Slashdot Mirror
IBM Announces First Linux-only Mainframes
A reader writes "The new Z-series mainframe for Linux, which costs $400,000 and is aimed at processing transactions at large businesses, is IBM's first mainframe computer sold without IBM's traditional z/OS mainframe operating system. More info at the IBM zSeries page" This is something that IBM and others of Big Iron vendors of *NIX have said - as Linux grows in maturity, they want to replace their *NIX with Linux. However, there's still work to be done in that area.
The link to the SourceForge Foundry is slightly broken. Correct link is here.
Does it support Hot Swapping?
I would think hot swapping would be one feature truely worthy of a mainframe operating system... especially if you can all of the different possible parts of a mainframe and still keep all of your applications running 24/7.
The article cites cost concerns, but how much does using a linux reduce the price of a $400,000 machine? (Cost of ownership may well go down, but I'm asking about purchase price.)
Prime numbers are exactly what Alan Greenspan says they are -S. Minsky
(nb: The last IBM big-box I worked on was a first generation AS400 so this question may be dated)
I recall licensing of IBM's OSs to be fairly expensive, have they cut prices at all to reflect the fact that a lot (the bulk?) of the vanilla Linux development happens outside IBM, therefore costing them nothing?
Trolling is a art,
More coverage from the reg
Cyborg_monkey
> what's a server?
A person who if you don't tip them when you leave the restraunt the next time you visit will spill soup all over you?
Or the pile of junk in the corner of the office that makes alot of noise, has various people standing over it and muttering dire curse relating to bill gates and all in the computer industry (assuming os = Windows) or in the case of linux... now where did I leave that boot stone-slate as its so rarley needed...
--- Users are like bacteria -> Each one causing a thousand tiny crises until the host finally gives up and dies.
I'd just like to correct something here: they aren't replacing the previous zSeries operating system, they're adding another choice. Now you can choose between z/OS, z/VM, and Linux. While there is something called Unix System Services that run within z/OS, it's not a stand-alone operating system; it's rund under z/OS, not by itself.
And with Linux, you do loose a lot of the RAS characteristics that z/OS provides, as well as 40 years of compatibility with existing workloads. Linux is being sold as something to run new workloads on, workloads that z/OS previously wouldn't have been considered for.
Silly signature limit . . .
"Nobody ever got fired for buying IBM" has today been replaced with "Nobody ever got fired for buying Microsoft." However, in the case of the IBM iServers and zServers, Linux is replacing a proprietary Unix, not a Microsoft OS.
This is a step forward for Linux (although perhaps a smaller one that at first glance, because you already could get IBM servers with Linux--these are just the first Linux-only servers) but not a step backwards for Microsoft.
That seems to be the trend now, anyway--remember when Amazon said they saved millions of dollars by using Linux? Those Linux systems replaced Unix systems, not Microsoft Windows systems.
ZDNet have a recent story about a company called Boscov's Department Stores replacing a lot of NT machines with one IBM zSeries. From the article: "Boscov's, with 36 locations in six states in the mid-Atlantic region, scrapped its client/server architecture and is in the process of consolidating 70 IBM NetFinity 8500 and 500 servers running Windows NT 4.0, on a recently purchased IBM zSeries 900 mainframe running SuSE Linux Enterprise Server 7 as a virtual machine."
and haven't touched z/os at all ... but was it a 'nix?
Jack Valenti and the MPAA are to technology as the Boston strangler is to the woman home alone
No. z/VM is the 'meta-OS'. It's pretty much analagous to VMware in what it can do, in terms of hosting other OSs underneath it.
z/OS is geared at high volume transaction, database, batch processing. it runs either z/VM or more typically natively or in an LPAR.
An LPAR is a 'logical partition', a way of dividing a m/f up into several virtual machines.
for now, these are static and implemented when a partition is 'booted' - IPL'd (initial program load) in m/f terms.
VM on the other hand supports hundreds, even thousands of dynamically generated virutal machines. You can run VM inside an LPAR providing two levels of partitioning. I expect VM and LPAR technologies will converge at some future point.
meanwhile everyhting can talk to each other over 'hipersockets' - memory to memory pipes that looks like a tcp/ip network to your software - blindingly fast
Is the mainframe wireless? Is it handheld?
Consulting Times has a article which gives a "real world" cost justification example.
I will not be pushed, filed, stamped, indexed, briefed, debriefed or numbered.....my life is my own.
Will this mean that IBM will finally replace OS/2 as the bootstrap and control server?
Replacing that with Linux would be a nice start!
For those that do not have the benifit of a 390 sitting behind them, it is very disconcerting to have that big black IBM monitor on top of it, because it is running OS/2 on a Celeron board inside the mainframe to control the whole show.
Architecture is the key. What's the difference between a 120 MIPS mainframe and 3000 MIPS desktop, and why is the 120 MIPS mainframe faster in mainframe type applications?
Architecture. Specifically, things like I/O, process handling, etc.
Don't get me wrong, I'm a strong believer that "desktop" type hardware can compete with the big boys, especially considering the cost diferences and the extra speed, boxes, redundancy, etc that you can buy with all that cash you save. But... there are times when the big mainframe architectures really do have a reason for being.
Just my $.05 (inflation, you know).
-- "Other than that, how was the play Mrs. Lincoln?"
IBM has made some bbbbbaaaaaadddd choices in software on the desktop over the years, but will stick linux to the forefront, they are advertising the hell out of it and this is good, it gives managment a confidence in Linux that would be nearly IMPOSSIBLE to gain elsewhere.
They did not make a bad choice in developing OS/2. They just outdid themselves. The win 3.11 compatibility was probably part of the reason there was so little OS/2 native software available. Microsoft didn't develop for OS/2, and they already had the "standard" for office suites.
OS/2 was (technologically) about 8 years ago where Linux wishes to be in the future. Only it wasn't open sourced and free.
On a low-end pentium they made an OS that would rock your socks with voice recognition, stability and a kick-ass shell. Allegedly, OS/2 scales like a champ if you stick multiple processors in it.
However, the market wasn't there. Why get a new OS to run windows? Now you'd need 2 os licences to run word!
Stop the brainwash
If a $400.000 server has (virtually) no downtime, and the $2000 server has several hours, that could really make a big difference in the balance sheet.
Not only does downtime mean lost transactions, it could also mean lost customer confidence.
Also, your $2000 estimate is off. A $2000 pc server, WITH a backup unit?
Stop the brainwash
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.
But most their savings are due to improved scalability and easier maintenance (especially for disaster recovery).
Read the article, all the arguments for the switch are there.
Store chain is sold on Linux [ZDNET]
Black holes occur when God divides by zero.
Not to sound like flamebait, but there have been alot of issues with 2.4 lately, it doesnt really seem stable enough that i'd put it on my mainframe, theoretically speaking. Problems range from fs corruption to sync() bugs, etc. Sure, its a nice desktop OS but I don't think it's ready for the mainframes.
I have been working in the mainframe world for a few years now, and one thing you have to understand about mainframe operations, is that since it's conception the #1 priority is UPTIME. Speed was number 8 or 9.
Only recently (last 7 years) has speed been a considiration, and that was thanks to the PC revolution. But again, you were alwsys dealing with two camps: Mainframe guys, and PC guys.
So all this means is that there is another choice for people who want the " 5 9's",the holy grail of computing, and not Windows, Unix or any other platform other than the mainframe can deliver that.
That $2000 server will have nowhere near the hardware reliability of the $400,000 mainframe. When the hardware fails on a mainframe it is a dire event, resulting in a team of engineers being put on the next flight out to the customers site.
Best Slashdot Co
(Shot of IBM's new server standing alone in a server room)
ANNOUNCER: "If you think we're overcompensating for something with our really, really big mainframe running linux..."(Cut to shot of a dozen small servers being carted off) "...You're absolutely right."
PCs crash a lot. They're made from cruddy hardware because the average consumer either doesn't know the difference, doesn't care, or can't afford anything better. Mainframes have uptimes in the years; some have benn going for decades. They usually have hot-swappable everythings, including the usual power supplies and disks, but also hot-swappable CPUs, memory, expansion cards (network, etc), and even motherboards sometimes. Finally, they have a high degree of self-awareness. Today's PCs are starting to get some of these features (your BIOS might know the speed of the CPU fan, wheeee) but the mainframes are way ahead. They're set up to figure out when things are about to fail. When a potential failure is detected, the mainframe will call the vendor and order replacement parts automatically. A service tech will usually be there within hours to replace the part, and the part will be taken back to the lab to see why it failed. The knowledge gained from the failing part is used to design the next revision so it doesn't fail.
When it comes down to it, CPU power isn't all that important in the mainframe world. They do a shitload of I/O, and they just work. An Athlon XP might run circles around a mainframe in Quake 3, but its components are slow and unreliable.
The average 16 processor mainframe is a 120 MIPS machine, whereas the average 1.5 GHz desktop system is a 3000 MIPS machine.
Mainframes run up to about 200 MIPS per processor and with multi-processor overhead a 16-way zSeries tops out somewhat below 3,000 MIPS. These are mainframe MIPS, not what you get as BogoMIPS out of Linux at boot (AFAIK, this is some quick integer timing loop calculation). There's a reason it's called BogoMIPS, troll.
IBM has successfully run over 40,000 Linux images on a mainframe (under VM). Try that on your 1.5Ghz desktop. Ever heard of Transactions Per Second (TPS) in four and five figures, I/O rates in GB/sec, multi-terabyte databases, 99.999% uptime for years? That's mainframe territory, and I sincerely doubt that you've ever seen it, or ever will.
Exactly. The point that most
A good engineer picks the best tool for the task at hand. Depending on the computing task, the best computer could be a mainframe, a MPP supercomputer, a commodity SMP server, a cluster of desktop PCs, or some other specialized architecture.
Commidity x86 hardware is great, and can do an acceptably good job on a wide variety of tasks, but it isn't the be-all and end-all of computers. Just because you haven't worked on anything else doesn't mean that other computer architectures are outdated crap.
Why is it that the proponents of "one nation under God" are so eager to get rid of "liberty and justice for all"?
Without Linux, how long would it take for Microsoft to take the whole server market away from Sun? The trend was set when they released two different versions of NT, "workstation" and "server". The fact that they don't have any version called "server" anymore may reflect a reality check they have done, realizing it will not be so easy after all.
As I said above it was mistated, not cash, but cash value assets.
:)
If you figure in interest on those amounts you will see as on average it is correct.
I also didnt mean there would be a lot left after 50 years
This at one time was actually a detailed finacial study done circa 1985 at that time it was longer...BUT ibm also had more long term high $ support contracts.
IBM has been in a better cash position before, but the value of not only IBM is in its physical property but its IP as well, IBM is probably one of the largets companies in the world in that respect.
Sig went tro...aahemmm.....fishing........
Amazon had a mix of Unix and M$ (more M$ than Unix).
Burlington was mostly Unix.
Boscov had 70 aix and >500 M$.
telia dropped mostly Solaris.
Home Depot is apparently going to drop all M$.
more and more are showing up and while they are replacing some unix, it is also replacing an equal or bigger percentage of M$.
As the economy worsens and the companies that are making profits are running linux, well...
It is exactly what happened in the late 80's early 90's when M$ was the correct way to go.
...we don't see too many Anonymous Cowards claiming that Linux is a "toy" operating system in this particular discussion.
I guess 400k$ is a little expensive for a toy!
Reminder: find a new sig
Lots o' programmers out there waiting for GNOBOL
Though I'm waiting for HIPERSOCKETS which would allow me to afford better use of OSA.
Is WLM support working yet?
Have they licked the scheduler problem yet? That was an inherent problem of the Linux kernel expecting to be the only OS instance on the hardware and constantly grabbing the clock to do more or less nothing.
Next stop - Checkpoint firewall code on a Linux instance on the mainframe and goodbye to that gated-ipchains crap.
- It's not memory - we've got half a gig.
- It's not IO - a mainframe has 16 channels, as opposed to one (ethernet) on the PC.
- It's not the size of data transfer(no graphics, just text).
Which leaves us with a CPU bottleneck. The problem is that the CPU's simply can't handle the processing load. There comes a point in time in which an efficient architecture can only do so much, and our shop has reached that point. Currently, we have jobs that we can't run because of the load they place on the system. Our operations are greatly constrained by the speed our 60 MIPS, 500,000 dollar machine. For half a million dollars, we could have a rack full of servers, almost no latency, and the freedom to run jobs on demand, rather than off peak hours.To be honest, it's a shame that IBM hasn't kept pace with current technology. Mainframes are very well organized internally, and a good example of how a machine architecture should be designed. But I feel kind of betrayed by IBM in that the cost of a mainframe is not commeasurate with its computing power - and corporate America is starting to notice...
The society for a thought-free internet welcomes you.
My company purchased one for web hosting
... are switched.
:)
The system uses VM as a base but has multiple instances of SuSE running. It is able to run up to 10000 instances of Linux which makes it a data center in a box.
There is no bus and the communication between the processor banks, memory,
First time I've seen it my eyes jumped out of the sockets.
Good Job IBM
The big vendors (including IBM) never said that they wanted to replace their proprietary unix systems with linux. IBM said (in a very marketingish type of way) that if Linux could do all the things AIX did, they would consider it.
In addition, AIX never ran on the zseries computers. So it has nothing to do with a mainframe running linux. The two are separate issues.
This is good news for Linux; but its not accurate to say that it has anything to do with linux displacing AIX, or any other unix.
Hot Damn! It's the Soggy Bottom Boys!
It's not the maintenance that is the problem, things like configuration management and data integrity are more important. If you have a hundred servers, then you have a hundred places to check that everything is in sync. If you are running a small shop with a dozen or so machines and one administrator then they can keep all the state in their heads. When you get up to hundreds then the state is larger than one person can easily cope with and you start having to communicate state to others. With hundreds of boxes, it is easy to overlook things, with fewer boxes, the communication is easier, and cheaper.
The other thing is CPU residency. Lots of small boxes wastes CPU power because they tend to be devoted to one task and are only capable of that task. The problem is, they are so small that you can't add other tasks to them so you need a new box... Generally, CPU residency on small boxes runs about 10%, with mainframes, this can rise to 90%. Take two tasks - one runs during the day, one runs during the night. Conventional wisdom would allocate two small boxes, one per task wasting them for most or their life. Mainframe usage would run them both on the mainframe - this gives each process more power when they run and doesn't waste the box when they don't. Most traffic tends to be peaky but only for a short period of time so if the box is large enough to hold them both, you get a saving whilst still making all the tasks faster.
Small boxes are good when you need maximum cycles per buck and the task is easily partitionable with minimal interprocess communication and the tasks are continuous. When the tasks are not easily partitionable, need lots of IPC or are peaky then larger boxes make sense.
The thing to remember is that where the scale is large, you need to make use of that scale to get maximum performance. You don't see chemical plants using hundreds of small vats, they use a few really big ones. With these systems they are used at a scale where communications and simply keeping track of what is going on is a major exercise and hence a major expense.
My Experience? Well - put it this way, the SunFire 6800 turned up a few weeks ago, the 4800 turns up on wednesday as part of a plan to replace a Tandem mainframe and they will be sitting next to quite a few racks holding Sun E3500s, E450s, E250s, t1s, HP netservers, IBM RS6000s and SGI Origin 2000s and indeed a MacOS server or twenty. A lot of our comms talk to Stratus mainframes and the machine room cooling plants are a more pressing problem than CPU speed.
hmm yeah - I'm no expert on Mainframe architecture, but from what I've read - it's down to pure I/O width, and massive redundancy/hotswap, belt&braces style robustness.
:-/)
:P) - we stopped buying huge multi CPU boxes, to handle a specific load - and re-designed our web server clusters to use many smaller (1U) rackabble boxes for all tiers of the system from front end caches, load balancers, firewalls, JSP processors and even the database nodes (with shared disk arrays). Need more back end database? Clone a few more 1U DB servers and connect em up! This meant we could stop worrying about how much traffic we would be getting to the sites so much - if it turned out we'd underspecced, we could add some more quite easily.
I also agree with you that "desktop" style machines running something like Linux *can* offer similar levels of reliabilty and performance, but in a completely different way. In a nutshell - instead of one ultra-robust machine with multiple redundant sub-systems, you go for multiple redundant machines (although you could define the cluster as the machine - in which case it's no different...hmm
I've successfully applied this pet theory of mine over the last 3 years wherever possible. Even things like ethernet switches - we used to buy Cisco 550X chassis which come with 2 of everything important, like PSU, routing module, supervisor module, backbone interfaces and so on, but they cost £35K each for the config we typically buy. Sure they hardly ever fail, and if a component fails, there's a backup. However - recently we started buying smaller cheaper swicthes - but lots of them - typically 3 where 1 would do: total cost about £15K for the same scenario
Web servers lend themselves easily to this too (especially if you use Apache and Tomcat (or whatever it's called this week
I always thought that IBM continued developing the Mainframe to support existing OS/390 customers with large complicated mission critical apps on them - I can see some use for a mainframe running Linux (and I bet their are more Linux savvy techies otu there than z/OS - which would help with recruiting admins for the box), but I still feel that the multiple-smaller-boxes-running-linux solution is a better bet - as it can be any size you want within reason - start off small for dev/testing, and then pile on the hardware for production.
I know that it doesn't highlight linux so much, but it's nice to see linux dunk the ball once.
I love the part where the 'middleware' character doesn't get any fan mail. No one wants his autograph.... hilarious. Even my computer-stupid girlfriend loves it.
Whoever does those IBM commercials is a genius.
My vote for the Super Bowl:
ANNOUNCER: "Now, All Your Base Are Belong To US!"
[shows a zSeries]
"Imagine a beowulf cluster... of these babies!"
Get your Unix fortune now!
Doubt that Sun is sweating bullets. They like Linux since it means that, as you pointed out, shops are filled with cheap linux boxes not cheap MS boxes. The shift from Linux to Solaris isn't that hard. When Sun will get worried is when Linux is really are more solid OS for high end computing, but then Sun will just need to make the switch over. It makes it's money much more on the hardware and 'systems' end of the things than selling a proprietary OS.
http://www-1.ibm.com/servers/eserver/zseries/900op en.html
Check out this page for a laugh... IBM says that their new servers will let you run 31 bit applications!
This site looks like it's got a fair few useful links.