In terms of repeating past mistakes, this has it all: mawkish and clumsy Kirk/Spock love; references to great episodes best left untouched by lesser writers; the rehashing of cool but fragile old plot devices that can't possibly hold up to the increased exposure*; and Romulans with overwrought names.
Warning: Spoilers, especially of the lunch variety.
IESB.net: JJ Abram's Star Trek Story Details
*This is typically done to "make the fans happy." See also Boba Fett.
Insanely cool story involving the New York Times, Hadoop, EC2, and lots of data. Lots of data.
I wonder how long copying 4TB of data from New York to Seattle took? They'd have saturated an OC-3 for two and a half days with that much. Maybe somebody just fedexed a bunch of drives to Seattle and they were loaded over a Gig-E LAN connection? At 6 hours of flying and 10 hours of file transfer, that's about four times faster.
[data transfer times via t1shopper.com]
Brand New weighs in on the new NYC Taxi Logo controversy, and finally tells us where to bring our pitchforks and torches.
I've been a loyal Red Hat user since 1997, when I installed Red Hat 4.2 on the Compaq Deskpro 2000 in my cube at work -- opting to use an old 3com 3c509 NIC instead of recompiling my kernel to work with the Compaq's built-in 10BT one. Red Hat has always been my distribution of choice in my web servers, and when Red Hat ended support for Red Hat 9 and introduced Red Hat Enterprise Linux, I was glad to pay for priority access to ISO downloads and Red Hat's then-peerless Red Hat Network update program.
But in recent years, the annual $349-per-server bill has become harder to justify. Felix made mention today of Red Hat's "nutty" pricing, so I thought I'd do a quick bit of research. Turns out Red Hat's actually in the middle of the pack if you don't count the free solutions.
Here's a chart of how much it costs to buy, and get software updates for some popular web server OSes.
| Product | 1 yr | 3 yr |
| Windows Server 2003 (Web Edition) | $399 | $399 |
| Red Hat Enterprise Linux ES 5 | $349 | $1,047 |
| SuSE Linux Enterprise Server | $349 | $873 |
| Solaris 10 | $240 | $648 |
| CentOS 5 | $0 | $0 |
I included Windows Server's "Web Edition" which doesn't let you run SQL Server. As none of the Linuces really include a MySQL license, I figured that was more-or-less Apples to Apples*. Before anybody lectures me about "libre" vs "gratis" or "free as in freedom" vs "free as in beer" let me just say: I know. I'm just trying to figure out how much money stuff costs. And for better or for worse, Microsoft's built a whole thing around their software updates, and they're pretty much free as long as the product is reasonably new, which Windows Server 2003 will surely be for the next three years.
So running a Red Hat server for a year is less expensive than running Windows for a year, but more expensive than running Windows for three years, and more expensive than running Solaris or any of a number of free/libre OSes no matter for how long. I wasn't able to find 3-year pricing for RHEL, which must be available, but Red Hat made it exceedingly hard to find, even in their clients-only extranet. If it does turn out that Red Hat doesn't discount for three year deals, then yeah, that's a little nutty, but presuming they do and their three-year price is competitive with SuSE's, then they're on the high side of normal. As for ROI or TCO or any of the more abstract stuff that includes staff costs and downtime and stuff like that, you're on your own.
CentOS 5 is a "clone" of Red Hat Enterprise 5, meaning drivers and RPMs that work in RHEL 5 will work in CentOS 5. But it's completely free (both kinds). And now that Yum, one of the free RPM package managers, is at least as good as if not better than up2date, and Bittorrent makes ISO downloads as fast as your copper can handle, I'm not sure why shelling out $349 a year makes sense from here on out.
* Speaking of Apple, which I wouldn't actually call a popular web server OS, a 10-user license (good enough for web hosting) is $499 with free updates.
Right now, the website where I work is hosted on two first-generation HP DL585's. They're big, loud, power-hungry comic superheroes, and I love them.
Having come from hosting on Dell hardware, HP is an absolute nerd dream. Unlike Dell servers, which tend to be made of customized commodity components, the internals of HP servers are full of thought and cleverness and custom engineering. Everything slides perfectly into place with little handles and arrows and lights and everything you need to feel confident when you're mucking around the insides of a pricey server. And HP's lights-out management tool (iLO) really does make visiting my servers unnecessary. With my Dells, I spent a few hours a week at my servers' colocation site. With the HPs, I physically touch these servers maybe twice a year. Even if the server's CPU is spiked and completely unresponsive, I can hop on to iLO and reboot the thing or get its network connection back up, etc. The web-based management interface even works on my Mac (even when Java is involved). Trust me, that's saying a lot.
You should know that I love researching, specking out, configuring, and installing new servers. To me, servers are the the sports cars of the nerd world. So sadly for me, I did too good a job with these DL585's. While the servers are three years old, they're still perfectly capable of handling the task they're given. Back in 2004, I spent a lot of time calculating how we'd engineer the system: how big and fast the disk and CPU would need to be, the best RAID to use, etc., and then we spent a lot of money to get a system that could live a long time. The good news is, all that planning paid off. The bad news is, all that planning paid off; I have no reason to buy new web servers. And I tried that whole asset-depreciation trick -- our Controller wanted no part of it.
There's great new server gear that has come out in the intervening years, but I'm pretty much watching it from the sidelines. Perhaps most interesting is the rise of blade servers. They were around in 2004 when I was putting our system together, but they were a bit bleeding edge and out of our range both technically and financially. Not anymore. The HP c3000 "Shorty" enclosure looks like exactly what I would use if I were putting together a web farm today. Easy, lego-like racking. No need to bring along skinnier staff people who are agile enough to get at the back of our servers. There are tons of blade choices, with easy and obvious integration. No cabling hassles.
There are blades for everything. Want Opteron? Yes. Intel? Of course. direct-attached storage? No problem. SAN? Yep. Even backup tape drives. The management is all centralized, and it has a cool LCD panel in the front that lets you bootstrap the enclosure and do health monitoring (it's designed for branch offices, so you can see a virtual version of the panel in the web interface).
When I've researched blade systems in the past, there were all sorts of strange questions leading to seemingly consequential decisions -- about the backplane, etc. Not so with the c3000. It doesn't seem to have the kind of pitfalls or hard choices a small shop like ours can't afford to get wrong. I'm sure it trades in some flexibility, but with the benefit of making blades a real option for us.
Of course, this is all conjecture. Unless traffic to my company's website suddenly quadruples, I won't have my hands on a c3000 for a few years. Maybe an enterprising HP product manager will read this and take pity on me and offer to trade me two happily chugging DL585 G1's for a stocked c3000. Hell, I'd even blog about it.
Mark Zuckerberg seems like a smart enough guy. Hell, he went to Exeter and Harvard (and like Bill Gates, Zuckerberg got out while the getting was good). But I think he's a few term papers short of a history degree when he said at the big Facebook announcement yesterday that "once every hundred years media changes."
Even if you're just talking about some higher level of abstraction than one normally does when referring to "media changes" (Zuckerberg: "The last hundred years have been defined by the mass media"), this is just wrong. Media, like the rest of human progress, is marked mostly by slow centuries of progress, followed by a 150-ish-year whirlwind of industrial innovation (call it an "industrial revolution" if you like).
Maybe he's reading his own press a bit too literally. Is he saying that the new Facebook advertising platform represents the biggest change in "media" since radio? Bigger even than the Internet itself? Man. Maybe Facebook really is worth $15bn.
And that's just the first two sentences of Zuckerberg's speech. Nick Carr makes short work of the rest:
Marketing is conversational, says Zuckerberg, and advertising is social. There is no intimacy that is not a branding opportunity, no friendship that can't be monetized, no kiss that doesn't carry an exchange of value. The cluetrain has reached its last stop, its terminus, the end of the line. From the Facebook press release: "Facebook’s ad system serves Social Ads that combine social actions from your friends – such as a purchase of a product or review of a restaurant – with an advertiser’s message." The social graph, it turns out, is a platform for social graft.
Ahem. That all being said, we're definitely going to explore using the new Facebook ad tools to find new subscribers.
The media might have treated the iPhone price drop and rebate as an oddity, but it wasn't the first time Apple gave a rebate to upset customers who paid extra to be early adopters.
Apple introduced the Apple III in late 1980, at $4,340 (almost $12,000 in 2006 dollars). Although it was released with Jobsian flair (Apple rented Disneyland for the day), it was underpowered (it could display fewer colors than the Apple II, which it nominally replaced), and defective (some chips would come unseated and Apple's workaround was to "drop the unit 2-3 in. onto a desk to reseat them"). Reviews were bad, and sales were terrible.
To try to save the product, Apple dropped the price in February 1981, to $4,190 (still, about $9,900 in today's money). Customers who had just shelled out for their Apple IIIs were livid. So Apple gave a $50 rebate to every customer who bought an Apple III before the price reduction.
The price break didn't save the product. Only 7,200 units were sold, of which about 2,000 were defective and needed to be replaced. Though the IIe, IIc and IIgs were still years away, Apple started looking in a completely new direction for its next generation of products.
The Apple III also coincided with the beginning of the end of Apple dominance in the microcomputer market. IBM introduced the PC in August if 1981 for $1,565 -- a still spendy $3,700 in today's money.
The monitor we had hooked up to my Apple IIe growing up was an Apple III monitor (see picture above). It was monochrome (and the color was green) and the glass had a kind of corduroy texture to it presumably to reduce glare. Even for its day, it was a terrible monitor. I can still remember how my Apple IIe had an 80-column card, and how my heart skipped a beat when I switched from 40 characters wide to 80.
[via Silicon User]
I hope Rudy Giuliani's happy now.
Omniture is buying Visual Sciences for $394 million.
Visual Site is the analytics package that the cool kids use. I guess the market consolidation is to be expected; where there used to be four "big" analytics packages (Omniture, WebTrends, WebSideStory, and Visual Sciences), there's only two now. WSS and Visual Sciences merged a while back.
Ultimately, I think this is good news for us Omniture customers, as there is some neat functionality Visual Sciences provides that presumably Omniture can now provide. And with Google Analytics still free, all two of the for-pay systems have huge incentives to keep their prices reasonable.
Your source for semi-interesting nerdosity, half-truths, and unwanted opinion. For everybody who enjoys smiling politely and pretending to agree.
subscribe