Hot week: Don't blame open source technology it's innocent.

Hot week: Don't blame open source technology it's innocent. Posted 23 hours ago | Times Read | SOURCE csdn| 0 Reviews | Author Wang Cloud hotspot open source Google Amazon IBM Transistor MongoDB Summary: This week's content: Open source technology is important, please do not give it for security problems, for the load after the 42 "monster" level problem, how should we solve? Google GCE vs. Amazon EC2, who is stronger, and Google and other Internet companies will become innovation centers? Four misunderstandings and four guesses based on IaaS; it is rumored that IBM will redesign the transistor; the Apple cloud stores the most dust-ridden users.

The content of our CSDN cloud channel is still wonderful this week, and the hotspots are still frequent. This week's Hot topics: open source, don't blame open source technology it is innocent; operational dimension, Load big escape: 42 Road monster Coalition and Seven escape rules; Cloud platform, Google GCE vs. Amazon EC2: Fast from Google's original story; Internet companies such as Google will become the center of innovation? New technology trends, IBM redesign transistor is expected to mature within 5-7 years.

1. Don't blame open source technology it's innocent.

Last month, OnRamp's free advertising service was forced to shut down as a result of hacking attacks, which had a severe impact on millions of of sites. OnRamp's parent company, OpenX, issued an official statement at the forum questioning the security of open source technology.

But the industry is more inclined to think: this is not an open source problem, and we should not put the blame on open source users and manufacturers. Open source economics and productivity make it almost the mandatory component of any modern software application. We have great benefits in open source-rapid development, re-use of validated components, allowing users to focus more time on proprietary software features. We cannot give up open source as the answer to the problem.

Now, we should see the real problem is how to reduce the risk and component consumption while achieving the benefits of open source. Therefore, we should strengthen the whole software development lifecycle protection measures on the component layer, and improve the integrity of the whole software supply chain.

2. Load escape: 42-way Monster Coalition and seven rules of escape

Horizontal scaling and vertical scaling help us deal with a large number of high load problems, yet excellent programming is still noticeable. An application with a design defect may not be able to perform in a low load situation, but as the load increases, various problems follow. In this article, Todd Hoff showed the reader the 42 "monster" problems that occurred after the load increased.

The heavy load of the "Monster Corps" is now, how the developers to complete the program design, so that the program is easy to expand and have high reliability? Let's take a look at the "7 rules for the attack on the Load Monster", which is, of course, what needs to be noted under the program's low level of expansion: limiting the proportion of resources used, Merge Aggregation, Delete Aggregation, Batch Aggregation , change Aggregation, Integration Aggregation, uninstall.

Of course, an article is unlikely to include all the problems that arise after the load is increased, and it is impossible to include all the solutions. But it is certain to do enough knowledge of their own system, based on sufficient information to make design decisions, to reduce the system after the expansion of the problems caused.

3. Google GCE vs. Amazon EC2: Fast from Google's own heritage

GCE (Google Compute Engine) has been widely watched by cloud-field workers since its release and is regarded as the most important competitor to the giant Amazon's service EC2 elastic Compute. But what is the performance? Take a look at the contrast between the two services SCALR founder Sebastian Stadil.

At boot speed, the GCE virtual machine can be successfully logged in less than 30 seconds of deployment and startup. On AWS, it takes about 30 seconds to get it running, and then you still have to wait a long time for the boot, at least 120 seconds, and up to 300 seconds. GCE is 4-10 times faster than AWS.

In terms of capacity, Amazon can increase and decrease the capacity of an instance at any time as needed, but not on GCE, at least not currently. The GCE write performance is 2-4 times faster than the two service read performance equivalent.

500M files in two different regions, AWS takes 242 seconds, and GCE only needs 15 seconds, Google fast Amazon 20 times times, in the same region to copy 500M files, AWS need 86 milliseconds, and GCE only need 20 milliseconds, GCE and AWS 4 times times faster.

GCE's powerful performance is certainly desirable, but don't forget that it is still in beta, so strong performance should be attributed to Google's own heritage. And the giant Amazon, after years of development has clearly formed a complete set of service system, which means that you can easily complete the establishment of large applications, if not worry about being locked, EC2 is a good choice.

4. Internet companies such as Google will be the center of innovation?

Google this month acquired the Canadian neural network start-up company Dnnresearch, which was founded last year by Geoffrey Hinton with his two graduate students, Alex Krizhevsky and Ilya. Geoffrey Hinton is a leading figure in the field of deep learning.

Figure: Geoffrey Hinton (right)

Geoffrey Hinton is the latest in a series of Google talent acquisitions. From the founder of Google X, Sebastian Thrun, and Andrew Ng,google, who leads Google's knowledge map, has absorbed a number of top scientists in deep learning, will the 21st Century Innovation Center shift from academic institutions to internet giants?

5. IBM redesign transistor is expected to mature within 5-7 years

Transistors are not only located at the core of the chip, but also the foundation of our High-tech society. IBM has proposed a new generation of transistor models, and IBM's breakthrough is to bring a new coating to the transistor, allowing the device to read ion signals rather than circuit signals. This is important because it helps chip makers put more transistors on a chip.

Technology giant IBM has not actually produced a chip that uses a new type of transistor, instead it just shows a rough circuit design. IBM expects the technology to mature over the next 5-7 years and believes it can actually be produced using the same production process as it is now.

6. Leading Google, Amazon, dropbox,apple cloud storage users most

Apple has the largest number of users in the cloud-storage market, about 27%, Strategy Analytics, and Dropbox, Amazon, Google and others followed, accounting for 17%, 15% and 10% of users respectively. Strategy Analytics's Digital media director, Ed Barton, said streaming video needs could become the next battleground.

7. MongoDB Release 2.4 version: 692 Update and 8 great pride

Following the release of the MongoDB 2.2 release in September 2012, 10gen released the latest MongoDB 2.4 version on March 19, 2013. Since the release, MongoDB's weak data structure model, easy to expand, open source has been widely loved by data workers, and in the Db-engines rankings, MongoDB is the first NoSQL database, a well-deserved popularity of the king!

Compare the previous version of 2.2,10gen to MongoDB 692 improvements, see the revised document. In these updates, 10gen enumerates 8 of the new versions, including hash-based fragmentation (hash-based Sharding), fixed arrays (Capped Arrays), text Search, GEO features, faster counts, The Working Set Analyzer (sharable Set Analyzer), the new V8 engine (newer V8 Engine), security, and so on.

8-12 respectively: In the War of Hadoop MapR won a new round of 30 million U.S. dollars financing, for talent Yahoo! and PINTERESR the same day to buy a recommendation system company, ten years 600 million dollars! Amazon Cloud computing likes to get the CIA big single! , OpenStack Foundation COO Mark collier:2014 will surpass AWS, start-up Appneta:4 time to get VC favor, financing 47.8 million dollars

"The First China Cloud computing Conference" will be held in June 2013 5-7th in the Beijing National Convention Center. Slam The Register!

If you have a wealth of practical experience, creative issues and are happy to share the results with others, welcome to volunteer or recommend lecturers; related questions and suggestions please @csdn cloud computing or join 147015990 (QQ Group: China Cloud Computing) to participate in the discussion.

The fifth session of China Cloud Computing Conference