They have specifically stated that they don't want to give the ability to change skills out-of-game, as that would remove one of the incentives for logging in.
Okay, step back and consider what they said.
One of the big reasons to login is to change skills? Not to have fun? Or run a mission? Or engage in combat? That feels like clear evidence of "vision over reality" where you put the customers through a PITA procedure just to maintain a pure vision.
(I have a love/hate relationship with EVE Online; played for about 18 months. There are some really wild and neat things in it, but I got tired of logging in, spending 3 hours at a gate-camp, and seeing maybe 10 minutes of action.)
If I could have changed skills out-of-game, or queued up a week's worth at a time, I might still have been a subscriber. Which is like the perfect customer for a company like CCP. Someone who pays their monthly fee, but barely puts any load on the servers.
In cases like that... never trust end-users to notify you of problems that can cause data loss if left unfixed.
On Linux, go with a simple bash script to monitor the arrays and e-mail / page you when they go bad. Then learn how to use Nagios or other monitoring software as a primary notification system. (The bash script then becomes a backup system for times when the monitoring software is having other issues.)
That said, the real problem here is why legitimate sites are service up the pop-under ads for antivirus 2009. Ad networks need to start vetting their clients. People should just start blocking all ads as a security threat.
A lot of these injections start as some sort of code-injection on legitimate sites. The criminals either do SQL injection attacks or break in via poor FTP username/passwords. They then append a bit of javascript to the pages that does the dirty deed. Or other attack vectors (Flash exploits, PDF exploits). So it's not always an infected ad (although those do exist and are becoming more frequent).
And if the Windows user is running as admin, the game is essentially over at that point and they've been infected.
Firefox combined with NoScript and FlashBlock cures a lot of these ills, but requires the end-user to maintain a whitelist of trusted sites. And to be sensible and cautious about adding sites to that whitelist. So if one of the sites on your whitelist is infected, you're still screwed.
Fortunately, it's moderately easy nowadays to run XP in limited-user mode. The vast majority of programs work just fine once you get them installed. Most viruses that do manage to infect the user can't do permanent infections, unless they use a local exploit to do privilege escalation. So a reboot kills off the running processes, and the malware becomes trivial to remove.
Well, the big problem is that the lengthening of the mean solar day is not a constant rate. You can't predict when leap seconds will be needed in the future with any accuracy.
Thirdly, the current gen phenoms even include 65W parts at reduced clock. I would not doubt that, if not at launch, shortly thereafter lower-wattage parts will come out. Even if not, underclocking may be an option.
For regular desktops at work, we're becoming very enamored with the 45W AMD parts. Those PCs are almost absolutely silent, especially when under a desk (Antec Sonata II cases).
I'm even considering switching over to 2.5" laptop drives inside to save another 5-8W. And probably make the systems even quieter. But the drives would cost too much.
Have to run the numbers again... 100W of power savings for a 24x7 PC is around $150/year in these parts. But it's only about $35/year for a PC that's only turned on during work hours.
19/22" LCD monitors almost pay for themselves in 3-4 years now, compared to our old 17" CRTs.
I usually try to wait until I can quadruple my main specs for a reasonable outlay of cash, and I think the time may be right.
Unless you're willing to wait 5-8 years now... the days of quadrupling your main specs are gone. CPU performance no longer doubles every 15 months.
Beyond that... I have a Phenom X4 at home. It competes favorable with Intel products that would have cost about the same amount. (Quite frankly, I don't give a rat's ass about CPUs that cost more then $300.)
8GB is probably affordable on a single-socket board. The 2GB DIMMs are not that expensive (but nowhere near as cheap as the 1GB DIMMs). 16GB is probably too expensive until 2010.
With Vista as my only option, my plan was to stick with XP as long as humanly possible. I have my own volume-licensed copy of XP Pro, so it's a somewhat realistic plan. If Windows 7 proves to be as high-quality as the pundits claim, that might just be enough to make me leave XP.
That's our current plan. We pushed through a WinXP upgrade across the board just before Vista shipped (because we knew Vista was a dog). All of the machines are dual-core, 2GB RAM, and will probably last us well into the 2011-2012 time frame. We also tossed a few copies of XP in a drawer for builds that we didn't plan for.
Still, I'm hoping that Win7 is at least decent and useful like XP. Because as we get past 2010, it's going to be harder and harder to keep XP running. Skipping Vista was easy if you timed things right, but skipping two generations is going to be difficulty.
(So nice not to have the majority of the PCs still running Win98 with 32MB or 64MB of RAM... that was a real PITA.)
I've used both Linux and Windows for servers for a decade.
I think Windows 2008 "core" mode is going to be too little too late. The more time I spend working with Linux servers, the power of the command line, the "everything is a file" mindset of Unix/Linux, and the sheer openness of the underlying tech - the less certain I am that Windows makes a good server product.
At least, if you don't want to spend lots and lots of money on add-on packages.
Some of the high points that have made my job easier in the past year:
- bash scripting
- LVM (flexibility, when our expected disk layout changed)
- having SecureCRT keep log files of all my sessions (so I can go back and figure out what I did 3 months ago in a certain situation)
- using FSVS and SVN to keep track of all changes on the server
- plain text configuration files that can be diff'd, grepped, and version controlled
- SELinux, the powerful iptables firewall
- a good security track record
Even better, the things that I learn on Linux server admin transfers mostly intact over to Solaris, Unix and BSD. And even Mac OS X. Whereas the things I learn on Windows only apply to Windows.
(Sorry for being long-winded, but I've spent a lot of time at the Linux command line in the past month. And things like LVM have saved my bacon a few times in the past year, allowing us to reconfigure servers on the fly when we forecast incorrectly.)
Reading, playing with the tech, making things go boom... and then fixing them.
The old Microsoft Certified DBA exams weren't that difficult, and there was some good things in there that were not just Microsoft-specific. (I finished the MCDBA cert back in 1999/2000, I've never recertified since then.) But I've been mucking with databases since the DBase III / CA-Clipper days and I can generally get around in 3 or 4 different database packages.
Beyond that, start playing with at least 2 or 3 different database products and learn their quirks. It's that or pay someone. In-depth knowledge of database concepts (normalization, tables, indexes, views) and SQL language (joins, unions, subqueries) pays off in spades. At least the major vendors now have freebie versions that everyone can play with.
Unfortunately, I have zero desire to put up with corporate behavior, so I'm staying in a small company where I can be the DBA, the sysadmin, the lead programmer. So my current focus is PostgreSQL and figuring out how fast we can move away from MS SQL Server. (The answer there is "soon".)
Besides, with the prices of Blu-ray players and discs now dropping rapidly, why bother with a solution that could tie up your broadband connection for long periods of time and the picture quality still can't compare to a real 1080p-resolution Blu-ray disc?
With modern codecs, excellent video quality can be had for 2-3 Mbps for a 720p video. The 1080p is about 2.25x more pixels, but since codecs scale well, you could probably do an excellent quality for 4-5 Mbps. Even 3Mbps could probably do well for the majority of video at 1080p.
Frankly, at 4Mbps for 1080p, you're probably not going to be able to tell the difference between that and Blu-Ray.
I may be biased, because I live in an area where the local cable company provides 20Mbps service. So 2/3/4Mbps streams don't phase me at all.
Am I the only one that has found this? I purchased a new Thinkpad ~6 months ago and it's terrible for a variety of reasons...
(shrugs) I have a T61p that I love (it's about a year old now?). It replaced an old (c2002) Toshiba Tecra. My T61p is the 15" 1680x1050 display, 3GB of RAM, and a Centrino Duo CPU running WinXP Pro.
They're still very solid units.
I don't care much for some of the ThinkVantage software, but I've learned to put up with its quirks. (They really need to sink some funding into fixing up that software.)
Eh, we just got done (mostly) with a 3 year upgrade cycle. I was hoping to go faster, but a lot of it depended on how much cash flow we had at the time.
We started back in 2005(?) right about the time that dual-core CPUs were still $300, but AMD was about to drop the prices on the X2s to below $150. At that point in time, 1/2 to 3/4 of our desktops were still running Win98 (and were from 1998-2000 purchases). The rest were running WinXP. A lot of machines were P3s with 300-800MHz CPUs, and the WinXP boxes were from 2001-2003 with a single-CPU (1.8GHz) and 512MB to 1GB of RAM.
We're now up to 95%+, WinXP Pro, Office Pro 2003, AMD Athlon64 X2 (45W versions where possible), 2GB of RAM, and (2) 250GB+ HDs in RAID1. The motherboards are all AM2 socketed. Average cost per box was about $1150, including $430 for Office Pro and XP Pro. We cut corners where it made sense, but didn't low-ball the units.
We have a few copies of WinXP Pro socked away in a drawer for new builds. We can get Office 2003 via a volume license with Microsoft (we're actually buying Office 2007 in packs of 5, and getting downgrade keys). Worst case, we buy Vista Ultimate OEM and get downgrade rights to WinXP Pro.
Expected lifespan for these units is at least 5-8 years. So the units that we put into place in 2006 will be starting to get long in the tooth comes 2011-2013. At that point, we can either decide to swap out the CPUs for quad-cores, or drop another gigabyte of RAM in. Because we bought dual-core systems, we expect them to age very well and not seem "slow" like single-core units usually do after 2 years. So we might even get an easy 10 years out of them, assuming no component deaths (such as capacitors leaking/bursting).
Starting in 2010, I expect to be evaluating Windows 7 (we're skipping Vista). Those desktops will probably end up being quad-core, 4GB or 8GB RAM, RAID1, and will probably still end up costing us around $1100 per seat (including licenses). But we won't be upgrading in bulk again until around 2012 or later.
We've still got a few single-core XP machines. They're used by our less-demanding users, but I have no compunction about replacing them with dual-core boxes once they wear out.
Basically, we jumped on dual-core as soon as the CPUs dropped below $300. The advent of multi-core CPUs means that systems should have very long useful lifespans compared to the single-core CPUs of yesteryear.
Not any more. I'm searching for one currently, and $130-150 seems to be the cheapest, and those are from sketchy sources. Love to know if you have a cheaper line.
The only way to get XP legally now is either to sign a license agreement with Microsoft (you'll probably have to buy in quantities of 5?) or buy Vista Ultimate OEM and pester Microsoft for a downgrade key.
(Vista Ultimate is the only OEM version that has downgrade rights.)
Let's say that we start with (15) 500GB disks. Next year I start dropping in 1TB disks, one-by-one. At first, we only use 500GB on the 1TB disks. But once we drop the 15th disk in, we can create additional partitions on the drives, create new Software RAID arrays, and then extend LVM volume groups across the new arrays.
Which lets us double our data density, without having to backup lots of data, tear the server down, put new disks in, build an all new array, then restore all that data. (Although I would hope that modern hardware RAID card management would allow you to grow the array by changing out disks.)
We might even upgrade the disks in the array a second time once 2TB drives get below $100 each. Past that point, and I'd expect the server to be getting long in the tooth with a need to be replaced.
A lot of it is not having to deal with the vagaries of finding the drivers to setup hardware RAID on a Linux box. Or deal with strange management interfaces, whereas if we simply use the RAID controller in JBOD mode, we can simply admin the box using Linux Software RAID, no matter what controller we use. And we can move those disks to another box, using a different RAID controller, and still have it work.
Software RAID also allows you to slice up the spindles into strange and varied shapes. Such as mixing RAID6, RAID10, RAID1 across the same set of spindles, depending on your access patterns or sensitivity to data loss. I might assign the first half of 3 drives to RAID1, the first half of the next 7 drives to a RAID10 array with hot spare, and put a RAID6 across the first half of the last 5 drives. Then across the back half of all the drives, I can put a large, seldom-used RAID6 array for bulk storage.
A lot of failure modes nowadays seem to be RAID slices deciding to desync with the rest of the array. Or bitrot which gets discovered during a weekly resync of the array. If a slice is an entire disk, you could be dealing with a large amount of resync work. But if you divide each disk into 4 slices and setup 4 arrays, a desync event might only affect one out of the four arrays. Giving you faster rebuild times.
Eh, for $5000, the upper limit is probably around 15-spindles. Including the cost of the case ($600), a good motherboard and CPU and RAM ($600), a good 16-port SATA controller ($1000), and all of the drives (15 drives @ $130 each adds up quick). Okay, well, that's only about $4000.
Figure RAID-6, 1 hot spare, net useful space of (12) 1.5TB disks. Which is about 18TB (probably closer to 16.5TB once you factor in over-head and the million bytes vs megabytes conversion).
The problem with that is power consumption. Build your own, and you'll be burning a lot of power unnecessarily because it's overkill. Contrast that with the ReadyNas Duo, which I own, that pulls on average around 30-40W. Much better and green.
The modern AMD 45nm energy-efficient parts have a max TDP of 45W. Which means that they're going to run at a lot less then that when not under load. Couple that with the new 500GB SATA 2.5" laptop drives and you're looking at a regular desktop tower with similar power requirements as the ReadyNAS. Or at least something in the ballpark of 30-60W.
As much as I like the ReadyNAS units (which are Linux-based and in an pinch data can be recovered using a Linux machine), I don't care much for the proprietary power-supply or other parts that I can't pickup at the neighborhood computer parts store.
(I've got a tits-up ReadyNAS NV sitting here beside me. Motherboard is completely fried, probably due to a defective power-supply. The unit is out of warranty, since it was bought before Infrant got bought up by NETGEAR. Waiting on the client to decide whether to spend $950 for a new unit, or whether they want us to simply build a whitebox PC using one of the NAS-like distros mentioned here.)
software RAID... is woefully slow compared to even a slow hardware RAID
Wrong. Go do your homework.
Well, maybe... what I've found over the years is that RAID-0, RAID-1, RAID-10 all perform extremely well as Software RAID without taxing the CPU.
But RAID-5 and RAID-6 do tend to become CPU-bottlenecked given enough spindles.
So if you're going with RAID-5 or RAID-6, makes sure to purchase a CPU with a high single-core performance (Linux Software RAID is currently not multi-threaded enough to split the calculations across multiple cores yet for a single array). The last RAID-6 box that I built, we made the mistake of going with a 1.8GHz CPU instead of a 2.4GHz or 2.6GHz CPU, which cut our performance by about 1/4 to 1/3 of what it could have been.
The spindles are definitely not at 100% utilization, but you can see the software RAID process pegging one of the CPUs at 100%. (Fairly modern motherboard, PCIe-based, pair of Opteron dual-core, 8GB RAM, 15 SATA drives.)
RAID-10 would have been a better choice for performance. Six drives for the array, with one hot-spare on standby. Net space would only be around 700GB, but you'd get pretty good performance.
Long as the PCs have a license sticker on the machine such as Vista or higher they have the right to downgrade for free.
Not true. The only version of Vista that you're allowed to downgrade to XP is the Vista Premium with all the bells and whistles.
Not sure about the VLKs, I know that the only way to get MS Office downgrades is to go with a volume license program for Office 2007, which then gives you the rights to downgrade to 2003.
I actually wouldn't be surprised if user space stayed 32 bit (or mostly 32 bit) for a very long time. The one thing Intel got right with the 386 is that its protected mode allowed for a mixture of 16-bit and 32-bit program contexts. AMD continued this tradition with x86-64. It's possible to have a 64-bit kernel with 32-bit user space applications. (Indeed, I've been tempted to set up a Linux machine that way in the past--put a 64-bit kernel under a 32-bit install.)
There's a lot of modern apps that are already pushing 500-800MB of memory usage. It's not far-fetched to see apps eating up 2GB or more within a year or three. And if the program is limited to 3GB (like it is in the x86 32-bit space), then you're up against limits that you wouldn't have if user space was 64bit.
Money. HDD's will keep getting cheaper. I'm betting on 2010.
For bulk storage, yes, magnetic HDs are cheaper (and will be for a good while). But we're already at the point that 64GB SSDs are inexpensive enough to be a good option. And that will handle a small segment of user needs. It won't be long until the 128GB SSDs drop enough in price that those too are a good option.
Right now, the 128GB is about the tipping point size. The 64GB drives feel a bit small for most users, but 128GB is pretty equivalent to the 120GB drives that are common on modern laptops.
So I'm thinking mass adoption of 128GB SSDs in laptops by Fall 2009, with wider use of the 256GB drives in mid-2010.
And if they can make the leap to 512GB SSDs and get the price under $200 or $300, I think that will pretty much doom the use of rotational magnetic HDs in laptops. Except for people that absolutely have to have 1TB of capacity.
Another option is to set up a second SSH daemon on a different port, and which only allows logins using public key (and possibly only by a specific user.) If you get blocked out on port 22, you can just use this side door to get in, as long as you've got your key.
Got a link for setting up a 2nd SSH daemon?
(The best move against this attack is either to go with port-knocking or simply move your primary SSH port to a non-standard port. We've chosen the latter solution on all of our public facing servers, along with requiring public-key logins.)
Slashdot Mirror
They have specifically stated that they don't want to give the ability to change skills out-of-game, as that would remove one of the incentives for logging in.
Okay, step back and consider what they said.
One of the big reasons to login is to change skills? Not to have fun? Or run a mission? Or engage in combat? That feels like clear evidence of "vision over reality" where you put the customers through a PITA procedure just to maintain a pure vision.
(I have a love/hate relationship with EVE Online; played for about 18 months. There are some really wild and neat things in it, but I got tired of logging in, spending 3 hours at a gate-camp, and seeing maybe 10 minutes of action.)
If I could have changed skills out-of-game, or queued up a week's worth at a time, I might still have been a subscriber. Which is like the perfect customer for a company like CCP. Someone who pays their monthly fee, but barely puts any load on the servers.
In cases like that... never trust end-users to notify you of problems that can cause data loss if left unfixed.
On Linux, go with a simple bash script to monitor the arrays and e-mail / page you when they go bad. Then learn how to use Nagios or other monitoring software as a primary notification system. (The bash script then becomes a backup system for times when the monitoring software is having other issues.)
That said, the real problem here is why legitimate sites are service up the pop-under ads for antivirus 2009. Ad networks need to start vetting their clients. People should just start blocking all ads as a security threat.
A lot of these injections start as some sort of code-injection on legitimate sites. The criminals either do SQL injection attacks or break in via poor FTP username/passwords. They then append a bit of javascript to the pages that does the dirty deed. Or other attack vectors (Flash exploits, PDF exploits). So it's not always an infected ad (although those do exist and are becoming more frequent).
And if the Windows user is running as admin, the game is essentially over at that point and they've been infected.
Firefox combined with NoScript and FlashBlock cures a lot of these ills, but requires the end-user to maintain a whitelist of trusted sites. And to be sensible and cautious about adding sites to that whitelist. So if one of the sites on your whitelist is infected, you're still screwed.
Fortunately, it's moderately easy nowadays to run XP in limited-user mode. The vast majority of programs work just fine once you get them installed. Most viruses that do manage to infect the user can't do permanent infections, unless they use a local exploit to do privilege escalation. So a reboot kills off the running processes, and the malware becomes trivial to remove.
Dude, don't forget Thunderbird 3! At least, I *hope* it'll be out in 2009. *sigh*
I'd settle for a version of Thunderbird that doesn't regularly corrupt its folder indexes... (mostly folders that are from an IMAP server).
Well, the big problem is that the lengthening of the mean solar day is not a constant rate. You can't predict when leap seconds will be needed in the future with any accuracy.
Thirdly, the current gen phenoms even include 65W parts at reduced clock. I would not doubt that, if not at launch, shortly thereafter lower-wattage parts will come out. Even if not, underclocking may be an option.
For regular desktops at work, we're becoming very enamored with the 45W AMD parts. Those PCs are almost absolutely silent, especially when under a desk (Antec Sonata II cases).
I'm even considering switching over to 2.5" laptop drives inside to save another 5-8W. And probably make the systems even quieter. But the drives would cost too much.
Have to run the numbers again... 100W of power savings for a 24x7 PC is around $150/year in these parts. But it's only about $35/year for a PC that's only turned on during work hours.
19/22" LCD monitors almost pay for themselves in 3-4 years now, compared to our old 17" CRTs.
I usually try to wait until I can quadruple my main specs for a reasonable outlay of cash, and I think the time may be right.
Unless you're willing to wait 5-8 years now... the days of quadrupling your main specs are gone. CPU performance no longer doubles every 15 months.
Beyond that... I have a Phenom X4 at home. It competes favorable with Intel products that would have cost about the same amount. (Quite frankly, I don't give a rat's ass about CPUs that cost more then $300.)
8GB is probably affordable on a single-socket board. The 2GB DIMMs are not that expensive (but nowhere near as cheap as the 1GB DIMMs). 16GB is probably too expensive until 2010.
With Vista as my only option, my plan was to stick with XP as long as humanly possible. I have my own volume-licensed copy of XP Pro, so it's a somewhat realistic plan. If Windows 7 proves to be as high-quality as the pundits claim, that might just be enough to make me leave XP.
That's our current plan. We pushed through a WinXP upgrade across the board just before Vista shipped (because we knew Vista was a dog). All of the machines are dual-core, 2GB RAM, and will probably last us well into the 2011-2012 time frame. We also tossed a few copies of XP in a drawer for builds that we didn't plan for.
Still, I'm hoping that Win7 is at least decent and useful like XP. Because as we get past 2010, it's going to be harder and harder to keep XP running. Skipping Vista was easy if you timed things right, but skipping two generations is going to be difficulty.
(So nice not to have the majority of the PCs still running Win98 with 32MB or 64MB of RAM... that was a real PITA.)
I've used both Linux and Windows for servers for a decade.
I think Windows 2008 "core" mode is going to be too little too late. The more time I spend working with Linux servers, the power of the command line, the "everything is a file" mindset of Unix/Linux, and the sheer openness of the underlying tech - the less certain I am that Windows makes a good server product.
At least, if you don't want to spend lots and lots of money on add-on packages.
Some of the high points that have made my job easier in the past year:
- bash scripting
- LVM (flexibility, when our expected disk layout changed)
- having SecureCRT keep log files of all my sessions (so I can go back and figure out what I did 3 months ago in a certain situation)
- using FSVS and SVN to keep track of all changes on the server
- plain text configuration files that can be diff'd, grepped, and version controlled
- SELinux, the powerful iptables firewall
- a good security track record
Even better, the things that I learn on Linux server admin transfers mostly intact over to Solaris, Unix and BSD. And even Mac OS X. Whereas the things I learn on Windows only apply to Windows.
(Sorry for being long-winded, but I've spent a lot of time at the Linux command line in the past month. And things like LVM have saved my bacon a few times in the past year, allowing us to reconfigure servers on the fly when we forecast incorrectly.)
Reading, playing with the tech, making things go boom... and then fixing them.
The old Microsoft Certified DBA exams weren't that difficult, and there was some good things in there that were not just Microsoft-specific. (I finished the MCDBA cert back in 1999/2000, I've never recertified since then.) But I've been mucking with databases since the DBase III / CA-Clipper days and I can generally get around in 3 or 4 different database packages.
Beyond that, start playing with at least 2 or 3 different database products and learn their quirks. It's that or pay someone. In-depth knowledge of database concepts (normalization, tables, indexes, views) and SQL language (joins, unions, subqueries) pays off in spades. At least the major vendors now have freebie versions that everyone can play with.
Unfortunately, I have zero desire to put up with corporate behavior, so I'm staying in a small company where I can be the DBA, the sysadmin, the lead programmer. So my current focus is PostgreSQL and figuring out how fast we can move away from MS SQL Server. (The answer there is "soon".)
Besides, with the prices of Blu-ray players and discs now dropping rapidly, why bother with a solution that could tie up your broadband connection for long periods of time and the picture quality still can't compare to a real 1080p-resolution Blu-ray disc?
With modern codecs, excellent video quality can be had for 2-3 Mbps for a 720p video. The 1080p is about 2.25x more pixels, but since codecs scale well, you could probably do an excellent quality for 4-5 Mbps. Even 3Mbps could probably do well for the majority of video at 1080p.
Frankly, at 4Mbps for 1080p, you're probably not going to be able to tell the difference between that and Blu-Ray.
I may be biased, because I live in an area where the local cable company provides 20Mbps service. So 2/3/4Mbps streams don't phase me at all.
Am I the only one that has found this? I purchased a new Thinkpad ~6 months ago and it's terrible for a variety of reasons...
(shrugs) I have a T61p that I love (it's about a year old now?). It replaced an old (c2002) Toshiba Tecra. My T61p is the 15" 1680x1050 display, 3GB of RAM, and a Centrino Duo CPU running WinXP Pro.
They're still very solid units.
I don't care much for some of the ThinkVantage software, but I've learned to put up with its quirks. (They really need to sink some funding into fixing up that software.)
Eh, we just got done (mostly) with a 3 year upgrade cycle. I was hoping to go faster, but a lot of it depended on how much cash flow we had at the time.
We started back in 2005(?) right about the time that dual-core CPUs were still $300, but AMD was about to drop the prices on the X2s to below $150. At that point in time, 1/2 to 3/4 of our desktops were still running Win98 (and were from 1998-2000 purchases). The rest were running WinXP. A lot of machines were P3s with 300-800MHz CPUs, and the WinXP boxes were from 2001-2003 with a single-CPU (1.8GHz) and 512MB to 1GB of RAM.
We're now up to 95%+, WinXP Pro, Office Pro 2003, AMD Athlon64 X2 (45W versions where possible), 2GB of RAM, and (2) 250GB+ HDs in RAID1. The motherboards are all AM2 socketed. Average cost per box was about $1150, including $430 for Office Pro and XP Pro. We cut corners where it made sense, but didn't low-ball the units.
We have a few copies of WinXP Pro socked away in a drawer for new builds. We can get Office 2003 via a volume license with Microsoft (we're actually buying Office 2007 in packs of 5, and getting downgrade keys). Worst case, we buy Vista Ultimate OEM and get downgrade rights to WinXP Pro.
Expected lifespan for these units is at least 5-8 years. So the units that we put into place in 2006 will be starting to get long in the tooth comes 2011-2013. At that point, we can either decide to swap out the CPUs for quad-cores, or drop another gigabyte of RAM in. Because we bought dual-core systems, we expect them to age very well and not seem "slow" like single-core units usually do after 2 years. So we might even get an easy 10 years out of them, assuming no component deaths (such as capacitors leaking/bursting).
Starting in 2010, I expect to be evaluating Windows 7 (we're skipping Vista). Those desktops will probably end up being quad-core, 4GB or 8GB RAM, RAID1, and will probably still end up costing us around $1100 per seat (including licenses). But we won't be upgrading in bulk again until around 2012 or later.
We've still got a few single-core XP machines. They're used by our less-demanding users, but I have no compunction about replacing them with dual-core boxes once they wear out.
Basically, we jumped on dual-core as soon as the CPUs dropped below $300. The advent of multi-core CPUs means that systems should have very long useful lifespans compared to the single-core CPUs of yesteryear.
Not any more. I'm searching for one currently, and $130-150 seems to be the cheapest, and those are from sketchy sources. Love to know if you have a cheaper line.
The only way to get XP legally now is either to sign a license agreement with Microsoft (you'll probably have to buy in quantities of 5?) or buy Vista Ultimate OEM and pester Microsoft for a downgrade key.
(Vista Ultimate is the only OEM version that has downgrade rights.)
One other reason that I like software RAID.
Let's say that we start with (15) 500GB disks. Next year I start dropping in 1TB disks, one-by-one. At first, we only use 500GB on the 1TB disks. But once we drop the 15th disk in, we can create additional partitions on the drives, create new Software RAID arrays, and then extend LVM volume groups across the new arrays.
Which lets us double our data density, without having to backup lots of data, tear the server down, put new disks in, build an all new array, then restore all that data. (Although I would hope that modern hardware RAID card management would allow you to grow the array by changing out disks.)
We might even upgrade the disks in the array a second time once 2TB drives get below $100 each. Past that point, and I'd expect the server to be getting long in the tooth with a need to be replaced.
A lot of it is not having to deal with the vagaries of finding the drivers to setup hardware RAID on a Linux box. Or deal with strange management interfaces, whereas if we simply use the RAID controller in JBOD mode, we can simply admin the box using Linux Software RAID, no matter what controller we use. And we can move those disks to another box, using a different RAID controller, and still have it work.
Software RAID also allows you to slice up the spindles into strange and varied shapes. Such as mixing RAID6, RAID10, RAID1 across the same set of spindles, depending on your access patterns or sensitivity to data loss. I might assign the first half of 3 drives to RAID1, the first half of the next 7 drives to a RAID10 array with hot spare, and put a RAID6 across the first half of the last 5 drives. Then across the back half of all the drives, I can put a large, seldom-used RAID6 array for bulk storage.
A lot of failure modes nowadays seem to be RAID slices deciding to desync with the rest of the array. Or bitrot which gets discovered during a weekly resync of the array. If a slice is an entire disk, you could be dealing with a large amount of resync work. But if you divide each disk into 4 slices and setup 4 arrays, a desync event might only affect one out of the four arrays. Giving you faster rebuild times.
Eh, for $5000, the upper limit is probably around 15-spindles. Including the cost of the case ($600), a good motherboard and CPU and RAM ($600), a good 16-port SATA controller ($1000), and all of the drives (15 drives @ $130 each adds up quick). Okay, well, that's only about $4000.
Figure RAID-6, 1 hot spare, net useful space of (12) 1.5TB disks. Which is about 18TB (probably closer to 16.5TB once you factor in over-head and the million bytes vs megabytes conversion).
Still not shabby at all.
The problem with that is power consumption. Build your own, and you'll be burning a lot of power unnecessarily because it's overkill. Contrast that with the ReadyNas Duo, which I own, that pulls on average around 30-40W. Much better and green.
The modern AMD 45nm energy-efficient parts have a max TDP of 45W. Which means that they're going to run at a lot less then that when not under load. Couple that with the new 500GB SATA 2.5" laptop drives and you're looking at a regular desktop tower with similar power requirements as the ReadyNAS. Or at least something in the ballpark of 30-60W.
As much as I like the ReadyNAS units (which are Linux-based and in an pinch data can be recovered using a Linux machine), I don't care much for the proprietary power-supply or other parts that I can't pickup at the neighborhood computer parts store.
(I've got a tits-up ReadyNAS NV sitting here beside me. Motherboard is completely fried, probably due to a defective power-supply. The unit is out of warranty, since it was bought before Infrant got bought up by NETGEAR. Waiting on the client to decide whether to spend $950 for a new unit, or whether they want us to simply build a whitebox PC using one of the NAS-like distros mentioned here.)
software RAID ... is woefully slow compared to even a slow hardware RAID
Wrong. Go do your homework.
Well, maybe... what I've found over the years is that RAID-0, RAID-1, RAID-10 all perform extremely well as Software RAID without taxing the CPU.
But RAID-5 and RAID-6 do tend to become CPU-bottlenecked given enough spindles.
So if you're going with RAID-5 or RAID-6, makes sure to purchase a CPU with a high single-core performance (Linux Software RAID is currently not multi-threaded enough to split the calculations across multiple cores yet for a single array). The last RAID-6 box that I built, we made the mistake of going with a 1.8GHz CPU instead of a 2.4GHz or 2.6GHz CPU, which cut our performance by about 1/4 to 1/3 of what it could have been.
The spindles are definitely not at 100% utilization, but you can see the software RAID process pegging one of the CPUs at 100%. (Fairly modern motherboard, PCIe-based, pair of Opteron dual-core, 8GB RAM, 15 SATA drives.)
7 250 gig Seagates, all IDE
RAID-10 would have been a better choice for performance. Six drives for the array, with one hot-spare on standby. Net space would only be around 700GB, but you'd get pretty good performance.
Long as the PCs have a license sticker on the machine such as Vista or higher they have the right to downgrade for free.
Not true. The only version of Vista that you're allowed to downgrade to XP is the Vista Premium with all the bells and whistles.
Not sure about the VLKs, I know that the only way to get MS Office downgrades is to go with a volume license program for Office 2007, which then gives you the rights to downgrade to 2003.
I actually wouldn't be surprised if user space stayed 32 bit (or mostly 32 bit) for a very long time. The one thing Intel got right with the 386 is that its protected mode allowed for a mixture of 16-bit and 32-bit program contexts. AMD continued this tradition with x86-64. It's possible to have a 64-bit kernel with 32-bit user space applications. (Indeed, I've been tempted to set up a Linux machine that way in the past--put a 64-bit kernel under a 32-bit install.)
There's a lot of modern apps that are already pushing 500-800MB of memory usage. It's not far-fetched to see apps eating up 2GB or more within a year or three. And if the program is limited to 3GB (like it is in the x86 32-bit space), then you're up against limits that you wouldn't have if user space was 64bit.
It's documentation is one of the bests among distros. They cover not only installation, but deep customization and administration too.
Yes, but the quality of the packages in their repository leaves a lot to be desired.
(I did Gentoo for 3+ years... gave up and switched to CentOS/RedHat so I could simply get things done without fiddling anywhere near as much.)
Money. HDD's will keep getting cheaper. I'm betting on 2010.
For bulk storage, yes, magnetic HDs are cheaper (and will be for a good while). But we're already at the point that 64GB SSDs are inexpensive enough to be a good option. And that will handle a small segment of user needs. It won't be long until the 128GB SSDs drop enough in price that those too are a good option.
Right now, the 128GB is about the tipping point size. The 64GB drives feel a bit small for most users, but 128GB is pretty equivalent to the 120GB drives that are common on modern laptops.
So I'm thinking mass adoption of 128GB SSDs in laptops by Fall 2009, with wider use of the 256GB drives in mid-2010.
And if they can make the leap to 512GB SSDs and get the price under $200 or $300, I think that will pretty much doom the use of rotational magnetic HDs in laptops. Except for people that absolutely have to have 1TB of capacity.
Another option is to set up a second SSH daemon on a different port, and which only allows logins using public key (and possibly only by a specific user.) If you get blocked out on port 22, you can just use this side door to get in, as long as you've got your key.
Got a link for setting up a 2nd SSH daemon?
(The best move against this attack is either to go with port-knocking or simply move your primary SSH port to a non-standard port. We've chosen the latter solution on all of our public facing servers, along with requiring public-key logins.)