Many years ago, I did consulting at $125/hour, but if I was going on-site (particular in another city), I'd bill by the day, at $1000/day (often putting in 10 hours a day, so $100/hour) with some sort of minimum in place, so I wouldn't drive or fly somewhere to only be needed for one day. The reason I could charge what I did was because I did some stuff that led to me being well-known. I wrote a book; it didn't sell very well, or pay me much, but when you say, "I wrote a book about system administration, published by No Starch Press" it raises you to the tier of people that can charge $125/hour without hesitation.

Also, when you have a long-running consultancy that specializes in one niche (mine was web caching, website acceleration, etc., though I also did good business in developer infrastructure, like building systems for packaging and testing and revision control, etc.), you'll have fewer customers, but they'll be customers who will value your skills. A jack of all trades, who is actually really fucking good at all trades, is incredibly valuable, but it's difficult to market that because it's difficult to prove you're really good at everything. Proving you're good in one field is just a matter of following the mailing lists and forums in your field, and being talkative and helpful, writing a blog about your subject, and eventually turning all that writing into a book or magazine articles or something.

Of course, I never took it to the next level, as this article discusses. I was still billing by the hour, and still being thought of as a (pretty high end, and difficult to find) commodity. So, I made a decent living, but it didn't scale.

The keys are two simple things:

1. Be able to quickly fix expensive problems.

2. Make sure people with expensive problems know about you.

I've routinely seen multi-million dollar projects held up for months over things a skilled sysadmin can solve in a day. You just have to be at the receiving end of that 911 call, and not afraid to quote a big flat fee for problem resolution.

I disagree with your statement. I am currently in Chicago just completing the infrastructure implementation for a large enterprise client where I specifically was brought on to do their tech infrastructure.

They pay a lot for this service.

Here is my scope of work:

Physical cable plant design

Physical network design (based on their corporate logical architecture)

Physical/Logical implementation (Procure, track, receive, rack, patch, power, config, test, coordinate, hand-off)

Turn up and test

Day-one operational support and hand-off.

Doing this for all their global offices.

Also, why do you only do remote work? Is there no one near you who needs a sysadmin? Could you travel? People value face-to-face interaction a lot more than online; if you only do remote work, then some guy in Bangalore can undercut your rates (hence the $20 an hour offers), and you're not really offering much over him.

And as the original article says, don't charge per hour. Charge per week, or per incident. And charge rates that make sense for that kind of time frame. Read the original article; it has good advice. Once you charge per week (or per incident, based on a standard weekly rate), you can start talking about "OK, you want to spend less, which things should I not do", not "Well, I think I'm worth $50 an hour, but since I need the money, I'll take the job for $20"

Yes, for systems administration it can work that way. Here where I work we don't really have any sysadmins (or everyone's a sysadmin, which is even worse). We have developers, and we have tech support who support customer systems, and we all kind of chip in on the system administration, but we don't have anyone who's job it is to maintain our own network. It finally got so bad that we've brought someone in to straighten it out. I don't know the details of how he's being paid, but I could easily imagine it as "it will cost $5000 for me to track down your problems, reconfigure your core router, set it up in a way that will be more maintainable in the future, and write the documentation for running that."

No, please do reply. Then go in wearing leather pants, with white powder on your upper lip and reeking of Jack Daniels. Use a pawn-shop electric guitar to smash the fuck out of their interview room. Then stab somebody with a katana on your way out.

It's the only way these people will learn.

I worked at a company that ended up spending probably several man-years worth of effort over a period of three years on puppetizing their environment and still doesn't have a one-click-deploy of their infrastructure and feels the pain of that acutely every few months when a new environment needs to be brought up by hand.

If someone'd been around offering 30, 50 or even 100k no-cure-no-pay work to get their environment into a one-click-deployable state in under a month or so I'm pretty sure I could've sold that internally at several occasions.

Instead you're a site scaling engineer, or a cloud automation specialist, whatever sounds like it'll solve the client's problems - which these days are frequently that a bunch of developers set up the production environment on the fly and it just about works but isn't reliable or convenient.

I think the future for sysadmin work is definitely 'DevOps' - and it's not really that much different from old school sysadmin. Yeah a pure Chef/Puppet-managed system is more like writing code than hacking away, but behind it all it's the same kind of work.

It is definitely possible to bill US$2k per week or more as a contract sysadmin, but trolling through Craig's List isn't the way to manage it!

Could you share some of the most important issues (technical or not) ? In advance many thanks.

I am really curious about your experience as we try to achieve the same objectives here with http://comodit.com

The situation works great for both of us.

I don't think Patrick's model is ideal for sysadmin work (who needs a sysadmin for an entire week?) ... but I think the retainer/on call model could be just as good ... get 10-15 companies paying $500/mo ...

It only worked a few times, but for those few times it was worth it.

Sorry to say it, but almost nobody hiring consultants gives a shit about cleaning up messes. If they did, they wouldn't need to hire consultants to clean up messes.

Don't clean up messes, or at least not the things that you perceive as messes but they don't. Just like the article says, solve problems that people with money know that they have. If somebody with a clusterfuck of a system is paying $30k/month for hosting and you can cut it in half with a month's work, then charge them $30k. They'll balk, but tell them they'll get paid back in 2 months. If they are too dumb to take that deal, fuck 'em.

I know it's painful to leave people with all the other messes they have. But if you're going to do charity work just because it makes you happy to tidy things, then do it for actual charities.

Or even specialize in when shit hits the fan? I mean the advice to startups are always 'find a customer with their hair on fire'.

http://youtu.be/ANPsHKpti48

The implication being that if a company only pays peanuts, they probably only employed monkeys before, and it'll be a head-wreck of a situtation.