Hacker News Reader

Comment by bombcar | Hacker News Reader

Yeah, mainframe coding is much closer to "code NASA runs on spacecraft" than what most people consider coding to be today. And the level of hardware support (mainframes are basically redundant on everything) and software support is unheard of. Consider that these things are so integrated into major banks, etc, that even minutes of downtime would cost millions upon millions, and think of all the things you can spend millions on to reduce the chance of it happening.

View on HN

> mainframes are basically redundant on everything

Not only that but that redundancy is hidden from everything. Already decades ago the Hungarian electricity distributor had a "RAID" storage unit where one bunch of disks were at the capitol, the other bunch connected with fiber at another city -- and you talked to this thing as if it were an ordinary disk. Short of an invading army nothing could cause data loss on that thing -- simple aerial bombing wouldn't because the unit in the capitol was tens of meters underground... that's not concerns you have on your average web app.

Yeah, one thing that I've been really disappointed with is how UNRELIABLE the "cloud" is - unless you engineer all the reliability in the software itself.

We have the technology and the capability to do things like run a VM and have it move between hardware when that hardware fails or needs replacement.

but that would cost money.

VMs can often be configured to do "live migration" but that doesn't protect a running instance if the hardware it is running on fails completely. There is such a thing as fault tolerant hardware/software stacks but, yes, most modern architectures shift application resiliency higher up the stack.

> the capability to do things like run a VM and have it move between hardware when that hardware fails

This is one of my favorite features of Google Cloud:

https://cloud.google.com/compute/docs/instances/live-migrati...

Unfortunately, Google continues to be the third-string provider in the US, so I don't often get the opportunity to work with Google Cloud.

I want to be able to justify using Google Cloud for projects, but the fact that it’s a Google product is the main obstacle to that. If history is anything to go by, relying on Google products and services seems like a great way to set yourself up for major problems down the line. Either the product you’re using is shut down on short notice or some automation flags a false-positive and you’re left screaming into the void trying to get yourself unbanned.

It’s not a live migration. It’s running an a VM in a sysplex which is in lockstep with its peer. There’s not a failover in the tradition sense..the I/O just goes from one side to the other.

Reply on news.ycombinator.com