DBase III/IV, C, C++, VB, Java, JavaScript).
OOP with the concepts of public functions and private variables is a big step forward in terms of the compiler being able to check your work. For anything where you are passing around complex business objects, some sort of OOP is necessary.
In the Java world, there are add-ons (Spring Roo) which simplify the definition of "business" objects. It handles a lot of the "grunt" work of defining plain old java objects (POJOs) by using AspectJ. With AspectJ being able to inject methods rather then defining them in your source code, you gain a lot of simplicity in your business objects. So that bit of CASE has come true, but at a limited scale.
Frameworks are another step forward. For many problem spaces, there are huge amounts of setup/teardown and communication needed to get stuff done. Having a good framework lets you abstract a lot of that out of your systems, letting you focus more on the business objects.
The open source movement and the advent of open source frameworks, languages, compilers, editors, and IDEs has also been a big paradigm shift. In order to build software for OS/2, I had to purchase something like a $300-$500 compiler, plus manuals, plus a UI framework. Now I can get a full fledged IDE in the form of NetBeans, Eclipse (or SpringSource Tool Suite), or a few others.
I don't miss the days of COBOL and Fortran77. Things have definitely improved over the decades.
Slashdot Mirror
FTP is fine for suitable purposes, which limits it to:
Anonymous upload directory (where downloads are prohibited), and anonymous downloads (public dissemination of information). Being a lowest common denominator, your clients don't need special software to push files to the uploads directory or to download files.
But beyond that, you should be using HTTPS pages or SCP or something else...
Well, my observation is that older Pentium 4 systems are kind of slow, even for simple web browsing. The issue is not with the CPU, but that a lot of PCs from that era come with little memory, like 250-500MB. I have seen a few friends neighbors upgrade because of that. But PCs from the time when Core 2 was around normally came with more memory, enough for office productivity or web browser.
It's a little of both really. I had a single-core P4 laptop (2GHz?) with 1GB of RAM and Windows XP on it. It was still slow to respond when doing stuff that hammered the CPU, because the CPU was just too busy and XP was lousy at giving the UI priority.
Yet I still use my Thinkpad T61s on a daily basis for work (Core 2 Duo, 2.2GHz, 8GB RAM, Win7 Pro). That second core makes a big difference in how responsive the machine feels. And for multi-taskers who like to have multiple things running at once, a quad-core CPU is not an extravagance.
The "fast enough for 99% of office work" came when the dual-core CPUs dropped to the $100 point. That was around 2007 and it was easy to get a XP machine with 2GB.
Once you have dual (or quad) core CPUs and 4-8GB of RAM, the next natural upgrade is SSDs. Which is the other magic ingredient that I installed into my 2007 era laptop to keep it going. The SSD makes it so that the CPU is the bottleneck and I can multi-task to my heart's content rather then twiddling my thumbs for minutes at a time as some slow program starts up. Or if I start something disk-intensive, I can still get other tasks on the machine done while that process runs.
The big advantage Yahoo! Mail had back around 2000-2005 was the large mailbox size, a larger limit on attachment size, etc.
This was back when ISPs only gave you a 2-10MB mailbox and limited attachment sizes to as little as 1MB.
Then GMail hit with their 1GB mailbox, all the other free webmail providers had to up their limits.
The most important thing for drives is proper cooling and avoiding manufacturing defects (bad batch).
I'd put "power-supply quality" to the list as #2. There's not much control over bad batches unless you spread your drive purchases for an array out over a period of weeks and source from multiple vendors. Or have a vendor who will pick/pack from multiple lots when filling your order.
The big issue, as you discovered, is that a RAID-5 rebuild puts a lot of stress on the remaining disks (RAID-6 has similar issues).
I much prefer RAID-10 (which is RAID-0 over RAID-1 pairs) because when you have to replace a drive, the rebuild only impacts the drives in that RAID-1 pairing. The rebuild time is related to the size of an individual drive, not the overall size of an array. A 12-disk RAID-10 array of 600GB drives takes just as long to rebuild as a 24-disk RAID-10 array of 600GB drives. Whereas a 24-disk RAID-5 array of 600GB drives takes at least 2x as long to rebuild as a 12-disk RAID-5 array of 600GB drives.
(Sometimes, there's no choice but to use RAID-6. You just need to understand the failure modes and that it is not as forgiving.)
Call me when LTO-6 drives are under $300 and the tapes are under $20 for 2TB of storage.
Right now, it's $2700-$3300 for LTO-6 tape drives and $65-$75 per tape.
The bigger issue is that EVE has (over the past 2-3 years) become "bring titans / supercaps or go home". A budding organization who wants to try and compete in low-sec? They'll get blobbed by super-caps and titans.
CCP really needs to exclude super-caps and titans from low-sec. Which would make low-sec the playground of capitals and smaller. (Titans, super-caps, and caps are already excluded from hi-sec, so this would be a natural progression.)
As beings raised in a mostly 2 dimensional plane, it's natural for a truly 3-dimensional no-gravity-bias large-scale interaction to bewilder us. I think this might be one of the things EVE got right.
Nope. EVE gets 3-d wrong. There is a gravity-bias in that all ships will by default orient "upwards" (like ships on the ocean surface) unless they are moving or pointing at something.
If you turn off your engines in EVE, you stop moving.
The downside of that is really nasty passwords, with changes of case, numbers and symbols end up written on Post-it notes and stuck on the fronts of computers.
That's a social problem - one best solved via social means. Like disciplinary action via management.
On the training side, we tell our users that it's fine to write passwords down in case they forget, but they need to be kept secure. Which means keeping them in your wallet next to your money, or in a safe-deposit box, or a lockbox.
The important flip-side of the issue is that we don't force users to change passwords every 30 days. Which means that once they learn it (takes 1-2 weeks on average if they use it daily), they don't have to worry that it will change 30 days later. So they get validation for memorizing it, their life gets easier and is not made arbitrarily more difficult every 30 days.
If I find the service useful, Firefox will remember my email and password for login and/or I'll be able to recover the password using their system.
/dev/urandom | head -c${1:-20};
If you are going to let Firefox remember the password for it anyway -- then why not come up with some random gibberish for your password in the first place?
For example: Extended Password Generator. Or putting the following shell script in your ~/.bashrc file:
passgen ()
{
tr -dc [:alnum:] <
echo
}
Better is:
tr -dc [:alnum:]
Which has the advantage of 62 possibles per digit instead of only 16 possibles per password digit. For an 8-character password, using 62 possibles per digit instead of only 16 is about 50,000x more secure.
Five platter drives have typically run hotter then normal and consume more power then a four platter drives. Heat is probably the top killer of hard-drives, closely followed by power issues.
It's best not to use 5-platter drives in a situation where the room can't be kept below 70-75F and/or where the drives have iffy airflow.
This is why you just buy whatever is cheap and rig up a RAID 5. A drive craps out and you throw another one in and keep on going.
RAID-5 on a system where you don't think you have high quality drives, a high quality power supply, battery backup (for the RAID card) plus a high quality UPS unit (preferably multiple UPS units hooked up to a set of redundant PSUs inside your case) -- is simply a bad idea. Sooner or later, you *will* lose the array due to a double-drive failure. Oh and make sure that you have plans to swap out drives on a regular basis and have a working backup plan.
RAID-6 is better, but not by much. It can at least deal with a double drive failure. But performance still goes in the gutter while it's degraded and/or rebuilding.
One of the more fault-tolerant setups is a 3-way RAID-1 mirror where you can lose 2 of 3 drives without losing data. The downside is that it is only 33% efficient while RAID-6 (1 spare, 2 parity, 5 data) is 62% efficient. A well configured RAID-10 setup also works well, but never gets much above 40-45% space efficiency if you set aside a hot spare for it.
Main reason why I prefer RAID-10 for larger arrays is that the time to rebuild a failed disk is linear to the size of a single disk within the array (because you have mirror pairs). With RAID-5 / RAID-6 the rebuild time scales with the total size of the array. For a 15-20 drive array, that means RAID-10 could rebuild the failed drive in 1/5 to 1/10 the time of the RAID-5 or RAID-6 array with the same number of spindles.
It's not Moore's Law -- but the problem that there is only so much performance you can get out of a single CPU core (due to heat / energy loss) and most programs are constrained to a single thread of execution.
Single-core performance has gone from doubling every 15-18 months to taking 8-12 years to double. That means machines from the mid-2000s are only 30-40% slower at single-threaded tasks then modern machines.
The only change-up has been the introduction of inexpensive dual/quad core CPUs which allow the O/S to spread the workload over more CPUs. This *does* have an effect that the O/S seems more responsive, but once you get into multi-core land, very few users are going to notice the difference between dual/quad/hex/octo core setups.
Multi-threaded programming is slowly becoming more common, but there's just a lot of programming issues where it's not easy to split your workload over multiple threads.
Moore's Law lets us put more CPU cores on the chip, but it's the heat/energy loss issues that are putting a limit on per-core performance.
Except for RAM, the vast majority of PC users will never fully max out their machine. They won't even get close to what the CPU can do. Even 10 years ago when someone asked me what kind of PC they should buy, I would tell them to buy the oldest machine they can find with twice as much memory as they think they need -- because in my experience, lots of RAM contributes more to the longevity of a machine than loads of CPU.
That was not very good advice 10 years ago. While true that OEMs typically sold you a minimum of RAM, hobbling your customers by telling them to buy the oldest machine in the shop was poor advice.
Modern advice is that you need at least a dual-core CPU (makes the O/S much more responsive) along with lots of RAM. For XP, my recommendation was a minimum of 2GB and once RAM got cheaper, 4GB. For Win7, a realistic minimum is 4GB, but 8GB is not that expensive and will work better long-term.
The drop in SSD prices also means that consumer SSDs are a strong recommendation for the primary O/S drive. Even on an older Vista/Win7 machine (that is at least dual-core), dropping in a SSD can breath new life into a machine that seems too slow to be useful. My Thinkpad T61p from 2007 is still a useful machine because it has Win7, 8GB RAM, dual-core CPU and a SSD.
Yeah, sure, OK, but that's why the time to start all this lengthy work was like 4 years ago, right?
Right, which industry do you work in again? Because here in the real world, we're struggling with recessions, budget cuts and trying to make payroll on a month to month basis. And that's after laying off 1/4 of our workers due to a drop in business.
With Microsoft giving in even a little bit, I think the big corps are going to put enough pressure on them to extend XP's patches another year.
Really? That 3.5GB RAM limit was a bit of a nuisance for some specific things, but realistically how many office computers ever run up against it?
Look around for your power users / multi-taskers in the office. They probably have their email open all day (and if it's Thunderbird, it's not light on the memory footprint for packrats). Plus web browser, plus word, excel, plus some sort of instant message tool for internal communication, plus some job-specific stuff.
They are quite likely using 2.0-2.5GB of RAM under XP. That will only increase over time. Putting 8GB of RAM and Win7 64bit on their desks is not really overkill any more. If it's DDR3, that 8GB RAM is probably the same cost as 4GB of DDR2. And you won't have to worry about bumping up their RAM if something changes down the road.
Developers? Even the most basic developers should be bumping up the RAM. And that's even if they aren't running VMs, multiple IDEs, some background servers or what have you.
As a long time AMD fan (if we whitebox build, it's always an AMD chip), I have to say "it depends".
For a lot of applications, per-core performance is what matters. And for the last few years, Intel beats AMD hands-down on per-core performance. As in 30-50% faster. That i3 for $200 is going to run rings around the AMD for $200. For a lot of single-threaded programs (many games are CPU-bound by a single thread), that 30-50% faster speed matters.
However, if your application is multi-threaded and the problem you are trying to solve (media transcoding) is easily done in parallel, then the AMD chips are a better fit.
The "Bulldozer" architecture was a dud. Lots of cores for cheap, but low performance per core under a lot of workloads. The Piledriver architecture is better and AMD is at least somewhat competitive again.
I'm very curious to see how well the new Steamroller (Kaveri) series chips perform.
Sun started selling x86 boxes back around 2005 or so. We have a pair of SunFire X2100 units using AMD Opteron chips inside.
So unless he specifically stuck with SPARC hardware, he was probably running Solaris on x86.
I wouldn't lump C# in with VB.NET. C# is a lot closer to Java and they both hit about the same market segment. Both are equally capable for large projects.
The main advantage of Java over C# is cross-platform compatibility for your business logic / business objects. It's the COBOL of the next 20-30 years. A good read is "Java: The Good Parts" (published in 2010 by O'Reilly).
Use C# and you're stuck running on Microsoft software stacks (or *maybe* Mono, if you are brave).
We spent 2 years looking at various languages as we moved off VB6/VBScript/ASP. Locking ourselves into C# was not an option. PHP and a lot of the other languages fail to scale past 100k lines of code. Ultimately settling in on Java / Spring for anything that was going to be more then 5k lines of code.
Having had to write my own SysV init scripts... I can't say that I'll be sorry to see them go. There's a lot of corner cases that they don't handle well (dependencies on another service are hard) and you constantly have to roll-your-own solution to common problems.
It's the difference between building a modern website using Java servlets (JSP) files vs a modern framework like Spring.
Yes, you can do very quick creations and tweaks of existing SysV init files, but as soon as you need to do anything more complex then start a standalone service, it gets hairy.
Still not sure systemd is the solution... but will try it in RHEL 7.
You should instead wait and move to CentOS 7 (or Sci Linux 7), which will be based on FC 19 or 20 and will be stable for 4-5 years.
Putting anyone on Fedora who isn't willing to rebuild their box every 1-2 years is a fool's errand.
The 172.16.0.0 to 172.30.255.255 range is even more obscure (and still allowed for private internet addressing).
Also, on some cable / mobile phone numbers, you get a 10.x.y.z address via DHCP and have to NAT to get to the outside world.
From the (mistaken? wise?) use of a .300 in an IPv4 address
Wise. It's like using phone numbers that contain "555". Prevents your fans from DDoS'ing some innocent IPv4 address that actually is in use (or someone's phone number).
In general, when the IPv4 or phone numbers are valid, it's probably a PR/Marketing thing to drive traffic to a website or to get you to call for a special offer.
I started programming in the late 80s, went professional / full-time after about 15 years of hobbyist level programming (COBOL, Fortran77, Pascal, CA-Clipper / DBase III/IV, C, C++, VB, Java, JavaScript).
OOP with the concepts of public functions and private variables is a big step forward in terms of the compiler being able to check your work. For anything where you are passing around complex business objects, some sort of OOP is necessary.
In the Java world, there are add-ons (Spring Roo) which simplify the definition of "business" objects. It handles a lot of the "grunt" work of defining plain old java objects (POJOs) by using AspectJ. With AspectJ being able to inject methods rather then defining them in your source code, you gain a lot of simplicity in your business objects. So that bit of CASE has come true, but at a limited scale.
Frameworks are another step forward. For many problem spaces, there are huge amounts of setup/teardown and communication needed to get stuff done. Having a good framework lets you abstract a lot of that out of your systems, letting you focus more on the business objects.
The open source movement and the advent of open source frameworks, languages, compilers, editors, and IDEs has also been a big paradigm shift. In order to build software for OS/2, I had to purchase something like a $300-$500 compiler, plus manuals, plus a UI framework. Now I can get a full fledged IDE in the form of NetBeans, Eclipse (or SpringSource Tool Suite), or a few others.
I don't miss the days of COBOL and Fortran77. Things have definitely improved over the decades.