Re^6: Memory addressing

We have legacy hardware that would cost $50k to $100k to replace. The hardware itself is excellent and includes modifications of our own and so would be very hard to replace anyway. However the interface is designed for win3.11, we have split it and run some of the interface functionality on win2k server but we are left lugging 500Mb - 2Gb files accross the network every few minutes which is a pain. The ideal would be to run the whole thing from win2k but we can still improve things physically - alter our network setup, etc. Writing a device driver may be the way to go since I would also see it as beneficial as a learning excercise. I'm looking on Amazon for "writing windows device drivers for dummies" but I fancied a bit of a poke at it with perl cos thats where I'm most at home. Sounds like the prognosis is write a device driver in C or do something else instead. Thanks again monks

Comment on Re^6: Memory addressing

Replies are listed 'Best First'.
Re^7: Memory addressing by QM (Parson) on Nov 28, 2005 at 01:44 UTC
We have legacy hardware that would cost $50k to $100k to replace. Um, gee, that was my point. How long do you think it will take you to write and debug a driver? How long does it need to run without rebooting or tweaking, etc? A week? A month? Let's say you're paid $40k/year, and you need 6 months to develop a "production" driver. You might think your time is worth $40k/year, but to your company, it's more like $80k/year. You work for 6 months, and it costs them $40k. What's the cost of ownership here? After the driver is developed, how much time will be spent keeping it updated with the latest requirements? 2 month/year? That's $10k/year. How about the hardware? What's its upkeep? What is that compared to new? What could you sell this hardware for? In the end, it may not make business sense to play with this, unless there is no cost in taking your (spare) time, and you just really want to write device drivers. Unfortunately, hardware and software is not backwards and forwards compatible forever, or there would be a lot of museums with useful capacity. -QM -- Quantum Mechanics: The dreams stuff is made of	[reply]

Replies are listed 'Best First'.

Re^7: Memory addressing
by QM (Parson) on Nov 28, 2005 at 01:44 UTC

We have legacy hardware that would cost $50k to $100k to replace.

How long do you think it will take you to write and debug a driver? How long does it need to run without rebooting or tweaking, etc? A week? A month?

Let's say you're paid $40k/year, and you need 6 months to develop a "production" driver. You might think your time is worth $40k/year, but to your company, it's more like $80k/year. You work for 6 months, and it costs them $40k.

What's the cost of ownership here? After the driver is developed, how much time will be spent keeping it updated with the latest requirements? 2 month/year? That's $10k/year. How about the hardware? What's its upkeep? What is that compared to new?

What could you sell this hardware for?

In the end, it may not make business sense to play with this, unless there is no cost in taking your (spare) time, and you just really want to write device drivers.

Unfortunately, hardware and software is not backwards and forwards compatible forever, or there would be a lot of museums with useful capacity.

-QM
--
Quantum Mechanics: The dreams stuff is made of

[reply]