I have a number of clients who rely on the Xserve as a solid, dependable and data-centre ready server platform and am VERY concerned with the recent announcement to kill the Xserve.
For a small design studio, the Mac Pro may be a valid replacement, for a small business with email and Office documents, the Mac mini Server is a wonderful option, however for enterprise business with a large number of Macs, there is an absolute requirement for a powerful and RACK MOUNTED server platform with redundancy built in.
I have large mission-critical Kerio Connect deployments, I have large HELIOS EtherShare installations with huge amounts of data and tight deadlines are always an issue.
I have Xserves installed in climate-controlled server rooms and datacentres, taking up a minimum of rack space and performing admirably and these customers, the ones that will happily spend hundreds of thousands of dollars every few years to ensure their hardware is up-to-date, will be left with no option for a serious Apple Mac OS X Server platform.
Sure, this doesn't affect the office with three iMacs and a Mac mini Server and it doesn't affect the design studio with a few Mac Pros and some MacBooks who went with a Mac Pro as a server because it was $100 cheaper than the Xserve, but for the customers who are spending the real money, they need a real server.
I know that I am not the only one in this situation and I ask that this decision please be reconsidered as it will have a serious knock-on effect to the image of Apple as a serious contender in enterprise environments.
OK, and as a counterpoint to your valid and very well reasoned point above, look at this for an example.
Boot a machine. Log in and have 9 applications launch simultaneously at login. On my workstation, this went from around 4 minutes until I had a usable desktop environment with the disk stopping thrashing to around 15 seconds. This is on a workstation with 8GB RAM and Quad Xeon, moving from a 7.2k Cuda ES to an Intel SSD.
Apps set to launch at login were: mail, browser, task monitor, InDesign, Photoshop, Illustrator, Bridge, iTunes and a text editor.
No special tuning of the system or anything else required, just swapping the HDD for an SSD.
Seriously though, go out and spend just a bit more on a Tier 1 manufacturer's SSD - Kingston or Intel are my faves, whack it in as the boot drive and prepare to be amazed.
You underestimate the lifespan of current flash devices. For example: Assume a device that can take 100k writes before failing, and your wear levelling algorithms are 50% effective. If you are writing continuously to this drive at 1MB/sec (24/7) you are going to get 6 years of use from it.
We can tweak the figures to be a bit more realistic - writing at 100MB/sec for 1% of the time (24/7) then you're still going to get 6 years use from it. I'd not expect a rotating hard drive to to much better than this for me...
Why the hell do you want a half a terabyte of SSD? Because it's the most expensive offering?
RAID 0 and RAID 1 are nowhere near SSD in terms of power consumption, throughput and IOPS.
In today's computing environment, RAM is plentiful, CPU cycles are cheap, storage is abundant yet IOPS will bring even a high end machine to it's knees.
I was migrating some data from an old laptop (2 year old MacBook Pro) to a new one (MacBook Pro with a small SSD). I don't know what it's like on Windows or lInux, but on OS X once you're hitting 500-800 IOPS on a 7.2k hard drive everything slows to a crawl. You CPU utilisation can be idle, your RAM usage can be well within the amount of physical RAM installed yet too many IOPS and you soon can't do much with the machine.
On this new machine, I was copying a mail spool to it (mbox folders) installing software and Spotlight (full text indexing) was running in the background. This machine (a laptop mind you, not a workstation) was pulling in 7500 IOPS and not breaking a sweat - it was quick, responsive and completely usable for interactive tasks.
In order to get 7k IOPS from spinning media, you're talking about Fibre Channel or iSCSI storage arrays costing tens of thousands of dollars.
I, for one, am more than happy to put up with a small boot drive (40-60GB) if it's an SSD and move my bulk storage to spinning media. After that experience I now carry a laptop with a 64GB SSD and a 500GB FireWire external drive for bulk data and I couldn't be happier with that setup. I've even made the boot drive (and apps drive) in my workstation a small SSD, with bulk data on spinning media. I can boot this machine in mere seconds and launch half a dozen apps at login and it just doesn't slow down.
If you haven't used a machine with an SSD in real life, don't knock it until you've tried it.
It used to be that adding more RAM to a machine was the cheapest way to speed it up as just about all machines used to be (more or less) RAM bound. Now it's IOPS and adding an SSD is the cheapest way to have a more responsive machine. Older machines will potentially benefit even more than a newer machine as the relative speedup can be even greater...
OK. so the mass of the moon is, oh about 7.346 x 10^22kg that's approximately 73459000000000000000 tonnes. If we extract, say, 1 million tonnes of stuff from the moon, that's about 1.3 x 10^-17 %, also known as a poofteenth of a percent. According to my calculations, this will be enough to move the moon closer to us by about 4.76 x 10^-11 metres or approximately the diameter of a hydrogen atom.
Life will be a lot easier for you if you run a calendar server that supports an open standard, such as CalDAV. Don't bother syncing the phone with the desktop computer, sync the desktop computer and the phone with your calendar server. Things will work a lot better that way, you can share your calendars with each other and you've got a single point to backup for all calendar information.
If your calendaring solution supports publishing via CalDAV or Exchange ActiveSync, then the iPhone will sync over-the-air for both of these systems. I have a number of clients running Kerio Connect (not Open Source, but runs on Linux if you're interested) that's essentially a drop-in replacement for Exchange and it supports ActiveSync. iPhones sync to this for mail, calendar and contacts very well. As do most other smartphones that aren't a BlackBerry.
If you don't like the idea of using ActiveSync, you can configure mail via IMAP and calendar via CalDAV and it also works very well.
Exactly! if you ask me the time, and I say it's "two" you immediately know I don't mean it's 00:02 (as in two past midnight) and you're also generally not asking me this at 02:00 (or 2:00 AM) as if I'm talking to you at that hour, I'm not going to be making any sense, so you know straight away that it's 14:00 or 2:00 PM.
Why is this so difficult for some people to understand?
I tell my software that a job is starting at 8 and finishing at 930. It knows what I mean.
How on earth could anyone assume that 13 is 1:30? or 00:13?
Would these be a reasonable assumption if you asked me the time and I said "twelve" - could anyone really think that I really mean 01:20 or 00:12? No, seriously, could they?
Developing in it is a pain in the arse for sure, but for a cross-platform database with a half-decent layout engine for creating decent looking forms, it's hard to beat.
Look, I'll concede that I'm not a coder - I can cut and paste and make small changes and that's about it. The code I have in FileMaker Pro to do what I want is well and truly WTF-worthy, however it saves the users of this database a large number of keystrokes. This improves their accuracy, it improves the speed at which data can be entered and most importantly of all, it is what the end users wanted (arrived at through a process of refinement) so I have happy end users who have a system they are happy to work with as it works the way they want it to. I have the computer doing the hard work, which is the way it should be.
For those who are interested, the code I have is (post it to thedailywtf if you wish, it works beautifully for me)
This took maybe an hour to research up and tweak to my needs and it's going to save far more than that in saved end-user time with the users of this database.
So, you'd rather an end user when forced by a database program that will only accept explicitly formed times to enter hh:mm AM or hh:mm PM (and not any close variant of this, such as hh:mm am) rather than entering something a lot quicker and less prone to error, such as 7p or 1245.
In the first case, entering 7:00 PM this takes 9 keystrokes, including the Shift key. Reduce that to 2.
In the second case, entering 12:45 PM this also takes 9 keystrokes and is reduced to 4.
If you're doing this hundreds of times a day, the extra time very quickly adds up, plus you need to think a lot harder about ensuring you enter exactly what the database wants as a timestamp, remembering capitalisation etc...
(I have a range of hours defined that are AM or PM - 700 is 7pm for instance - this is completely arbitrary and works perfectly for the intended use)
If I'm booking an onsite technician for instance, they are not going to be onsite at 7am, I'm not sending anyone out that early in the morning, yet it's completely possible they'll be doing work after hours starting at 6 or 7. In the same light, I'm not sending anyone out to start work at 8pm, but would quite happily have someone going out at 8am.
In this case, what is "retarded" about having a rule that determines that 8.. 11 is AM and 12, 1... 7 is PM?
In the extremely rare situation that this rule doesn't apply, enter the time as "7" for instance and it gets corrected to 07:00 AM, change the A to a P and you're done. Either that, or enter the time as "7p" and it's put in as 7:00 PM
Or would you rather have to enter every single time value as hh:mm:ss AM|PM explicitly?
The user initially loved the idea of the proper date field - a small tweak to it and they would have been completely happy.
Defaulting to the year makes sense, as long as you have a well defined rule for it that the user can understand - if you commonly enter records for Jan of the following year in December, make it so that a month of 01 when entered late in the year is for the following year - similar to how most vendors worked around the y2k bug by having a rule that anything after (for example) 40 was 19nn and anything before 40 was 20nn
Interact with the end user, refine your model of what they want and you will end up producing a better product and have happier end users.
So in that case it's a process of refinement - implement what they say they want and seek feedback on the implementation rather than implementing what they say they want (even though their spec was incomplete) and saying "There, that's it. Take it or leave it"
This is a prime example of not listening to what the users want.
It is trival for the software to figure out what the current year is. With a small amount of effort done once on the part of the programmer to convert a day/month (or month/day) date into a full date, the user would have been happy, the software would have worked as they wanted it to and it would have been quicker and less error-prone for them to enter the data.
Instead the software vendor implemented what they thought the user wanted and more importantly didn't listen to the user completely, they implemented half of what the end user wanted and this resulted in more work having to undo the work that had been done to revert the system back to the "old 'n busted" way it was before.
Customers ask for things they want, but the developer needs to be willing to listen to them,
I had a similar thing happen recently, however this was for a database I was developing for my own purposes.
It has a field type of time, but it's really strict - you must enter a time as hh:mm[:ss] AM|PM anything else beeps at you as being invalid (duh)
With some coding effort and a liberal amount of google searching, I was able to have this field exhibit a lot more intelligence and be infinitely more user-friendly. I now have it so that you can enter just about anything that can be interpreted as a time and it'll sort it out. I get the computer, not the user, to do the hard work.
Now, I can enter 800 and it will be 08:00 am (I have a range of hours defined that are AM or PM - 700 is 7pm for instance - this is completely arbitrary and works perfectly for the intended use)
I can enter 1525 and it will enter 3:25 PM, I can enter 4 and it will enter 4PM, I can enter 9 and it will enter 9am. I can enter 12:34 and it will also take it...
It's now a lot quicker for me to quickly enter a few numbers rather than enter numbers separated by colons and an explicit am or pm. It's also a lot less error prone as there's less thought involved, less keystrokes and no need to use a shift+key stroke combination.
In your example, a few more minutes of coding effort to detect a supposedly invalid date (I know what 10/20 is, the user knows what 10/20 is, you know what it is, so tell the computer what it is) and everyone would have been happy.
Rubberhose (Pronounced Marutukku) is transparently deniable encryption, developed by (among others) Julian Assange. This seems to do exactly what you're trying to do, so even if you want to go ahead and implement it yourself from scratch, it's worth reading up on what they've done to get some ideas and avoid some potential pitfalls.
Rubberhose is a computer program which both transparently encrypts data on a storage device, such as a hard drive, and allows you to hide that encrypted data. Unlike conventional disk encryption systems, Rubberhose is the first successful, freely available, practical program of deniable cryptography in the world. It was released in an earlier form in 1997, but has undergone significant changes since that time. The design goal has been to make Rubberhose the most efficient conventional disk encryption system, while also offering the new feature of information hiding.
Rubberhose is a type of deniable cryptography package. Deniable cryptography gives a person not wanting to disclose the plaintext data corresponding to their encrypted material the ability to show that there is more than one interpretation of the encrypted data. What deniable crypto means in the Rubberhose context is this: if someone grabs your Rubberhose-encrypted hard drive, he or she will know there is encrypted material on it, but not how much -- thus allowing you to hide the existence of some of your data.
Blizzard have something like over 11 million subscribers. Let's say that each of them are paying $10 per month. (It's $15 for month-to-month payments, and as low as $13 per month if you buy a 6 month block) That is a revenue stream of 110 Million Dollars per month. That's over 85 million Euro per month. 600k per month is a drop in the ocean compared to this.
Slashdot Mirror
Dear Apple,
I have a number of clients who rely on the Xserve as a solid, dependable and data-centre ready server platform and am VERY concerned with the recent announcement to kill the Xserve.
For a small design studio, the Mac Pro may be a valid replacement, for a small business with email and Office documents, the Mac mini Server is a wonderful option, however for enterprise business with a large number of Macs, there is an absolute requirement for a powerful and RACK MOUNTED server platform with redundancy built in.
I have large mission-critical Kerio Connect deployments, I have large HELIOS EtherShare installations with huge amounts of data and tight deadlines are always an issue.
I have Xserves installed in climate-controlled server rooms and datacentres, taking up a minimum of rack space and performing admirably and these customers, the ones that will happily spend hundreds of thousands of dollars every few years to ensure their hardware is up-to-date, will be left with no option for a serious Apple Mac OS X Server platform.
Sure, this doesn't affect the office with three iMacs and a Mac mini Server and it doesn't affect the design studio with a few Mac Pros and some MacBooks who went with a Mac Pro as a server because it was $100 cheaper than the Xserve, but for the customers who are spending the real money, they need a real server.
I know that I am not the only one in this situation and I ask that this decision please be reconsidered as it will have a serious knock-on effect to the image of Apple as a serious contender in enterprise environments.
OK, and as a counterpoint to your valid and very well reasoned point above, look at this for an example.
Boot a machine. Log in and have 9 applications launch simultaneously at login. On my workstation, this went from around 4 minutes until I had a usable desktop environment with the disk stopping thrashing to around 15 seconds. This is on a workstation with 8GB RAM and Quad Xeon, moving from a 7.2k Cuda ES to an Intel SSD.
Apps set to launch at login were: mail, browser, task monitor, InDesign, Photoshop, Illustrator, Bridge, iTunes and a text editor.
No special tuning of the system or anything else required, just swapping the HDD for an SSD.
... and I wasn't having a dig at you. ^_^
Seriously though, go out and spend just a bit more on a Tier 1 manufacturer's SSD - Kingston or Intel are my faves, whack it in as the boot drive and prepare to be amazed.
How is this comment modded Insightful?
"I bought one SSD, one that many other people have had problems with, and I had problems with it. I'm never buying another SSD"
You underestimate the lifespan of current flash devices.
For example:
Assume a device that can take 100k writes before failing, and your wear levelling algorithms are 50% effective.
If you are writing continuously to this drive at 1MB/sec (24/7) you are going to get 6 years of use from it.
We can tweak the figures to be a bit more realistic - writing at 100MB/sec for 1% of the time (24/7) then you're still going to get 6 years use from it. I'd not expect a rotating hard drive to to much better than this for me...
Why the hell do you want a half a terabyte of SSD? Because it's the most expensive offering?
RAID 0 and RAID 1 are nowhere near SSD in terms of power consumption, throughput and IOPS.
In today's computing environment, RAM is plentiful, CPU cycles are cheap, storage is abundant yet IOPS will bring even a high end machine to it's knees.
I was migrating some data from an old laptop (2 year old MacBook Pro) to a new one (MacBook Pro with a small SSD). I don't know what it's like on Windows or lInux, but on OS X once you're hitting 500-800 IOPS on a 7.2k hard drive everything slows to a crawl. You CPU utilisation can be idle, your RAM usage can be well within the amount of physical RAM installed yet too many IOPS and you soon can't do much with the machine.
On this new machine, I was copying a mail spool to it (mbox folders) installing software and Spotlight (full text indexing) was running in the background. This machine (a laptop mind you, not a workstation) was pulling in 7500 IOPS and not breaking a sweat - it was quick, responsive and completely usable for interactive tasks.
In order to get 7k IOPS from spinning media, you're talking about Fibre Channel or iSCSI storage arrays costing tens of thousands of dollars.
I, for one, am more than happy to put up with a small boot drive (40-60GB) if it's an SSD and move my bulk storage to spinning media. After that experience I now carry a laptop with a 64GB SSD and a 500GB FireWire external drive for bulk data and I couldn't be happier with that setup. I've even made the boot drive (and apps drive) in my workstation a small SSD, with bulk data on spinning media. I can boot this machine in mere seconds and launch half a dozen apps at login and it just doesn't slow down.
If you haven't used a machine with an SSD in real life, don't knock it until you've tried it.
It used to be that adding more RAM to a machine was the cheapest way to speed it up as just about all machines used to be (more or less) RAM bound. Now it's IOPS and adding an SSD is the cheapest way to have a more responsive machine. Older machines will potentially benefit even more than a newer machine as the relative speedup can be even greater...
OK. so the mass of the moon is, oh about 7.346 x 10^22kg that's approximately 73459000000000000000 tonnes. If we extract, say, 1 million tonnes of stuff from the moon, that's about 1.3 x 10^-17 %, also known as a poofteenth of a percent.
According to my calculations, this will be enough to move the moon closer to us by about 4.76 x 10^-11 metres or approximately the diameter of a hydrogen atom.
Life will be a lot easier for you if you run a calendar server that supports an open standard, such as CalDAV.
Don't bother syncing the phone with the desktop computer, sync the desktop computer and the phone with your calendar server.
Things will work a lot better that way, you can share your calendars with each other and you've got a single point to backup for all calendar information.
If your calendaring solution supports publishing via CalDAV or Exchange ActiveSync, then the iPhone will sync over-the-air for both of these systems.
I have a number of clients running Kerio Connect (not Open Source, but runs on Linux if you're interested) that's essentially a drop-in replacement for Exchange and it supports ActiveSync. iPhones sync to this for mail, calendar and contacts very well. As do most other smartphones that aren't a BlackBerry.
If you don't like the idea of using ActiveSync, you can configure mail via IMAP and calendar via CalDAV and it also works very well.
13 is also 00:13
13 is also 1:30
13 is also 13:00
which one of those is right when I type 13?
This is crazy talk. I'm just glad that you aren't writing any parsing software for me.
13 is "thirteen hundred hours" which can also be entered as 1300, 13:00, 1p, 1:00 p, 100p, 1:30 PM, 1:30 etc...
Exactly! if you ask me the time, and I say it's "two" you immediately know I don't mean it's 00:02 (as in two past midnight) and you're also generally not asking me this at 02:00 (or 2:00 AM) as if I'm talking to you at that hour, I'm not going to be making any sense, so you know straight away that it's 14:00 or 2:00 PM.
Why is this so difficult for some people to understand?
I tell my software that a job is starting at 8 and finishing at 930. It knows what I mean.
How on earth could anyone assume that 13 is 1:30? or 00:13?
Would these be a reasonable assumption if you asked me the time and I said "twelve" - could anyone really think that I really mean 01:20 or 00:12? No, seriously, could they?
Hell yeah, doesn't it ever!
Developing in it is a pain in the arse for sure, but for a cross-platform database with a half-decent layout engine for creating decent looking forms, it's hard to beat.
Look, I'll concede that I'm not a coder - I can cut and paste and make small changes and that's about it.
The code I have in FileMaker Pro to do what I want is well and truly WTF-worthy, however it saves the users of this database a large number of keystrokes. This improves their accuracy, it improves the speed at which data can be entered and most importantly of all, it is what the end users wanted (arrived at through a process of refinement) so I have happy end users who have a system they are happy to work with as it works the way they want it to. I have the computer doing the hard work, which is the way it should be.
For those who are interested, the code I have is (post it to thedailywtf if you wish, it works beautifully for me)
This took maybe an hour to research up and tweak to my needs and it's going to save far more than that in saved end-user time with the users of this database.
Let(
Self = GetAsNumber( Start Time );
Let(
inter =
GetAsNumber(
Case(
Length( Self ) = 1;
Self & "00";
Case(
Length( Self ) = 2;
Self & "00";
Case( Length( Self ) = 3; Self; Case( Length( Self ) = 4; Self ) )
)
)
);
Let(
suffix =
Case(
Right( Start Time; 2 ) = "am";
" am";
Case(
Right( Start Time; 2 ) = "pm";
" pm";
Case(
Right( Start Time; 1 ) = "a";
" am";
Case( Right( Start Time; 1 ) = "p"; " pm" )
)
)
);
Case(
inter > 0 and inter < 900;
Left( inter; 1 ) & ":" & Right( inter; 2 ) &
If( not IsEmpty( suffix ); suffix; " pm" );
Case(
inter 900 and inter < 1000;
Left( inter; 1 ) & ":" & Right( inter; 2 ) &
If( not IsEmpty( suffix ); suffix; " am" );
Case(
inter 1000 and inter < 1200;
Left( inter; 2 ) & ":" & Right( inter; 2 ) &
If( not IsEmpty( suffix ); suffix; " am" );
Case(
inter 1200 and inter < 1260;
Left( inter; 2 ) & ":" & Right( inter; 2 ) &
If( not IsEmpty( suffix ); suffix; " pm" );
Case(
inter 1300 and inter < 2160;
Left( (GetAsNumber( inter ) - 1200 ); 1 ) & ":" & Right( inter; 2 ) &
13 turns into 13:00 which is 1pm.
130 turns into 1:30 PM (as explained before 7 is defined as PM)
How is this so difficult to understand?
So, you'd rather an end user when forced by a database program that will only accept explicitly formed times to enter hh:mm AM or hh:mm PM (and not any close variant of this, such as hh:mm am) rather than entering something a lot quicker and less prone to error, such as 7p or 1245.
In the first case, entering 7:00 PM this takes 9 keystrokes, including the Shift key. Reduce that to 2.
In the second case, entering 12:45 PM this also takes 9 keystrokes and is reduced to 4.
If you're doing this hundreds of times a day, the extra time very quickly adds up, plus you need to think a lot harder about ensuring you enter exactly what the database wants as a timestamp, remembering capitalisation etc ...
Did you miss the part where I said:
If I'm booking an onsite technician for instance, they are not going to be onsite at 7am, I'm not sending anyone out that early in the morning, yet it's completely possible they'll be doing work after hours starting at 6 or 7.
In the same light, I'm not sending anyone out to start work at 8pm, but would quite happily have someone going out at 8am.
In this case, what is "retarded" about having a rule that determines that 8 .. 11 is AM and 12, 1 ... 7 is PM?
In the extremely rare situation that this rule doesn't apply, enter the time as "7" for instance and it gets corrected to 07:00 AM, change the A to a P and you're done. Either that, or enter the time as "7p" and it's put in as 7:00 PM
Or would you rather have to enter every single time value as hh:mm:ss AM|PM explicitly?
The user initially loved the idea of the proper date field - a small tweak to it and they would have been completely happy.
Defaulting to the year makes sense, as long as you have a well defined rule for it that the user can understand - if you commonly enter records for Jan of the following year in December, make it so that a month of 01 when entered late in the year is for the following year - similar to how most vendors worked around the y2k bug by having a rule that anything after (for example) 40 was 19nn and anything before 40 was 20nn
Interact with the end user, refine your model of what they want and you will end up producing a better product and have happier end users.
So in that case it's a process of refinement - implement what they say they want and seek feedback on the implementation rather than implementing what they say they want (even though their spec was incomplete) and saying "There, that's it. Take it or leave it"
This is a prime example of not listening to what the users want.
It is trival for the software to figure out what the current year is. With a small amount of effort done once on the part of the programmer to convert a day/month (or month/day) date into a full date, the user would have been happy, the software would have worked as they wanted it to and it would have been quicker and less error-prone for them to enter the data.
Instead the software vendor implemented what they thought the user wanted and more importantly didn't listen to the user completely, they implemented half of what the end user wanted and this resulted in more work having to undo the work that had been done to revert the system back to the "old 'n busted" way it was before.
Customers ask for things they want, but the developer needs to be willing to listen to them,
I had a similar thing happen recently, however this was for a database I was developing for my own purposes.
It has a field type of time, but it's really strict - you must enter a time as hh:mm[:ss] AM|PM anything else beeps at you as being invalid (duh)
With some coding effort and a liberal amount of google searching, I was able to have this field exhibit a lot more intelligence and be infinitely more user-friendly. I now have it so that you can enter just about anything that can be interpreted as a time and it'll sort it out. I get the computer, not the user, to do the hard work.
Now, I can enter 800 and it will be 08:00 am (I have a range of hours defined that are AM or PM - 700 is 7pm for instance - this is completely arbitrary and works perfectly for the intended use)
I can enter 1525 and it will enter 3:25 PM, I can enter 4 and it will enter 4PM, I can enter 9 and it will enter 9am. I can enter 12:34 and it will also take it...
It's now a lot quicker for me to quickly enter a few numbers rather than enter numbers separated by colons and an explicit am or pm. It's also a lot less error prone as there's less thought involved, less keystrokes and no need to use a shift+key stroke combination.
In your example, a few more minutes of coding effort to detect a supposedly invalid date (I know what 10/20 is, the user knows what 10/20 is, you know what it is, so tell the computer what it is) and everyone would have been happy.
Rubberhose (Pronounced Marutukku) is transparently deniable encryption, developed by (among others) Julian Assange.
This seems to do exactly what you're trying to do, so even if you want to go ahead and implement it yourself from scratch, it's worth reading up on what they've done to get some ideas and avoid some potential pitfalls.
1600GB RAM = 2GB RAM per core. That's a power of 2.
Blizzard have something like over 11 million subscribers. Let's say that each of them are paying $10 per month. (It's $15 for month-to-month payments, and as low as $13 per month if you buy a 6 month block) That is a revenue stream of 110 Million Dollars per month. That's over 85 million Euro per month. 600k per month is a drop in the ocean compared to this.
Delicious? You keep using that word. I do not think it means what you think it means.
... Kotaku explaining how Sony fucked over a well-liked privately owned importer that Sony used for its own executives...
But was it Sony purchasing the items as a company, for their execs, or was it the execs purchasing items for themselves?
New punctuation: "~" at the end of a line to indicate 'Snarky'. http://harns.blogspot.com/
You forgot to put a tilde at the end of that last sentence...