I've had trouble thinking of Docker as anything but a, fairly thin, wrapper to LXC. When Docker announced a move to support BSD, that made sense, because of the existence of jails. LXC and jails obviously have a fair bit of over-lapping functionality, so supporting the BSDs isn't a huge leap...but Windows? Microsoft must have a team well into work on containerization, otherwise this new partnership would have to implement everything from scratch. And given the mayhem that cgroups brought to Linux, I can't see that happening quickly.

I generally agree except for the install-time bit. My last job was as the technical director for a school division. We ran mostly Windows on user machines and mostly Linux on the back-end. I did a lot of installs of both Windows (XP & 7) and Linux (Ubuntu Server) over those years.

Getting XP or 7 to the point where one could image it onto a bunch of other machines took us at least a day. We didn't leverage AD as much as we could. If that were the case, it likely would have taken less time, but I'm not sure how much. Installing updates was, by far, the most time-consuming part. We tried to schedule that for the end of the day so the bulk of it would be done by morning. Next up was setting up the software suite, and then the virus scan and disk compaction. Finally, the little details about the desktop environment needed to be dialed in before imaging.

Setting up Ubuntu Server on a box took under an hour. Our procedure for setting up a file-server (which included several other services) could be executed in at most three hours. Heck, on Friday, I upgraded an Ubuntu desktop box (with a RAID array that needed to be preserved) from 12.04 to 14.04 in a bit over an hour. The hard part was backing up the most important bits of data. The actual upgrade was painless...except for the corrupt USB-key I made the first attempt with.

We didn't really image Linux machines because they were generally back-end servers, but it was something I looked into with some depth. Basically, because of the effort Canonical put into supporting Ubuntu on VMs, it was a piece of cake. I was actually more interested in network booting ala the Linux Terminal Server Project, but the truth is that I would have been drawn-and-quartered if I put any distribution of Linux onto a machine when a teacher didn't demand it. The teachers ran the show. A situation that was mostly fine, but horrible in some areas (ie. security).

The current craze of 3D printers use additive methods (FDM). This makes it impossible to print a whole class of pretty simple shapes. For example a scan of myself with my arms held up 60-degrees above horizontal: printable (assuming there's not much overhang in my belly area, and ignoring my chin). A scan of myself with my arms held down 60-degrees below horizontal: unprintable without support. A sphere, for example isn't printable on a standard Reprap-type printer without support. You can add support or print models in pieces, perhaps with indexing to make them easy to assemble, but these can be challenging operations.

There are a lot of problems that still remain that make printing as a hobby a fairly frustrating experience. Automated bed-leveling will be nice when it becomes standard. Some sort of dissolvable material for printing support (using dual-print heads) would also be a huge leap. The software has a ways to go too.

But if your aim was to keep a couple of printers busy day-in and day-out in a business, it would likely be worth learning how to efficiently jump these hurdles.

Exactly. There are lots of decent router distributions (and not all are Linux). I haven't tried all of them, but so far my favourite is pfSense. But for the sort of hardware in wireless routers (the subject of this article) OpenWRT is more appropriate. Not having to worry as much about spying is nice.

Just be prepared for a significant wait between the release of 802.11ax hardware and OpenWRT support. Support for 802.11n took quite some time.

I got something similar yesterday: Picked up the phone and get a long pause before an employee of "Visa" greets me from an obviously packed and busy call-center. At this point, my Spidey senses were tingling. I would call his accent East Indian, which is fine. He tells me that I had qualified for a lowered interest rate. "Great!", I says. He then confirms my name and address, which didn't bother me as they are all over the place. But I hang up as soon as he asks for my date-of-birth.

They must get a bunch of people with this scam.

It occurs to me that several discussion forums I'm on know my date of birth. That now seems like an unacceptable risk.

More likely the software running on the Cisco boxes. Reflashing the software would be a simple task and wouldn't even involve a screw-driver.

You might be right. $3500 for a router was too much for us. That wouldn't get us in the enterprise door.

You are correct. There was no smoking gun. At least, not in any of the stories that I've seen. I mis-wrote. Apologies to Huawei.

Putting open source routing software on a rack-mount PC equipped with a few NICs is looking better all the time. Since the open source routing software solutions are getting quite good, this is doable. I did it and wouldn't go back:

About three years ago I noticed that our Cisco routers were a bottle-neck, worryingly old, and I was the only member of my staff comfortable with their CLI. We definitely did not have the budget to buy new Cisco routers, so I looked into HP and D-Link layer-3 switches. They were still too expensive. We used OpenWRT on some wireless routers, so the idea of using open source routing software was not new to us. Tested using plain Linux as a router. That worked, but was (way) over my staff's head. Tried Vyatta on the same hardware. At that time Vyatta's web-interface was a joke, making it no better than plain Linux for our purposes. (The web-interface may have improved since then and as a virtual router in a VM environment, Vyatta looks quite good.) Untangle was decent, but all of the interesting features had to be bought, which nullifies most of the advantages of it being open source. Heard about pfSense on the Linux Action Show and gave it a try.

Testing pfSense and learning its feature-set convinced us that it could do everything we needed (NAT, routing/firewalling between VLANs and the outside world) as well as do some other nice tricks (VPN concentrator, web caching/filtering, nice graphs of important stats, logging web usage, acting as a DHCP and DNS server, etc.). Basically, pfSense does everything that OpenWRT does and more since it expects to be run on more powerful standard hardware. Since it runs on standard hardware, the community isn't as fragmented as with OpenWRT, and more of pfSense's users are applying it in a professional environment, so the community support is quite good. The paid support is excellent. Being able to replace a failing router or NIC with something we had on the shelf is nice too.

So we had an open source routing solution that fit our needs, and much better than Cisco's offerings. But shifting all of our routing from Cisco to pfSense was a bold move. The Huawei story was the clincher for us. If Huawei did it, Cisco could too. That realization lead to my decision to always use an open source solution on network edge devices. This story seems to support that decision.

Which historic numbers are those?

I'm waiting on the edge of my seat to see a comprehensive study comparing the gamut between little and much government, but my expectation is for freshly picked cherries.

You realize that the Medieval Warm Period wasn't a global event, right? It was limited to the North Atlantic. Other regions of the world experienced both cool and warm periods over the same period. Read more on the topic here.

Having worked in more established institutions, mostly public ones, I'm envious of folk at start-ups. It seems like there could actually be a team, where everyone is mostly focused on the same goal. In established organizations this often isn't true and work becomes more about protecting one's position than getting a job done.

In an established organization, even if you are competent in your position, the management structure can mean you have little control over your destiny. Everyone between yourself and the tippy-top of the org-chart can decide your fate in a heart-beat. If any of those folk aren't worthy of the position (which happens a *lot* in public institutions) you have to have an exit strategy ready. Even if everyone above you is stellar now, tomorrow is a different day. I've had that sinking feeling, the one where I realize my direct superior doesn't have a clue and is making terrible decisions that directly, and devastatingly, impact myself and my staff. It has happened three times in the last five years and escalated in severity as each was of my superiors was replaced with someone worse. Serving at the pleasure of someone who is incompetent, and trying their damndest not to be found out, is tenuous.

So make sure your personal life is in order. Keep debt under control. Have at least six-months of living expenses in liquid savings. Live in a center where there are several other options for your skill-set. A partner with a stable income is wonderful insurance. Take your health seriously. Make sure that you are prepared for an unavoidable downward change in your health. That will happen eventually. Cultivate a network of friends and relatives that would stick their necks out for you in a time of need, by sticking your neck out for them when you can.

If one is personally prepared for things to go sour, the stress of personal loss won't add to the professional stress. Knowing that you can safely give your notice (no sense burning any bridges) goes a long long way.

And get paid to do it?

Unless you are doing desktop support (which doesn't pay) or developing applications that will run directly on the devices that sit in front of users (which can pay very well or not at all), you will likely be doing backroom work, where Linux dominates. Backroom work pays nicely and there's lots of it. True, one can do backroom work from a Mac easily enough, but a Linux desktop has its productivity advantages, like being able to spin up dozens of LXC containers in a way that mirrors the production environment, which is exactly the reason why my work desktop is a Linux box.

If your work is aimed at desktop support of Macs or developing applications for Apple products, a Mac makes sense, as does a Windows box for Windows work, but the amount of real work opportunities I'm seeing with Linux has grown a great deal in the last five years, like quadrupled. We're not talking about a tiny slice of the pie anymore. It's significant and growing fast.

Regarding the Homeland Security code audit of Linux and the 14x fewer bugs figure: Can we get a reference? I've done some Googling, but came up empty. That would be a nice feather in one's cap when some idiot manager starts ragging on Linux's security. (Which is happening to me right now.)

Without saying too much, I have some first-hand experience with traditional and some not-so-traditional composting toilets. That experience leads me to believe that a manufactured composting toilet, a user-built Jenkins-style composting toilet, and a multi-chambered composting toilet, can all be easily scaled up. The Jenkins-style system, in particular, is dead-easy to scale.

The challenge is that the western world has grown used to the flush-toilet. All that extra moisture presents a problem to any composting system. A small constructed wetland (basically a lined pit or trench filled with pea-gravel and topped with a moisture prefering mixture of plants) could handle the extra moisture easily enough. A small constructed wetland a few feet in diameter and depth can process an amazing amount of waste liquid. The system scales well too.

So scaling is not an issue. As the system scales, I also feel that management becomes less of an issue as well (because it becomes feasible to hire expertise in system design and maintenance).

I'm leary of biochar. Frankly, its following seems a bit cultish. I understand the carbon sequestering angle, but the issue I can't get around is that the process of generating biochar volatilizes a lot of useful nutrients, almost everything actually. The result is, in many ways, similar to peat moss. Lots of carbon, not much else. I've dumped loads of peat moss into low-quality garden soil, and it helps, but not like compost does, not even close.