Friday, October 19, 2012

Can Canonical Make Skunkwork Work?

This post is in regards to Mark's recent announcement that Ubuntu 13.04 will be using a skunkworks approach to some of its more wow-factor features.

I know some people are going to cry out from their basement, screaming "community, community, community!" Let's face it though, one thing Linux is missing is being able to release something and people say "holy shitballs, batman!"

Anyway, I totally applaud Mark and Canonical for this decision. If it all goes well, here's my vote for naming 13.10 the Snazzy Skunk.

Tuesday, October 2, 2012

The GOP/DEM Circlejerk

This is slightly off from my normal blog topics, but it is election season. As I stare off with an uncontrolled dazed look on my face, pondering which candidate to vote for, a deep depression has set in.

I could choose Obama. At least I'll know what I'm getting, even though I disagree with his wealth distribution policies and bailing out too-big-to-fail companies that are sending us in a spiral of debt.

I could choose Romney, since I'm Republican, but I honestly don't like him as a person, and I feel he will serve money hungry people more than a strict capitalist agenda.

I could vote Libertarian, but, let's be honest, that will just take votes away from one of the other candidates. At this point, a vote for anything other than Romney is like voting for Obama.

My depression begins when I consider that the two primary political parties are only interested in preserving themselves. Considering that Democratic and Republican platforms have morphed every election campaign into what they think will get them voted into power, or attempt to separate themselves from one another, I can never be sure that the party I choose on my beliefs really fits who I'm voting for.

Every election it's the same thing: "Your problems are because of <insert opposite party>." Neither party will say "our platform allowed these problems to happen" or "the bills we voted caused a downturn in the economy, we'll revert that and try something else."

Let's consider this. Say you work at a company. You and another co-worker handle a particular product or service together. You work well together, bouncing ideas and patting each other on the back. One day, your manager quits, and since your company promotes from with-in, you know that you or your co-worker are going to replace him. You'll get more money, access to the executive bathroom, more vacation days, etc.

So during the manager's last couple weeks, you work your ass off. You make sure that the higher ups responsible for the decision see your hard work. You are basically campaigning for this new position. Your co-worker, being prudent, does the same thing.

Now, instead of working closely together, you are both trying to make yourself look more important and more manager-like. During these few weeks, arguments ensue, and instead of trying to build a great product or service, you are positioning, spreading distrust against your co-worker and generally being unproductive, all for the power of that promotion.

In the end, your co-worker wins out. You blame it on him bringing the top-shelf bottle of liquor to the Christmas party and taking the boss out to golf with some no-name NBA player that his cousin is towel-boy for. It's definitely not because he's better.

Now you stew in your hatred toward him as he becomes YOUR boss. At first, you continue to do your job, until one day he tells you that he thinks the product would be better if you changed one aspect. This doesn't sit well with you. "Who does he think he is?" It wasn't your idea, so it must be crap. You toss all logic aside and just hate his suggestion, regardless of the fact that it will save tons of money and is generally better at solving a certain problem than your solution would have been.

You do as you're told, but you don't have it in you to put in 100% effort. You don't necessarily sabotage the product, but the solution he asked you to implement is definitely half-assed and you don't care.

Now the product goes into production, and sure enough, the implementation falls flat, the one that you worked on. However, it was his idea, so you don't care. When the firestorm comes down from above, you blame the problem on your back-stabbing co-worker that took yer job. "He wanted it implemented this way, but I wanted to do it different." Sure enough, he's canned and now you have the manager job.

You got what you wanted. The power courses through your veins like an espresso hopped up on Redbull. You are the man. Now, two people work for you, one of whom was there when you did the poor implementation, and overheard you complain about your manager at the water cooler. He knows what you did, and has a distaste for how you got your new job. He thinks you are incompetent and undeserving: He is just like you used to be.

You get my meaning here? How can we expect Congress or the President to do anything useful if their entire intent is to make their affiliate party look good, not to mention, self-serve their lust for power? How is the President supposed to get anything done when his counterparts in Congress are trying to set him up for failure so they have bullet points on the next presidential election?

I wish I had some sort of proposal to fix this, but alas, I'm not a political scientist, so all I can do is bring the bad news.

Friday, September 28, 2012

Power Architecture Related Tracks Proposed for UDS-r

On my ramp-up toward UDS-r, I've created some blueprints and pinged some related folks to get them into the proper tracks.

I'm hoping to get a lot of interest and discussion around there, so here they are:


So this covers a wide range of topics. The most in-depth one is the Virtualization blueprint. As of yet, I've not seen a lot of broad support for non-x86 in OpenStack and related software. While it works (I've set it up), it just doesn't do a lot to make me happy happy.

The boot loaders blueprint is basically an RFC. The idea of Power architecture on a non-embedded system not having OpenFirmware is about on par with Dell selling an Intel system without a BIOS. The Power systems do have u-Boot (Das Boot), but that's not as robust as it needs to be. I'm thinking something like grub2 being compiled agains the u-Boot API that u-Boot can load modularly or perhaps something like the kexec based loader that the ps3 used.

Finally, the kernel development is a host discussion that needs to be hammered out with the Canonical Kernel Team so we can all be happy and not step on the primary architectures, but still being able to spread some support for newer Power equipment.

Cheers and see you in Denmark!

EDIT: Updated link for boot loaders blueprint

Wednesday, September 26, 2012

The Rack Revolution

As I sit here in my cozy home on my comfy couch, I am bewildered and amazed at just how far things have come in the last decade.

Let's take a quick inventory of my immediate surroundings:

  • Laptop
  • WiFi
  • Smart Phone
  • HD TV
  • High Speed Internet
  • Server Farm

Hmm...that last one's a bit different from the old days. I used to have a nice collection of loudly humming, room-warming servers in my garage. As a telecommuter, I needed it. My blog was running on it, my email was running on it and my firewall was running on it.

What happened? Well, we all know the answer to that question: things consolidated into the "Cloud." Instead of under-the-table boxes running our local services, we now have providers doing the heavy (literally) lifting for us.

So what do they run on? Practically the same loudly humming room-warmers that we used to keep under our desks. However, in recent years, the move is being made to lower the operating costs of these rack farms into quiet, low-powered, self-cooling, maintainable animals.

As most places have tried to just tone down, or spread thin, some have been making the move to efficiency. Enter the reverse revolution of the CPU to something more applicable to today's computing needs. Instead of powering with high-wattage x86 chips, many are dipping their toes into the shallow end of the alternative-processor kiddy-pool.

And with that I introduce an amazing NEW and WILD CPU: PowerPC!

Oh, you've heard of it? It's legacy and old-hat, you say? I must be thinking of a different PowerPC CPU then. The company I've been gainfully employed with for the past 1.5 years seems to be using something quite different than your grandmother's Power chip. Not quite the behemoth of the IBM Power7 iron (in size nor noise), but not the wussy of your old PowerMac neither.

We're talking multiway SoCs with full floating-point running at a fraction of the wattage of just about anything else on the market. Add with it full hardware virtualization (via KVM), and you begin to see where in the market this is headed.

We've already been engaging multiple Linux and software vendors to give a complete and first rate experience on this new class of hardware. You'll have multiple choices when it comes to supporting and administrating, whether it's one system or a room full of racks filled.

So here's my not-so-humble way of introducing you to Servergy, Inc.. They've been around for 3 years, but expect to be hearing a lot more about us in the coming months. If you're going to be at a Linux or Cloud/Server related event in the near future, chances are you will run into one of us. I'll actually be at Ubuntu's UDS-r in Copenhagen at the end of October. I'm hoping to have a live demonstration while I'm there.

Cheers

NOTE: In this article I am speaking solely on my behalf. None of what I've said can be taken as a statement by the company I work for: Servergy, Inc.

Tuesday, July 17, 2012

The Community Conundrum: PowerPC

In my recent work, I've been dealing a lot with PowerPC. As an old Mac user, I've had a soft spot for PowerPC for ages. Like most people, until recently, I've considered PowerPC an aging and dying architecture. Even with IBM selling PowerLinux systems, the lack of cheap hardware for developers has left a hole not easily filled, no matter how man old PowerMacs you buy in eBay.

However, there are a lot of PowerPC platforms that do fill this gap left by PowerMac. Some are even 32-bit platforms that can compete in today's markets.

So why have you never heard of them? Why can't you download Fedora or Ubuntu to install on your PowerPC of today? Several reason:

  • Distributions don't really support it.
  • The "community" behind it is driven at the kernel and low-level, not at the distribution level (see last bullet item).

This circle of support appears to be the hold up. Convincing even community supported architectures like Ubuntu and Fedora to support these different kernel flavors is met with archaic skepticism, and is usually concluded with "there is no community" to which I usually respond "because there is no support."

Something has to give here. Linux and Open Source isn't where we want the chicken-and-egg scenario to happen. You can't walk up to a Linux distro with a community and say "Here we are, let's do this" in much the same way as you can't go to a community and say "Come over here with us. We don't support you yet, but we'd like you to prove that you're worth it."

So where to begin...