I noticed that your homepage is at UIUC. Have you spoken with the NCSA people over at Beckman?
I'm a manager with a large company that used to be an NCSA partner. One of the things they loved to demo was a biology workbench, so do open collaborative stuff. Now, I'm a mathematician by training, not a biologist, so I don't know if this was just demo-ware or not, or even connected to what you do.
Anyway, I worked on projects with NCSA a few years ago and they were building a lot of great tools for fundamental research. Perhaps this is no longer the case. But it's surprising that your adminstration can't get with the CS administration and learn that open source is, in fact, a good thing.
I'd volunteer, naturally, but I've been to central Illinois in the winter before and nearly froze to death:-)
As an aficionado of German beer, I'm sure that this will be the first real application. They just want to get the military to pay for some cool toys along the way.
So, as I understand it, a working group (WG) member creates a standard and then says, "Oh, hey, great standard guys, but now you're all going to have to pay me for it".
Is this not a conflict of interest? Should the WG member be immediately voted off? Perhaps they should be tarred and feathered, run out of town on a rail?
I prefer the latter approach, it may reduce the number of bogus patent claims.
In fact, I have a simple JCL interpreter for Linux. I read someone whinging one day that Linux was hard to use. Methinks, "hard to use, I'll show you hard to use!" Imagine 14 lines of JCL to call IEBGENER to copy a file....
Porting it to C# / Mono would somehow be wrong. I've done enough wrong already.
To receive the service, homes must be no farther than sixth-tenths of a mile from the neighborhood box.
That's not very far.
I live in town (Southlake, TX) and can't get DSL, can't get a modem to work at more than 28.8. Too far from the phone switch. The only consolation is the cable company, now if they could only define customer service.
The mainframe is MIPS per CPU, so the 16-way box is 16*120. Also, 120 MIPS is slow these days for a mainframe.
Write a simple memory intensive program and try it on a mainframe and try it on a PC. I guarantee that you won't get 3000 MIPS out of a desktop, even if the data fits in cache. Many reasons for this....
The s390 ISA is definitely CISC, you can copy a whole string with MVCL, that count's as one instruction. Do this on RISC machines and it might take a loop and execute dozens of instructions. Hence "Meaningless..."
About 2 years ago I wrote some C code to recursively quicksort 20M random integers and tried it on a bunch of platforms. A mainframe that was about 1 cycle behind fastest available gave me about the same single processor performance as a 1GHz PC, both a little slower than Alpha.
The big differentiator is memory architecture. How much time do you lose for a cache miss? Most processors only operate at 20-30% of theoretical maximum speed on big problems.
Memory speed has not kept up, that 2GHz box you dream about is not twice as fast as a 1GHz box, particularly if you're crunching a lot of data.
Sabre runs on a cluster of 8 IBM-compatible mainframes, running IBM's TPF operating system. It's about a dozen TB, handling >13,000 TPS. The system has to run 7*24 with zero planned downtime.
Pricing and ticketing functions are handled by a separate cluster, it will be migrated to open systems, first implementation is within weeks. The code is C++ & Java, not Lisp. Can't say more, but there's a lot of detail in Sabre press releases, interviews and public presentations if anybody cares to look.
The current fare searching is based on some advanced algorithms, written in C. The design team features a number of CS and math PhDs that did the algorithm design.
One major difference between ITA/Orbitz and Sabre/Expedia/etc is that Orbitz does *not* use the indsutry's standard seamless availability protocol to check for seat availability in real time. ie. Orbitz caches stale data and you take your chances if the seat is really available. Generating a shitload of irrelevant and/or unavabilable options is actually not that difficult. Most of the major computer reservation systems have contracts with the airlines specifying last-seat availability, hence the limitation on the number of options displayed.
Another difference is having to deal with all the complexity of the airline industry. Another posting on this site mentioned "state". ITA doesn't print tickets, doesn't handle international fares and doesn't take the debit-memo risk. If you mess up a fare rule, price a ticket too low and the airline notices, you pay the difference. Guess what? ITA relies on the Worldspan (mainframe!) systems to handle ticketing, accounting and all the other functions.
BTW - if it's not already obvious, I work for Sabre and am very familiar with the subject matter.
Suggestion time - might Slashdot readers be interested in an interview with one of our senior techies? I'd be happy to help set it up. The legacy assembler stuff is a small part of our business and there's a lot of interesting technology.
Most of the large airline systems run on IBM's Transaction Processing Facility (TPF). IBM keeps updating it, even added TCP/IP a few years back, but it's essentially 1960's technology. By the way, go to a web site like Expedia or even the new Orbitz site and you're still hitting mainframe assembler code in the background somewhere...
Try these on for size: most applications are written in assembler, manually divided into 4K blocks. No virtual memory, all storage preallocated at sysgen into fixed size blocks (woohoo - no fragmentation!). No filesystem, all you get is a shitload of blocks (381 bytes, 1055 bytes, 4K) and it's up to the programmer to do the rest.
I've seen code on these systems that was written in 1970-1972 and is still in use today, taking thousands of transactions per second. Somehow I don't see W2K apps lasting 30 years.
This would clearly be fun. Can you imagine some petrol-head outfitting his shoes with little turbochargers and nitrous-oxide injection? As soon as you make it a sport, people will take it to extremes.
My wife & I signed up for Marcus / @Home service. It is great when it works, but with outages of several hours, a few times per week, the overall service level is far below a dial-up modem. We are thinking seriously of going back to dial-up!
Here's a session from Unix to OS/390 (MVS), I use it all the time, in both directions. We also have mainframe jobs ftp files to servers, even to a Linux box.
/home/snfs01/alanw [!] > ftp mvsk Connected to mvsk.dev.sabre.com. 220 MVSSYK -- FTP Server, Enter command or HELP Name (mvsk:alanw): z6504 331 Enter PASS command Password: 230--- GREETINGS --- *** TCPaccess (FTP) for MVS-SYK *** Logged in - User=Z6504 Working directory "Z6504." 230 The local host for the control connection is 144.9.3.2. ftp> ls 200 OK, Ready 125 Transfer started AABKTFST.FAC1 ISPF.ISPPROF MESSAGE.LOG ROSS.DATAPULL TEST.SAS TKDATA 226-Transfer complete. 75 bytes sent in 0.04 seconds (1875 bytes/s) Path name: Z6504. User=Z6504 Data bytes read: 63. 226 End of reply. 75 bytes received in 0.08 seconds (0.91 Kbytes/s) ftp>
See - no problem! I've also used the command shell and tools like awk on the mainframe. But wait, I was doing that in 1994....
Much of my CGI work has been done in C++, not because it's fast, but because I'm familiar with it and I can access everything. I have programs that call CORBA services, access databases and make system calls, all running on NT.
I'd like to use a scripting language to do this, but couldn't find something with ODBC, CORBA and the other pieces I need. For example, a particular CORBA interface may not have been tested with Visibroker, which is what I have, and I didn't want to spend a week figuring out how to compile/link all the add-ins I'll need.
Invitation - if somebody has a better way to integrate all this stuff, please let me know.
I loved Delicatessen, saw it on the big screen when it was released - it's one of my favourite French films. I can't imagine it with subtitles, though.
I've had occasion to use this and thought it was pretty cool. There have been printed versions of these, but the online one is better.
Another interesting idea that I've seen printed is a musical theme dictionary, if you can plunk out the first few notes by ear then you can look up the sequence. Has anyone done this online? Would someone sue you for it, since printed and/or recorded music is a pretty touchy subject on the Internet.
My favourite sequence, not listed, is:
s(n):= 1,2,1,1,1,1,1,1.....
n=1,2,3,4,... is the number of people in an elevator and, if one of them farts, s(n) is the number of people who are sure who did it.
Hmmm, it's so hard to get tech workers in the US, the INS is a hopeless bureaucracy. I speak from experience, I'm an Aussie living in the US.
So, what happens next? US companies will have foreign workers telecommute, no messy forms to fill out. The taxes that would have been paid in the US are now paid outside. Congress is gonna love it!
Having lived and working in Australia, Europe and the US, I can tell you that international taxation is a nightmare. It's primarily based on residency, you have to tell the tax office where you were for every day of the year.
I don't see anything that would stop you living anywhere you want, even some nice tax haven.
The Egyptians thought of building a restaurant at the Cairo airport years ago. It's across the street from the airport Moevenpick hotel. It's kind of tacky and has corrosion in places. If I remember rightly, it is (was?) a 707.
I went there with some friends, all airline employees on business in Cairo, for dinner. The memorable part was when a guy laughed so hard after a joke someone told that he ripped the seat from the floor and fell in the aisle. Oh well, you had to be there...
Most commercial databases prefer raw partitions and you don't run a file system at all. The database manager has log files for commit/rollback processing. They run faster on raw partitions, since the OS is not "in the way". The DBMS actually doesn't need much of what a general purpose OS offers, that's why Oracle was promoting the "raw iron" concept (BTW what ever happened to that?).
Last I heard, maybe it's changed, Linux didn't support raw partitions.
I don't see what's groundbreaking about this. Our cat has a microchip implant, it was done about 2 years ago. If she gets lost, any veterinarian in the area can check her with a scanner and have her safely returned.
I don't enviseage having lights turn on for her when she enters a room, but it would not be difficult to do.
Slashdot Mirror
I noticed that your homepage is at UIUC. Have you spoken with the NCSA people over at Beckman?
:-)
I'm a manager with a large company that used to be an NCSA partner. One of the things they loved to demo was a biology workbench, so do open collaborative stuff. Now, I'm a mathematician by training, not a biologist, so I don't know if this was just demo-ware or not, or even connected to what you do.
Anyway, I worked on projects with NCSA a few years ago and they were building a lot of great tools for fundamental research. Perhaps this is no longer the case. But it's surprising that your adminstration can't get with the CS administration and learn that open source is, in fact, a good thing.
I'd volunteer, naturally, but I've been to central Illinois in the winter before and nearly froze to death
Unbreakable beer glass.
As an aficionado of German beer, I'm sure that this will be the first real application. They just want to get the military to pay for some cool toys along the way.
So, as I understand it, a working group (WG) member creates a standard and then says, "Oh, hey, great standard guys, but now you're all going to have to pay me for it".
Is this not a conflict of interest? Should the WG member be immediately voted off? Perhaps they should be tarred and feathered, run out of town on a rail?
I prefer the latter approach, it may reduce the number of bogus patent claims.
Alan.
Why not? Just gotta port an interpreter.
In fact, I have a simple JCL interpreter for Linux. I read someone whinging one day that Linux was hard to use. Methinks, "hard to use, I'll show you hard to use!" Imagine 14 lines of JCL to call IEBGENER to copy a file....
Porting it to C# / Mono would somehow be wrong. I've done enough wrong already.
That's not very far.
I live in town (Southlake, TX) and can't get DSL, can't get a modem to work at more than 28.8. Too far from the phone switch. The only consolation is the cable company, now if they could only define customer service.
MIPS = Meaningless Indicator of Processor Speed
..."
The mainframe is MIPS per CPU, so the 16-way box is 16*120. Also, 120 MIPS is slow these days for a mainframe.
Write a simple memory intensive program and try it on a mainframe and try it on a PC. I guarantee that you won't get 3000 MIPS out of a desktop, even if the data fits in cache. Many reasons for this....
The s390 ISA is definitely CISC, you can copy a whole string with MVCL, that count's as one instruction. Do this on RISC machines and it might take a loop and execute dozens of instructions. Hence "Meaningless
About 2 years ago I wrote some C code to recursively quicksort 20M random integers and tried it on a bunch of platforms. A mainframe that was about 1 cycle behind fastest available gave me about the same single processor performance as a 1GHz PC, both a little slower than Alpha.
The big differentiator is memory architecture. How much time do you lose for a cache miss? Most processors only operate at 20-30% of theoretical maximum speed on big problems.
Memory speed has not kept up, that 2GHz box you dream about is not twice as fast as a 1GHz box, particularly if you're crunching a lot of data.
What do you want to know?
Sabre runs on a cluster of 8 IBM-compatible mainframes, running IBM's TPF operating system. It's about a dozen TB, handling >13,000 TPS. The system has to run 7*24 with zero planned downtime.
Pricing and ticketing functions are handled by a separate cluster, it will be migrated to open systems, first implementation is within weeks. The code is C++ & Java, not Lisp. Can't say more, but there's a lot of detail in Sabre press releases, interviews and public presentations if anybody cares to look.
The current fare searching is based on some advanced algorithms, written in C. The design team features a number of CS and math PhDs that did the algorithm design.
One major difference between ITA/Orbitz and Sabre/Expedia/etc is that Orbitz does *not* use the indsutry's standard seamless availability protocol to check for seat availability in real time. ie. Orbitz caches stale data and you take your chances if the seat is really available. Generating a shitload of irrelevant and/or unavabilable options is actually not that difficult. Most of the major computer reservation systems have contracts with the airlines specifying last-seat availability, hence the limitation on the number of options displayed.
Another difference is having to deal with all the complexity of the airline industry. Another posting on this site mentioned "state". ITA doesn't print tickets, doesn't handle international fares and doesn't take the debit-memo risk. If you mess up a fare rule, price a ticket too low and the airline notices, you pay the difference. Guess what? ITA relies on the Worldspan (mainframe!) systems to handle ticketing, accounting and all the other functions.
BTW - if it's not already obvious, I work for Sabre and am very familiar with the subject matter.
Suggestion time - might Slashdot readers be interested in an interview with one of our senior techies? I'd be happy to help set it up. The legacy assembler stuff is a small part of our business and there's a lot of interesting technology.
Most of the large airline systems run on IBM's Transaction Processing Facility (TPF). IBM keeps updating it, even added TCP/IP a few years back, but it's essentially 1960's technology. By the way, go to a web site like Expedia or even the new Orbitz site and you're still hitting mainframe assembler code in the background somewhere... Try these on for size: most applications are written in assembler, manually divided into 4K blocks. No virtual memory, all storage preallocated at sysgen into fixed size blocks (woohoo - no fragmentation!). No filesystem, all you get is a shitload of blocks (381 bytes, 1055 bytes, 4K) and it's up to the programmer to do the rest. I've seen code on these systems that was written in 1970-1972 and is still in use today, taking thousands of transactions per second. Somehow I don't see W2K apps lasting 30 years.
Hmmm... I wonder if they could isolate the THC gene from cannabis sativa, then I could be permanently out of my tree...
I think I met a lawyer the other day who's already done this, judging by his dopey demeanor. I'll bet the bastard's patented it already too!
Alan.
This would clearly be fun. Can you imagine some petrol-head outfitting his shoes with little turbochargers and nitrous-oxide injection? As soon as you make it a sport, people will take it to extremes.
I'm curious what you'd call something even bigger in the future, is there a natural progression in the naming of these things?
My wife & I signed up for Marcus / @Home service. It is great when it works, but with outages of several hours, a few times per week, the overall service level is far below a dial-up modem. We are thinking seriously of going back to dial-up!
Here's a session from Unix to OS/390 (MVS), I use it all the time, in both directions. We also have mainframe jobs ftp files to servers, even to a Linux box.
/home/snfs01/alanw [!] > ftp mvsk
Connected to mvsk.dev.sabre.com.
220 MVSSYK -- FTP Server, Enter command or HELP
Name (mvsk:alanw): z6504
331 Enter PASS command
Password:
230--- GREETINGS ---
*** TCPaccess (FTP) for MVS-SYK ***
Logged in - User=Z6504 Working directory "Z6504."
230 The local host for the control connection is 144.9.3.2.
ftp> ls
200 OK, Ready
125 Transfer started
AABKTFST.FAC1
ISPF.ISPPROF
MESSAGE.LOG
ROSS.DATAPULL
TEST.SAS
TKDATA
226-Transfer complete. 75 bytes sent in 0.04 seconds (1875 bytes/s)
Path name: Z6504. User=Z6504 Data bytes read: 63.
226 End of reply.
75 bytes received in 0.08 seconds (0.91 Kbytes/s)
ftp>
See - no problem! I've also used the command shell and tools like awk on the mainframe. But wait, I was doing that in 1994....
Much of my CGI work has been done in C++, not because it's fast, but because I'm familiar with it and I can access everything. I have programs that call CORBA services, access databases and make system calls, all running on NT.
I'd like to use a scripting language to do this, but couldn't find something with ODBC, CORBA and the other pieces I need. For example, a particular CORBA interface may not have been tested with Visibroker, which is what I have, and I didn't want to spend a week figuring out how to compile/link all the add-ins I'll need.
Invitation - if somebody has a better way to integrate all this stuff, please let me know.
I loved Delicatessen, saw it on the big screen when it was released - it's one of my favourite French films. I can't imagine it with subtitles, though.
Alan.
I wonder how fast someone will try and overclock one of these...
I've had occasion to use this and thought it was pretty cool. There have been printed versions of these, but the online one is better.
:= 1,2,1,1,1,1,1,1.....
Another interesting idea that I've seen printed is a musical theme dictionary, if you can plunk out the first few notes by ear then you can look up the sequence. Has anyone done this online? Would someone sue you for it, since printed and/or recorded music is a pretty touchy subject on the Internet.
My favourite sequence, not listed, is:
s(n)
n=1,2,3,4,... is the number of people in an elevator and, if one of them farts, s(n) is the number of people who are sure who did it.
Alan.
Actually, I thought it was kind of cute. Better than stuffing an article full of banner ads every paragraph or three.
Hmmm, it's so hard to get tech workers in the US, the INS is a hopeless bureaucracy. I speak from experience, I'm an Aussie living in the US.
So, what happens next? US companies will have foreign workers telecommute, no messy forms to fill out. The taxes that would have been paid in the US are now paid outside. Congress is gonna love it!
Having lived and working in Australia, Europe and the US, I can tell you that international taxation is a nightmare. It's primarily based on residency, you have to tell the tax office where you were for every day of the year.
I don't see anything that would stop you living anywhere you want, even some nice tax haven.
The Egyptians thought of building a restaurant at the Cairo airport years ago. It's across the street from the airport Moevenpick hotel. It's kind of tacky and has corrosion in places. If I remember rightly, it is (was?) a 707.
I went there with some friends, all airline employees on business in Cairo, for dinner. The memorable part was when a guy laughed so hard after a joke someone told that he ripped the seat from the floor and fell in the aisle. Oh well, you had to be there...
Most commercial databases prefer raw partitions and you don't run a file system at all. The database manager has log files for commit/rollback processing. They run faster on raw partitions, since the OS is not "in the way". The DBMS actually doesn't need much of what a general purpose OS offers, that's why Oracle was promoting the "raw iron" concept (BTW what ever happened to that?).
Last I heard, maybe it's changed, Linux didn't support raw partitions.
This may be cool. Once they figure out what the eyeballs send to the brain, they might be able to make an artificial eyeball.
I hope the artificial ones don't come in a visor that looks like an air-filter from a '75 Buick.
Has PC Week coughed up the cash? And, if you're listening, how does a geek spend this well earned money?
Actually, on-line reservations were done in about 1960. I have seen code from the Sabre reservations system that was originally written in 1972.
This patent will never be enforced, Easy Sabre was selling reservations on Compuserve in 1980's.
I don't see what's groundbreaking about this. Our cat has a microchip implant, it was done about 2 years ago. If she gets lost, any veterinarian in the area can check her with a scanner and have her safely returned.
I don't enviseage having lights turn on for her when she enters a room, but it would not be difficult to do.