Remove Ads From ‘Special Offers’ Kindle for $30

Sonim's rugged XP3300 Force does not blend

Our old friend Tom Dickinson is back with a brand new edition of Will It Blend?, but it seems his ultra-destructive Blendtec blender has met its match -- the equally rugged Sonim XP3300 Force we tested at MWC. Now, Sonim doesn't have a perfect track record, and Tom definitely let the smoke out here, but apparently the leftover wreckage was still functional. Better luck next time, blender!Sonim's rugged XP3300 Force does not blend originally appeared on Engadget on Wed, 01 Jun 2011 20:20:00 EDT. Please see our terms for use of feeds.Permalink CNET  |   | Email this | Comments

MORE BUSINESS INSIDER SECRETS REVEALED: An Inside Look At The Guts Of Our Tech Infrastructure

Inspired by Business Insider's Full Monty article, in which CEO Henry Blodget went into great detail about the financial workings of Business Insider as a company, I thought a similar revealing article about our technical infrastructure was in order. We've written infrequently before on the inner workings of the tech here, especially with regards to MongoDB, and the interest has always seemed strong, so let's go into some depth today about the BI server architecture and our development and monitoring tools, and then I'll take your questions. If you aren't of a technical background, be warned: this article is about to get nerdier than a Wikipedia editors' meetup. Put some MC Hawking on in the background and read on. Business Insider runs on a multi-tiered server architecture that enables us to scale to meet the unpredictable demands of the web media business. The entire site runs on only about a dozen production servers and is easily able to handle the 4 million+ page views we may find ourselves needing to push out on a high-traffic day. The heart of the site and its lowest tier remains MongoDB, of course. MongoDB gives us blazing fast performance and even when the site is heavily loaded the database servers barely edge over 10% CPU utilization. One optimization we've made at the database level since my predecessor Ian White's article on the topic is to move our analytics collections onto a separate server. This way, the relatively large volume of write requests to the analytics collections doesn't cause locking and takes their load off of our main database. Both of our Mongo databases are configured as replica sets with a slave server for failover purposes, but we haven't yet needed to do any distribution of read requests to the slave due to Mongo's raw speed. The next tier up from MongoDB are our web application servers. These run Apache with PHP 5.3, cached using APC, to build our pages and make the magic happen. The application servers each run an instance of memcached so we can cache commonly used page fragments and database query results in memory for speed. The webservers also contain lightweight MongoDB instances running as arbiters for our MongoDB replica sets. Arbiters vote to decide which database server should become the master in a failover situation, but they don't host data or receive queries. All they do is sit there and vote occasionally, reminiscent of a Congressman except a MongoDB arbiter consumes far fewer valuable resources. Hi-yooo! Above the web servers are our front-end caching servers. These are machines that run nginx and Varnish and function as reverse proxies to the web app servers. Varnish caches complete pages and some user specific page fragments so the web app servers don't have to generate a page from scratch for every individual visit to the site, which helps us handle the unpredictable high-traffic events that can be caused by the latest celebrity philanderer, market crash, or end of the world prediction. It also helps us easily scale pages that involve a lot of number crunching and database lookups, like the new data center pages we're currently rolling out. We use a fully customized Varnish configuration so we can do active cache purging and keep our pages up to date. For example, when an editor posts a new story we actively purge the verticals the story appears on, and when anyone posts a comment we purge that article page. This helps us achieve a high cache hit ratio while keeping our content as fresh as possible. A fast nginx web server instance sits in front of Varnish and handles on-the-fly gzipping of content as well as rate limiting. Static content such as images, CSS and JavaScript are not cached by Varnish; it's not necessary because they get delivered directly to clients via our content delivery network. For our internal site search, we use a Google Search Appliance that crawls the site and ingests our posts into a full text search index. We're outgrowing this, so we're looking into either upgrading or migrating to Solr or a similar open source document-based search engine. Yes, we could do full text searching with MongoDB, but I prefer using a tool more specifically suited to that task. A separate server runs our cron jobs that perform automated background tasks like generating short URLs for new posts via bit.ly and populating the Hive with links, among other miscellaneous tasks. Some other points of interest about our tools and how we write code:

We use git with github for distributed source control and revision tracking. We use Munin to measure our server performance and draw pretty graphs to help us diagnose problems. We would set up Nagios for monitoring, but our hosting provider does a good job sending us alerts when there's a problem so it hasn't been necessary. We use Hoptoad to alert us to application errors that may happen on the site. Any uncaught exception notifies the tech team and the Hoptoad web interface shows us a stack dump so we can start debugging the issue. All of our servers run Linux, specifically CentOS. The server hardware specs vary depending on the machine's role, our web servers have dual core CPUs and 8GB of RAM, while the database and caching servers are quad core with twice as much RAM. Most developers do their work on a MacBook Pro, but we do have a couple of the more hardcore hacker types running Linux desktops on their machines. Most of us use Textmate on the Mac to write our code, but there's a smattering of NetBeans, Eclipse, and straight up vi as well. We each develop locally using Linux virtual machines that contain our entire application stack on a single VM, then push code to a shared integration server for testing and QA. Backend developers share the sysadmin duties amongst ourselves. We don't have a dedicated systems administrator, since our servers are managed hosts it's not generally necessary, and when any important admin work needs to be done a backend developer will wear two hats for a while. We use Catchpoint to monitor page load speed, something we are focused on improving over the balance of the year.

One future improvement to the infrastructure being considered is using a Hudson server to perform continuous integration and automated running of our testing suite of low-level phpunit and high level Selenium tests to improve release quality. Also ongoing are more enhancements to the Varnish caching strategy to do more intelligent purging so we can raise our cache TTLs even higher and achieve even better scalability under high loads. Scalability is what keeps me up at night: traffic has doubled since I joined BI eight short months ago and it shows no sign of slowing. As a final note, since I know it's going to be asked about by our wise and gentle commentariat, any perceived slowness of the site isn't caused by our server infrastructure. Most of the slowness people experience is on the client browser side rendering our relatively complex HTML, retrieving images, and executing JavaScript analytics and ad requests unfortunately. Faster servers can only accomplish so much if the content being served is complicated with a large number of third-party client side elements, but such is life in the ad-revenue-based web journalism world of 2011. Someone has to pay the bills. There you have it, you've now been exposed to the official BI technical Full Monty. Hope you enjoyed it. Fire away with your questions and I'll do my best to answer them as long as you're not asking me to reveal our root password. (Okay, fine, it's "passw0rd". We use a zero instead of an O because we're clever like that.) For the latest news, visit Business Insider. Follow us on Twitter and Facebook.Join the conversation about this story »

Wanted: Business Insider Seeks Experienced PHP Coders

Business Insider is looking for two awesome experienced backend PHP developers to join the fast-paced and exciting world of being an engineer at a journalism startup. We're looking for some strong PHP developers who have tech chops, a startup mindset, and the ability to work hard and play hard. Business Insider is a great place to work. We're laid back, we don't let bureaucracy get in our way, and we play-ping pong in the office every day. It's honestly one of the best places I've ever worked, personally. Since the last time I put the word out looking for engineers back in September, our page view traffic has almost doubled and our unique visitors per month have almost tripled. It's an exciting time to be here, and we face unique challenges every day due to the demands of our huge audience and the 24/7 news cycle. The tech team also handles basic administration of our scalable server infrastructure, and we get exposure to cutting edge technology such as MongoDB, due to our close association with 10gen. Our product team pushes the boundary on feature development, resulting in a varied and interesting assortment of tech challenges to solve. Position requirements are as follows:

Skilled back-end programmers who can write lean object-oriented PHP 5 code Strong front-end JavaScript and CSS knowledge a plus Broad understanding of web architecture and how it applies to scalability Enjoys building user-focused interfaces, using Ajax when appropriate Experience with systems administration of Linux servers a plus Experience with NoSQL document-based DB stores such as MongoDB a major plus Objective-C iPhone/iPad development experience also a major plus Strong communication skills Based in New York -- you will need to be able to work out of our office on Park Avenue South. Some telecommuting is fine.

We offer a competitive salary and good benefits. Please email a cover letter, resume, and a short code sample to pax@businessinsider.com. Thanks in advance. Please follow Business Insider on Twitter and Facebook.Join the conversation about this story »See Also:LAUNCHING THE ENGAGE-O-METER: See How Many People Are Reading Business Insider Right Now!The Original "Scariest Jobs Chart Ever" Is Still Plenty ScaryBusiness Insider Is Hiring A Research Analyst

John Doerr And Reed Hastings Put $11 Million Into Education Startup

Netflix CEO Reed Hastings and super VC John Doerr just invested in DreamBox Learning, an adaptive math learning platform. The startup just raised an $11 million round from Hastings (through his Charter School Growth Fund), Doerr (through his private investment fund), not Kleiner Perkins, and others. The company plans to use the investment round to expand the product and curriculum as well as increase distribution. In the past year, half a million elementary school students used the platform, viewing more than 11 million lessons. An independent study found that students who used DreamBox for four months, improved their test scores by about 5%. As a nation, the United States is lagging behind in math, but educational startups may soon change that. Another popular startup Khan Academy recently raised $5 million, which uses videos and practice problems to teach a range of courses including physics and computer programming. In October, Founders Fund invested $33 million in education startup Knewton, which has an adaptive learning algorithm that has been used to power a college online math readiness course. Please follow SAI on Twitter and Facebook.Join the conversation about this story »See Also:Google: Whatever Numbers You've Heard About Google+, They're Way LowThe 20 Best New Startups Of 2011Welcome To Our New SAI: Enterprise And West Coast Reporters!

Guess What The Biggest Topic On Facebook Was This Year

The death of Osama bin Laden. 10 percent of all status updates (in English) mentioned Osama bin Laden in the days following his death, according to a Facebook blog outlining the top ten global trends in 2011. Coming in second was Green Bay Packers beating the Pittsburgh Steelers in the Super Bowl. Charlie Sheen was winning in March, if you recall. Each month engagement centered around the hottest current events. For instance, conversations about the Royal Wedding were really popular during April. Mentions of the marriage shot up 600-fold, according to the Facebook post. This is what your status updates revealed:

The blog post also looked at the memes that emerged this year. In it, you'll see planking -- you know, where people lie down in an unusual place. It hit a spike after Max Key, the son of New Zealand Prime Minister John Key uploaded a photo to Facebook, then celebrities gave the meme a second wind, but then it just sort of disappeared. If you don't know what "lms" is or "tbh" -- then you're clearly not spending enough time on Facebook. Please follow SAI on Twitter and Facebook.Join the conversation about this story »See Also:Here's What Facebook's Management Is Constantly Thinking AboutREPORT: Facebook Buying Gowalla, A Foursquare CompetitorBOMBSHELL: Huge Company Bans Internal Email, Switches Totally To Facebook-Type-Stuff And Instant Messaging

Underdog Social Network, Tagged, Is Building A "Pandora For People"

Today, the valley's underdog social network Tagged acquired startup Topicmarks to help it improve its people-discovery algorithm. Topicmarks has machine learning technology, which takes long form content and summarizes it (so you don't have to read all of the text). Tagged CEO Greg Tseng said "we are a site about meeting new people. We use the same expertise [as Topicmarks], but look at user profiles and comments to extract meaning and match people together." Tagged plans to incorporate the machine learning technology into its algorithm. "Facebook's algorithm is about surfacing the best content from your friends. Facebook is a content-matching algorithm," Tseng said. "Google is for web pages, Netflix is for movies, ours is for people. We call it internally a Pandora for people." Unlike Facebook which helps you stay connected with friends, Tagged helps you meet new people by engaging in social games or giving you a way to flirt online. Tagged is really popular in the middle of America where urban interactions are sparse. But 70 percent of Tagged users are international, and are spread over 200 different countries. What's been challenging is getting both people to be interested in each other, Tseng admitted. Unlike Pandora where the recommendation is made to you, the connection on Tagged has to be reciprocated or else it won't work. "We really need to be the best in the world at people matching and thus acquiring Topicmarks fits right into that," he said. The Topicmarks team will join Tagged's San Francisco office, but financial details of the acquisition haven't been disclosed. Please follow SAI on Twitter and Facebook.Join the conversation about this story »See Also:Can Electric Bikes Be A Billion-Dollar Business? This Garage Entrepreneur Thinks SoSean Parker Invests In Music Startup StageItIt Sure Sounds Like Gowalla Screwed Over Investors With A Side Deal To Go To Facebook

Here's Who Was At The Super Exclusive Kleiner Perkins Party Last Night

Top Silicon Valley VC firm Kleiner Perkins held its annual holiday party last night at the fancy Menlo Circus Club, and a lot of high-profile people were there. With an ex-vice president in the house, even spouses weren't allowed in. The spots were reserved for investors and their portfolio founders. Here are some partners and their friends who we saw schmoozing well into the night:

Former VP of the U.S. Al Gore

Co-founder of Sun Microsystems Bill Joy

Former Morgan Stanley analyst (and current Kleiner partner) Mary Meeker

Former Twitter engineering head Mike Abbott, who just joined the firm last week

Longtime partners John Doerr and Brook Byers and founding partner Tom Perkins 

Bing Gordon, who joined a couple years ago and led Kleiner's moves into gaming and social

Venture capitalist Vinod Khosla

Economist Laura Tyson

HP Chairman Ray Lane

Ex-MySpace CEO Owen Van Natta, who just left his role as chief business officer at Zynga last month

Former Hulu CTO Eric Feng -- his new company Erly is a Kleiner investment

Facebook e-commerce leader Jared Morgenstern

Everyone was dressed up except for an entrepreneur with a DropBox sweatshirt and another with flannel jammy bottoms on. Parked outside, of course, was a Fisker -- the super sexy electric sports car from the latest startup trying to wean drivers from gasoline. Please follow SAI on Twitter and Facebook.Join the conversation about this story »See Also:Twilio Raises Another $17 Million To Help The World CommunicateJohn Doerr And Reed Hastings Put $11 Million Into Education StartupCan Electric Bikes Be A Billion-Dollar Business? This Garage Entrepreneur Thinks So

No, Sean Parker Didn't Just Get Ousted From Founders Fund

A source close to Founders Fund recently told us that billionaire Sean Parker was spending less time as a VC these days, and more time building startups. Another source heard that Parker was actually out at Founders Fund, the Silicon Valley VC firm he joined after leaving Facebook in 2006. So we decided to figure out what was really going on. Bruce Gibney, a partner at Founders, says that Parker is still a member of the Founders team. Parker's title at the firm has changed in connection with the launch of Founders fourth fund ($625 million), but his role really hasn't. According to a source close to Founders, Parker is now executive general partner, instead of "managing partner."  But he'll be doing the same thing he has always done: Referring deals to the firm and spending a lot of his time focusing on the operations of one of Founders' investments. In the third Founders fund, Sean spent a lot of time working with Spotify. Now, Sean is spending a lot of time working a new startup called Airtime, which he started with Napster co-founder Shawn Fanning. Please follow SAI on Twitter and Facebook.Join the conversation about this story »See Also:Sean Parker's Wild Birthday Party Involves A Story About A Fire Extinguisher And Security GuardsSean Parker Invests In Music Startup StageItSean Parker: Gowalla Lost To Foursquare Years Ago, And Didn't Listen To My Ideas

Watch Tim O'Reilly Play With This Cool Sonar Device For Blind People

Last night at the launch of startup advocacy group Engine Advocacy, we saw Tim O'Reilly walking around with his eyes closed -- guided only by a sonar device. Standing nearby, the inventor of the device Steve Hoefer watched anxiously, hoping O'Reilly didn't run into anything as the device is still in prototype stage! We asked O'Reilly to show us how it worked:

Please follow SAI on Twitter and Facebook.Join the conversation about this story »See Also:HEY DC, LISTEN UP: We Are The Voice Of Startups In Silicon ValleyYou Won't Believe How Big This Profitable 5-Person Startup IsThis Russian Model Is Headed To Y Combinator