Hacker News Reader

Comment by Tichy | Hacker News Reader

What are these things used for? Why can't they ever go away? I don't really believe it...

They're used by people running massive systems that want all of their data to be accessible at incredibly high speeds.

We're talking about systems designed for throughput. For processing huge amounts of data in overnight batch processes. For five nines uptime. And for still being compatible with the code you wrote 30 years ago that still manages millions of customers.

Right, and incredibly high speeds is definitely accurate. I can't remember the amount of instructions/second our mainframe can handle, but I remember it being in the multiple hundreds of thousands.

Several hundred Kilohertz? :-)

It IS incredible - but modern Computers are even more incredible.

Oh, don't get me wrong, I'm not defending it. It frustrates me on a daily basis, especially without having a relational database for some critical information.

:) I don't think he's talking about clock cycles...

How fast would one of those monsters be compared to, say, a modern day Mac Pro?

The actual CPUs are pretty impressive, and operate at 3-4.5GHz range. The system will have up to ~80 of these, so it would effectively blow away a Mac Pro. Let's be honest though, at the prices IBM fetches for these (millions for the top end), you could build a VERY healthy Hadoop cluster.

The reason why these are kept around is that the custom software that runs on them often implements very complex business rules that are otherwise undocumented (the code is the documentation). If you're talking about a financial institution, the mainframe IS the business.

As far as reliability, these things are rock-solid and proven, but mainstream distributed systems design has pretty recently surpassed the mainframe model IMHO. It's relatively simple (compared to just a few years ago) to implement clustered data processing using MapReduce, replicate the data across multiple datacenters, and not care about individual node failure.

80 processors? Big potential for writing concurrent apps. Do mainframe developers exploit that?

When you consider that not only do these systems run a lot of batch processing daily, they also have user interfaces through things like CICS, which users are allowed to login and initiate jobs. Most of the time it has it's own form of concurrency in which it will lock certain datasets for processing and release them when it's done. Of course, it's still possible to get deadlocks and other issues because of this.

The architecture is very different.

You have every piece of I/O handled by its own subsystem; in the old days when they had lots of green screens they had FEPs, front end processors, that took care of all the busywork for displaying and accepting data - the main CPU never saw any of that. Same thing with networking, with disk I/O, you name it.

Reply on news.ycombinator.com