4 ms·

Doesn't matter: the industry has moved-on. They had to make it open decades ago. Now it's too late.

gnu8 6y ago
I don't think that is really accurate; nothing processes transactions like an IBM mainframe. To draw a crude automotive analogy, no matter how much you soup up a truck, it is never going to be a locomotive. It is simply a different scale of machine.
- astrange 6y ago
  Are there actual objective benchmarks for things like this? Everything you actually see out of IBM is so bad (blockchain 5G AI cloud nonsense) and Moore's law exists, so it's hard to believe this stuff is that good as opposed to just having a really good moat. I am surprised how many government/bank systems run on 50s-70s code; it makes you wonder if we kept in practice writing those kinds of things since then.
  - bawolff 6y ago
    IBM is a shell of what it used to be. I don't know if what the grandparent is saying is true, honestly i highly doubt it, but its unfair to compare old ibm with its current incantation of marketing fluff and bad high priced consultants.
  - rbanffy 6y ago
    > Are there actual objective benchmarks for things like this? Probably not, but they are machines designed for throughput, with a multitude of intelligent peripherals that pipe data in and out of the CPU, which will be running at close to 100% speed all the time. The CPUs themselves are quite impressive (they present at Hot Chips regularly) and each cluster of 4 sockets share almost a gigabyte of L4 cache. Everything is also constantly checked, audited, logged for security and reliability. I wouldn’t be surprised if these machines thoroughly smoked a rack of souped up x86s at a similar budget in transaction processing applications.
- 908B64B197 6y ago
  Problem is, it's a machine that's from one and one vendor only. You can buy a truck anywhere and run it on almost any road. The locomotive has to be from the one locomotive company and good luck running it outside of (really expensive) tracks.
anthk 6y ago
Not even close. Mainframes can hot swap stuff while running jobs.
- lstamour 6y ago
  So can distributed systems? All you need is a queue in the right place or some other kind of data store? It’s relatively easy to decouple data models from processing, you just have to consider failure scenarios? I mean, the distributed non-mainframe way of doing things is you just use more units of hardware to make up for deficiencies? I imagine the hardest thing to port is well, the equivalent to “if we scale up this machine and buy lots of RAM, our data fits in RAM and we can run this code that much faster,” but even then, can mainframes really handle that much more data by virtue of being mainframes with all their legacy? The way I’ve heard it described, it’s not the features of a mainframe that sell it now, those are commodity. Instead it’s the existing code that ties companies to mainframes - they don’t want to rewrite logic they likely don’t actually understand the details of if it’s already been tuned and tweaked to work for their business? To that end the future of mainframes is probably virtualization of hardware and improved static analysis tools to better support codebases we don’t understand because they’re decades old, etc.?
  - rbanffy 6y ago
    You are right in the sense that a modern mainframe is reminiscent of a cluster of specialised computers connected through multiple ridiculously fast interconnects. The difference is that IBM offers the software you’ll need to manage the environment.
  - toomanybeersies 6y ago
    Aren't we facing the exact same problem, but now rehashed in the cloud? A lot of companies are running applications that rely heavily on AWS managed services. If you're tied to DynamoDB or Athena, you're in no better position than being tied to IBM mainframes.
    - lstamour 6y ago

gnu8 6y ago

I don't think that is really accurate; nothing processes transactions like an IBM mainframe. To draw a crude automotive analogy, no matter how much you soup up a truck, it is never going to be a locomotive. It is simply a different scale of machine.

astrange 6y ago
Are there actual objective benchmarks for things like this? Everything you actually see out of IBM is so bad (blockchain 5G AI cloud nonsense) and Moore's law exists, so it's hard to believe this stuff is that good as opposed to just having a really good moat. I am surprised how many government/bank systems run on 50s-70s code; it makes you wonder if we kept in practice writing those kinds of things since then.
- bawolff 6y ago
  IBM is a shell of what it used to be. I don't know if what the grandparent is saying is true, honestly i highly doubt it, but its unfair to compare old ibm with its current incantation of marketing fluff and bad high priced consultants.
- rbanffy 6y ago
  > Are there actual objective benchmarks for things like this? Probably not, but they are machines designed for throughput, with a multitude of intelligent peripherals that pipe data in and out of the CPU, which will be running at close to 100% speed all the time. The CPUs themselves are quite impressive (they present at Hot Chips regularly) and each cluster of 4 sockets share almost a gigabyte of L4 cache. Everything is also constantly checked, audited, logged for security and reliability. I wouldn’t be surprised if these machines thoroughly smoked a rack of souped up x86s at a similar budget in transaction processing applications.
908B64B197 6y ago
Problem is, it's a machine that's from one and one vendor only. You can buy a truck anywhere and run it on almost any road. The locomotive has to be from the one locomotive company and good luck running it outside of (really expensive) tracks.

anthk 6y ago

Not even close. Mainframes can hot swap stuff while running jobs.

lstamour 6y ago
So can distributed systems? All you need is a queue in the right place or some other kind of data store? It’s relatively easy to decouple data models from processing, you just have to consider failure scenarios? I mean, the distributed non-mainframe way of doing things is you just use more units of hardware to make up for deficiencies? I imagine the hardest thing to port is well, the equivalent to “if we scale up this machine and buy lots of RAM, our data fits in RAM and we can run this code that much faster,” but even then, can mainframes really handle that much more data by virtue of being mainframes with all their legacy? The way I’ve heard it described, it’s not the features of a mainframe that sell it now, those are commodity. Instead it’s the existing code that ties companies to mainframes - they don’t want to rewrite logic they likely don’t actually understand the details of if it’s already been tuned and tweaked to work for their business? To that end the future of mainframes is probably virtualization of hardware and improved static analysis tools to better support codebases we don’t understand because they’re decades old, etc.?
- rbanffy 6y ago
  You are right in the sense that a modern mainframe is reminiscent of a cluster of specialised computers connected through multiple ridiculously fast interconnects. The difference is that IBM offers the software you’ll need to manage the environment.
- toomanybeersies 6y ago
  Aren't we facing the exact same problem, but now rehashed in the cloud? A lot of companies are running applications that rely heavily on AWS managed services. If you're tied to DynamoDB or Athena, you're in no better position than being tied to IBM mainframes.
  - lstamour 6y ago