Cloud Computing Week Jevin Review (2.13-2.18)

The architectural challenges behind the tumblr:150 of billions of months of browsing

Like many new websites, the famous Light blog service Tumblr is facing the bottleneck of system architecture in the rapid development. 500 million times a day, peak 40,000 requests per second, 3TB of new data storage, more than 1000 servers, in this case how to ensure the smooth operation of the old system, smooth transition to the new system, Tumblr is facing enormous challenges.

Tumblr was very typical of lamp applications. is currently evolving into a distributed services model based on Scala, HBase, Redis (well-known open source K storage), Kafka (Apache project, distributed publish-Subscribe message system from LinkedIn), Finagle (by Twitter's open source fault-tolerant, protocol-neutral RPC system), there is also an interesting cell-based architecture to support dashboard (csdn Note: Tumblr's distinctive user interface, similar to the microblogging timeline).

Tumblr's biggest problem now is how to transform it into a massive web site. The system architecture is evolving from lamp to the most advanced technology portfolio, and the team also has to develop new capabilities and infrastructure from a small start-up to a fully armed, ready-standby formal development team.

37signal Founder: SaaS start Easy to harvest

As an application development company that was born with Ruby on Rails, 37signals is 13 years old. A few days ago, one of 37signals's founders, Jason Fried, published an article about SaaS experience. In 2006, Jason Fried was included in TR35 (awarded to the world's greatest innovator, who was not more than 35 years old and therefore named) by the MIT-issued journal Marvell Review.

Hacker's way: Everything can be automated

Refusing to repeat the work and replacing it in an automated way is the hacker's way!

In many companies, human resources and finance departments will process a large number of spreadsheets every day. I think that every month tens of thousands of people repeat similar things, wasting time. If they have a bit of programming experience, I think they will agree with me and try to find a better solution. This is what I call the mentality of recruiting non-developer positions. Even if they can't design their own solutions, they should know that there must be a better solution!

Solving ideas

1. Quickly define requirements

Record details of each fee, including date, amount, and description

can query by month or by person, generate report

Keep copies of data

2. Consider the relationship between the bill and the workflow, which is a major problem

Invoices may be physical or electronic

Costs are randomly generated

If everyone keeps their own custody for one months, the bill may be lost

Each person must fill in the monthly expense item, takes too long, the efficiency is too low

3. More brain, less pain in a smart way

To keep everyone from keeping receipts for a long period of time and looking for electronic invoices painfully from thousands of emails, we can consider storing the invoices electronically.

NASA's big computer era has ended, salute!

NASA announces that their big computer age is officially over. The large computer model, which was held until the last minute to leave office, is IBM Z9, a Marshall space Flight Center. NASA says they are not bad, still reliable, powerful and safe, but that their size is as large as a fridge, a "must change" situation. But anyway, these large computers have made a lot of contribution to human space exploration, so we must remember them well.

Super computer simulations show that a nuclear bomb could detonate an asteroid colliding with Earth.

This picture shows the impact of a mega-ton nuclear explosion on a planet-threatening asteroid.

Scientists are designing a way to avoid the end of the real world, and they believe that nuclear bombs can make the earth avoid collisions with asteroids. It sounds like a disaster movie plot, but it's true.

Physicists at the Los Alamos National Laboratory used the world's most powerful supercomputer to estimate the destructive power of nuclear explosions to space rocks. While the United States and other national aerospace departments have mapped the nearest asteroid's position, one of them has caused a catastrophe by unexpectedly crashing into the earth, so they believe that preparing for this potential danger is necessary and a significant plan.

Scientists at Los Alamos National Laboratory have used supercomputer Cielo to complete this extremely complex simulation experiment. The supercomputer has a total of 32,000 processors, making it capable of computing 1350 trillion times.

Secrets of Free cloud storage: contributing redundant storage

Free cloud storage sounds unlikely, but a new enterprise, Symform, can provide users with a new type of storage service free of charge, which can be used to build a distributed storage network.

After joining the network and providing 150GB of local space, you can use 100GB storage space anywhere on the web. In simple terms, this seems like an unfair deal. But having a cloud storage point is attractive, prompting many businesses and individual users to start using cloud storage services. Morgan Stanley's survey of cloud users last year showed that all types of cloud services grew at 60% a year, while cloud storage usage increased to 67%.

By adopting Remote Storage services, organizations simplify their backup and disaster recovery processes. If you have a data backup outside of the data center, this will help you mitigate the impact of the local disaster on your business. With remote data backup, you can still be complacent, even if your storage volume has multiple failures.

The unveiling of New Hampshire office

In the New Hampshire State Nashua City of Dell, there is a Facebook-style graffiti wall, the company encourages employees to create, and employees draw a star to the name of NetApp.

Dell has a big goal: to be the top three storage providers, but they have to surpass their fifth-ranked rival,--netapp, a company in the California Sun Valley.

As a 60 billion-worth Dell, this goal is not remote. It's just that in most people's minds, Dell is still more connected to laptops, and there's still a lot to go to become a leading enterprise-class storage provider. Dell's office in Hampshire originally belonged to one of the 5 storage companies EqualLogic Division,dell had acquired over the past 5 years. Once upon a time, it was the 1980 office seat of computer giant Digital equipment Corporation. Like Dell's frequent acquisitions, Dell hopes to be as successful as Digital Equipment Corporation in its heyday.

Employees here hope to reproduce the history of Digital Equipment Corporation by posting posters in the corridor. "You will soon see many very interesting innovations," said Lazarus Vekiarides, director of software engineering. ”

Just as the posters on the walls, or the sights on NetApp's name, are like a totem representing the new Dell. Dell is doing what it can to shift the focus of business to servers and storage, and acquisitions are a great way to achieve goals quickly.

Commercial companies are moving data centers to virtual servers, which necessarily reduces the need for physical servers, and the market for server manufacturers, including Dell, is squeezed. However, virtualization will bring a larger storage market, so Dell, HP and IBM are redoubling their efforts in the storage market.

Microsoft Store in India compromised by black user password

Microsoft's online store in India was suddenly covered by hackers ' pages today, and the security of its Web site was also a concern. According to the Wpsauce website, the landing page of the Microsoft Store India Sub-station was temporarily occupied by a team of hackers named Evil Shadow and placed in the Windows merchandise page their bizarre v-geek (movie, English translation: V for Vendetta) page. The team also cracked the user database and converted the user's account password to a text file, proving that the site had been cracked by the team and claiming that "unprotected systems will be cleansed".

Flash era: Traditional mechanical hard disk hard to meet the needs of cloud computing

Flash memory will be an important part of cloud computing and virtualization, and it's time to discard traditional hard drives. Cloud computing and virtualization have dramatically changed the server infrastructure, but the real push for a recovery in the storage industry is the rise of solid-state drives and flash memory.

Flash will continue to bring much-needed impetus to the enterprise sector. Flash is becoming more and more important from Fusion-io last year's IPO, to Nutanix and Tintri, which have recently emerged in a series of venture-backed start-up companies.

With the increase in the number of applications and databases that are deployed through virtualization and cloud computing, traditional hard disk storage sequences are becoming a serious performance bottleneck and are gradually being marginalized by the market. Virtualization and cloud computing infrastructure can reduce server usage, which will have a greater impact on infrastructure based on mechanical hard drives.

Although the server, network, and processor performance follow Moore's Law, the performance of the hard drive has not changed much. Flash, on the other hand, can dramatically improve performance. For example, conventional solid-state drives are 400 times times faster than mechanical hard drives. This is also the Amazon (microblogging) AWS Cloud computing Services of the new version of the DYNAMODB database using solid-state drives, and abandoned mechanical hard drive reasons.