Hacker News Reader

Comment by keyle | Hacker News Reader

     I now understand why I'm not – and will never be – rich. My first (and last) employer complained about my preference for stable and reliable solutions, equating it with lesser profits. According to him, unstable solutions requiring frequent maintenance meant more revenue. To me, a job is well-done when it works consistently, not when it demands constant fixes.

This is a great point. Haven't you noticed the teams that have downtimes constantly are hailed like heros for saving the day countless times. Yet the guys that have services running 24/7 for years without a hiccup will be the first ones on the chopping block when finances get tight.

I am also in that boat. I too like systems that never fail, defensive programming, proactive attention. I'll never be rich and most of my company has no idea what I do. It's still the right thing to do.

Boring that never fails = Quality.

View on HN

I've been a project manager for nearly 20 years after "hanging up my technical boots" and trust me it's the same shit for us as well - quietly get on with it, manage the challenges and the team, anticipate and deal with the problems upfront? You are an unremarkable "safe pair of hands".

Run around panicking with your arse on fire constantly firefighting issue after issue (almost all of which you should have anticipated and already dealt with) making your constantly "red" project go "green"? You're a fucking hero that wins any internal little award/prize for rescuing the project.

Still annoys the crap out of me after 20 years.

Now apply that principle to politics: representatives who actually work to, uh, represent their districts, tend to quietly go about their business and are rarely promoted. To get promoted you need headlines, which you won't get if you're actually good at your job.

How does a senator get promoted?

Power broker, party positions of influence, special committee positions, 'speaker of the house' or become president.

Depends on the buyer. Many avoid GCP because of a perceived lack of policy / customer service reliability.

In defense of your job, make sure you are constantly communicating how much uptime / how stable / reliable your services are. One reason why the people with lots of downtime are hailed as "heroes" is because they get more face-to-face time with upper management.

Of course, some managers will be dumb and equate instability with revenue.

Think bigger. Start a quality IT consulting business and possibly be rich. There’s a small fraction of industry who will pay for it. Likely plenty for many ‘lifestyle businesses’ per region.

There used to be good money in that, setting up and securing email/filesharing/website for small businesses.

These days they largely opt for the GSuite/Squarespace stack or similar to fulfil those needs around here.

I have been wondering about this recently.

Actually I have been considering investing & going BSD only to escape the insanity that has now also contaminated the "Linux world"

Now from my initial observations the opportunities in the saner world are quite scarce and hard to find. But I am pretty sure those companies often face a skill shortage too.

what insanity are you referring to? systemd?

> also

Probably complaining about the rat race / grind mindset which leads to lot of people without strong fundamentals gaming the system and getting high end jobs.

Systemd doesn't bother me that much actually.

The whole Docker & container craze vs how things are done in FreeBSD worry me much more.

But this is not so much about the technology but more about the fact that as Linux became mainstream & the default choice it doesn't filter out anymore the companies who have no idea what they are doing.

I agree about the Docker & container craze, but doesn't FreeBSD have something a bit similar - jails - since 2003 if not earlier?

Wow, mid-2000 release date, I thought Solaris containers were older, but definitely not!

Yes but jails had since the beginning much clearer and stronger security features.

"void linux" is a great distro if you want BSD-like but still linux. It also uses runit instead of systemd.

Isn't all this stuff in the cloud now, most startups I goto don't even have a single server in the office

Are you implying every system is stable and reliable now, because it is in the cloud? My guess is that it helps on the hardware front, but there is more to a stable system.

Yup, definitely resonates here. We design replacements and upgrades to keep ancient stuff in factories running. After the initial fixup, we often never hear from a customer again until a different machine breaks.

Not sure how to be better at making money.

> Yet the guys that have services running 24/7 for years without a hiccup will be the first ones on the chopping block when finances get tight.

Not really my experience, as long as the stable services costs of operation are good. I have witnessed more chopping on the R&D side, where things are never "stable and reliable", but innovative and expensive.

No need to change or outsource something that works technically and financially. Until it becomes irrelevant.

I don't disagree, but I think the opposing take would be the cliche "if you're not breaking some things then you're not moving fast enough" (by whatever measure of "fast" or progress management would like to use).

What angers me the most about this mindset is that unless your dev teams are full DevOps (this does not exist IMO), and truly can understand and create scalable and reliable infrastructure, all this means is “we are externalizing tech debt onto other teams.”

You built everything in AWS via ClickOps? Sounds like an SRE problem.

Your DB is just a giant JSOB object in Postgres? Sounds like a DBRE problem.

It’s extremely frustrating, as someone who has been / is both of those other groups, to see projects being hastily thrown together, knowing that one day it’ll be your problem to solve.

If you have a stable setup that does not need changes as you grow, chances are you're not growing fast enough, or at all.

New and thus inherently unstable things may allow order of magnitude jumps in efficiency, and, again, getting onto a growth trajectory.

Dropping VMS and Solaris in 1997 and switching to Linux was switching from stable software and hardware to unstable, but such that had a growth potential. Switching from bare metal and colocation to AWS in 2009, the same. Switching from highly advanced ICE to electric motors and huge lithium batteries in 2007, the same.

If all that you value is stability [ed: of your setup], the amount of wealth you own will also stay the same, at best. Stable things need to be stepping stones into the unstable, uncertain future that may bring something bigger.

From direct experience, instability and uncertainty less to your customers getting angry on twitter and fucking off rather quickly. All they want is to do what they did yesterday without being poked in the eye.

The whole move fast, break things and grow quickly is a cancer that compromises everything not a commercial win. When you put yourself before your customers you fucked up.

I'm not talking about breaking things and having downtime.

I'm saying that if you have the same infrastructure over 13 years, and you did not have to upgrade it, or replace its parts, or even migrate off it, you're likely not growing.

All the above things can be done with minimal disruption for the customers, ideally in a completely transparent way.

Perhaps you don’t need to grow? Growth is not the only business model. I mean failing to do maintenance is a sin of course but anything else, there is no one true course.

I originally answered to a comment about never becoming rich. No growth means not getting rich, this is what I'm trying to point at.

Of course growth and getting rich is not the only worthy target. Say, OSS is not about getting rich, but it's wonderful and worthy without doubt, in my eyes.

At a practical level, this is NetBSD story is a story of a machine that just does it's job and doesn't change. For example, there is no active experimentation, no machine learning added on, etc. It's like a wood-mill; good that it make 2x4s but also not a highly dynamic business. Probably not a business that's seen a lot of competition or having to adapt to changing markets and products.

I know of at least one semiconductor company pushing the state of the art on 1990s litho processes, running 1990s control software on their stepper that runs on Solaris on SPARC32 (SPARcstation 5). Anecdotal, but their production engineers have stated that not relearning how to use critical parts of their process every few years has helped them stay focused on actual R&D.

This is of course an extreme, probably unusual example.

Reply on news.ycombinator.com