The Eternal Mainframe 225
theodp writes "In his latest essay, Rudolf Winestock argues that the movement to replace the mainframe has re-invented the mainframe, as well as the reason why people wanted to get rid of mainframes in the first place. 'The modern server farm looks like those first computer rooms,' Winestock writes. 'Row after row of metal frames (excuse me—racks) bearing computer modules in a room that's packed with cables and extra ventilation ducts. Just like mainframes. Server farms have multiple redundant CPUs, memory, disks, and network connections. Just like mainframes. The rooms that house these server farms are typically not open even to many people in the same organization, but only to dedicated operations teams. Just like mainframes.' And with terabytes of data sitting in servers begging to be monetized by business and scrutinized by government, Winestock warns that the New Boss is worse than the Old Boss. So, what does this mean for the future of fully functional, general purpose, standalone computers? 'Offline computer use frustrates the march of progress,' says Winestock. 'If offline use becomes uncommon, then the great and the good will ask: "What are [you] hiding? Are you making kiddie porn? Laundering money? Spreading hate? Do you want the terrorists to win?"'"
Deep (Score:3, Insightful)
Wow, so deep. Computer is the Internet, Internet is the computer.
Mainframes are specialised equipment, server farms are almost generic computers with redundancies. The real difference is the cost. Today's server farms would cost many factors more if they were built with specialised mainframes, there is no other real difference, they are really there for the same purpose.
Running Virtualization Software.... (Score:4, Insightful)
Just like the mainframe. [wikipedia.org]
Privacy (Score:5, Insightful)
That "may" be true if the authorities are not abusing their power, or trying to gain more power than the people want them to have.
As soon as you have even a potentially oppressive regime, privacy becomes essential.
Ending maintenance also ends control (Score:5, Insightful)
One of the points I found the most insightful is that the geeks don't like to take the time to make things work anymore. I remember a colleague saying that there was no better way to kill a hobby than to get it as a job.
The days of tweaking the OS and hardware as a common practice among the majority of geeks is gone. The field is too broad now. You have to pick which stack, and where on it, you want to hack.
Mainframes is for those.. (Score:5, Insightful)
..that have very big amounts of data, complex data structures and can't afford any errors (especially data corruption) caused by hardware limitations.
Banks is an example.
No because (Score:5, Insightful)
Are you making kiddie porn? Laundering money? Spreading hate? Do you want the terrorists to win?
Because I don't want every goddamn marketer out there trying to sell me their shit. I don't want to have to deal some horseshit like this [forbes.com] because businesses feel entitled to stick their noses into my business.
No, you are NOT offering me "convenience" - you are prying.
As it is, I CAN create a dossier that would make an East German Stazi agent cream his pants by just hitting the credit bureaus, Google, ChoicePoint, ISPs, Cell phone companies, and every other business entity out there that has this need to collect consumer data.
Something to hide?
Well, just ask the atheist, gay or lesbian, peace protestor or Muslim who has their identity known what happens to them.
The uncle of the Marathon bombers who had his face plastered all over the place is headed for some serious shit. You just know that folks are going to vandalize his house, harass him, and give him a lot of shit just because he's related to those kids and a Muslim.
People are hateful, ignorant, cruel, shallow and just stupid - until proven otherwise. Therefore, it is imperative to keep one's secrets.
Re:Deep (Score:4, Insightful)
For a certain type of workload, at a certain level of necessary uptime, mainframes start becoming cost effective. Fun things like where IBM will install as many CPUs as you want, but only charge you for their time when you use them. This can be very cost effective for businesses with seasonal volume shifts. At some point, paying IBM $1000 an hour for their support is cheaper than paying 20 creeps with greasy hair to change hard drives, stack servers into a rack and fuck up the rollout of new VMs. It's kind of like trucks versus trains. Each have their place, but neither is very good at emulating the upsides of the other.
Rinse, lather, repeat (Score:5, Insightful)
Not networked, networked, not, networked, on and on. Each cycle begets a new cycle. Now it's just called "the cloud."
Mainframes and server farms the same? Precisely. (Score:2, Insightful)
The general thinking of comparing the two is that both systems are the ones running running the show, storing the data, and being accessed by dumb-clients that only serve as terminals.
Obviously server farms and mainframes are very different from a back-end technology standpoint, but from a viewpoint of the user they are identical in every single way. You log in with your user specific credentials, you do your work using the server's processing power and save your work in the servers storage medium. Your client likely is even set to network boot from a server supplied boot image via PXE. If your local machine is nothing but a terminal to access the backend machine, then you are for all practical purposes operating in a mainframe environment.
Re:and the mainframe never went away (Score:5, Insightful)
Besides the cost, banks are also averse to risk, and change causes risk.
Wait a minute: did you somehow sleep through 2008? Banks love risk, so long as it's someone else's money they're churning.
Re:Mainframes is for those.. (Score:4, Insightful)
The people who use mainframes for big data (like banks and insurance companies) and the people who use clusters and racks of servers for big data (like search engines, social networking sites and other web companies) have totally different requirements.
Valid Observation, Poor Argument (Score:4, Insightful)
Mr. Winestock's parallels between server farms and mainframes are reasonable, if unoriginal, and the same can be said for his concerns over privacy and social control. His attempt to claim the former as the causative agent for the latter, however, goes wrong right from the start: 'Mini/micro-computers were supposed to kill the mainframe.'
Not so. They came about firstly because technological advances made them possible, and also because some smart people realized that they would allow us to do things that, in practice, we could not do before. The pioneers of these developments were not interested in reproducing, much less replacing, mainframe computing.
Turing showed us that the form of our hardware doesn't dictate what we can do with it. To understand the arc of privacy erosion and social control, we need to examine social history and human nature, not the artifacts of technological advance.
GOOGLE DOESN'T CARE ABOUT ERRORS !!! (Score:4, Insightful)
"Google, for example, has pretty full-featured job control layered on top of their server farm."
Google has never cared about errors.
Who gives a damn if what absolutely positively SHOULD have been the very first result is instead the fourth or the fifth result, or if it appears on page two of the results, or if it somehow magically disappears into the ether because commodity server #XJ42 in rack #43HB on aisle #521JJ in column #447F in building #QQZ1 in server farm #H61M happened to have crashed just as the query response was being assembled?
Especially if the query involved "Justin Bieber", "Lindsay Lohan", or "Natalie Portman Hot Grits".
IBM, on the other hand, has always cared about errors - has always, in fact, been FANATICAL about errors.
If you send a query to an IBM mainframe, then you're expecting umpteen-sigmas of confidence that the mainframe will actually be up and running, that you'll get an actual response, and that the response, when it finally arrives, will be 100% CORRECT.
Especially when the response is something along the lines of "DANGER: CHILD KNOWN TO BE ALLERGIC TO AMOXICILLIN. ALLERGIC RESPONSE INCLUDES ANAPHYLACTIC SHOCK. PRESCRIPTION REQUEST THEREFORE INVALID AND REFUSED."
Last TCO I was involved with... (Score:4, Insightful)
You haven't tried the IBM kool-aid yet. Those people whose jobs currently rely on mainframe expertise are very happy with them. They do have better error-checking but everything else is at least an order of magnitude out of whack with commodity hardware price/performance, and in many cases, several orders. You can reduce some of the costs on their zSeries by buying specialised processors for DB2, Java, and Linux (~100K a pop) so you don't have to may for MIPS usage but the costs are still astronomical for the performance. If it was cost effective, don't you think Amazon would be running its cloud services on them?
The last TCO I was involved with actually showed that the mainframe was the more cost effective approach for the use case at hand.
As for Amazon, well that is hard to say. If when they first started, they knew how successful they were going to grow and how quickly, maybe they would have gone with a mainframe solution.That's the nice thing of TCO analysis, it eliminates, or should eminiate, any platform bias the decision makers have. Then again, it also depends on really knowing what future growth patterns and expected use cases are or it is just more GIGO.
Re: GOOGLE DOESN'T CARE ABOUT ERRORS !!! (Score:4, Insightful)
You have no idea what you're talking about. Dropping the occasional search result is fine, but what about failing to record billing for the ad system, dropping mails you were supposed to receive in your Gmail account, or failing to save the doc you were editing? Google does a lot more than serve search results, and most of that needs to work every single time.
The fact of the matter is that even the most expensive hardware eventually fails, so your software needs to be able to deal with it and fall back to working units. Once you've written your software to handle hardware failures, you can run on really cheap hardware. And, it turns out that buying a lot of really cheap computers some of which are broken all the time gets you way more computing power than trying to buy a few really robust machines.
Re:Giving up the dream (Score:2, Insightful)
There was a time when we expected computers to become so easy that everyone could use them. We've given up that dream. Now it's all "managed" again.
I blame the users. If they bothered to learn even a little about how things work, they wouldn't give up their freedom so easily.
Unbelievable. You blame the USERS for computers being hard to use.
You just demonstrated what went wrong with computing. Right. There.
Computer geeks: your libertarian paradise is not what other people want or need in their daily lives.