How to see the domestic database SequoiaDB open source

In general, I think there are a few points it

1) Compared to mongo or Chinese complete documentation, as a Chinese coder. English documents to see or headache ah.

2) it should be said that the open source community side of the support is still relatively fast, in the group question basically the same day will be answered, and then in the beginning of the configuration and docking program when the original students also in the area of the hand to teach our engineers. Still very powerful

3) overall use and migration conversion time will not get on, but now it is said that more SQL support, have not tried, sounds very powerful, but their native operating statements are still very good understanding

How to think about yandex open source clickhouse this columnar document database

Yandex in the June 15, 2016 open-sourced a database for data analysis called ClickHouse, which is an extraordinarily big deal for conservative Russians. What's even more surprising is that this columnar storage database runs ahead of many popular commercial MPP database software, such as Vertica.If you haven't heard of Vertica, you must have heard of Michael Stonebraker, winner of the 2014 Turing Award, inventor of PostgreSQL and Ingres (Sybase and Michael Stonebraker, winner of the 2014 Turing Award, inventor of PostgreSQL and Ingres (Sybase and SQL Server were inherited from Ingres), and founder of Paradigm4 and SciDB, founded Vertica in 2005, which was later acquired by HP, which made HP Vertica the high-performance representative of the MPP columnar-storage commercial database that Facebook purchased to use for its users. purchased Vertica data for user behavior analysis.

Simply put, ClickHouse as an analytical database, there are three major features: one is fast running score, the second is more functionality , the third is the literary model

1. fast running score: ClickHouse running score is Vertica's 5 times faster:

ClickHouse performance exceeds most of the columnar storage databases on the market. Compared to traditional data ClickHouse to 100-1000X faster, ClickHouse still has a very big advantage:

100Million dataset:

ClickHouse than Vertica is about 5 times faster than Hive is 279 times faster than My SQL is 801 times faster than My SQL

1Billion datasets:

ClickHouse is about 5 times faster than Vertica, MySQL and Hive can't do the job anymore

2. Many functions: ClickHouse supports various scenarios of statistical analysis of data

- Supports SQL query,

- Supports a wide range of library functions (e.g. IP conversion, URL analysis, etc.). For example, IP conversion, URL analysis, etc., estimate calculation/HyperLoglog, etc.)

- Support for arrays and nested data structures (Nested Data Structure)

- Support for offsite replication and deployment of the database

3.Arts and crafts: Currently, there are a lot of limitations on ClickHouse. Born to serve the petty capitalist

- Currently only supports Ubuntu systems

- Does not provide design and architecture documentation, the design is very mysterious look, only open source C++ source code

- Ignore the Hadoop ecosystem, go their own way

How to think of the Alibaba announcement of the opening of open-source AliSQL database

In fact, it is a little bit similar to

Google opened up Android for everyone to use for free

Certain technologies are not hard for others to imitate, and patents don't last very long

Google probably thought it would be better to make it public all at once, so everyone can get it together and capture the market quickly

How to look at hacking into databases

Intranets. Insiders and outside hackers work together to get it done. It's easier if you're an insider.

What do you think of TokuDB, the US-developed database?

Tested TokuMX, performance is really good, but stability is a concern. mongodb 3.0 introduced the wiredtiger engine, and the gap with tokumx has narrowed

Studied TokuMX and the index data structure used by TokuDB, it's very clever. It's a very clever design, and while the tree depth is doubled, the insertion time is really reduced dramatically.

In the end, I didn't adopt them.

How to see the free open source CRM

Free open source CRM is basically difficult to meet the actual business needs of the enterprise, you can consider a support for personalized customization of the CRM, the CRM will be good, it can be customized according to user needs, in the shortest possible time and let the user see the results. Satisfaction and then pay, no worries. The customization tool is simple and fast. Users can operate completely on their own to meet future business changes. In addition, it is based on the SAAS model of the online rental situation, which can save the enterprise to buy hardware, installation and debugging, and the cost of upgrading the cost of the later. Regular after-sales visits can also solve many problems in use.

How do you think Facebook has open sourced React Native

Tom Ohino, a member of the React Native project, published React Native: Bringing modern web techniques to mobile (off the wall), which describes the design concepts of React Native in detail. Ohino argues that Native is a must at this stage, despite its higher development costs, because the web's user experience still can't outperform Native:

1. Native's native controls have a better experience;

2. Native has better gesture recognition;

3. Native has a more appropriate threading model. Although Web Worker can solve some of the problems, such as image decoding and text rendering still can't be rendered in multiple threads, which affects the smoothness of the web.

What Ohino didn't mention is that Native can achieve richer and more detailed animation effects, which is ultimately a better human-computer interaction experience at this stage. I think these examples are persuasive and are a direct result of the emergence of React Native.

Figure 3 - Ohino shared React Native at F8 (Keynote)

Learn once, write anywhere

"Learn once, write anywhere" is similarly from Ohino's article. Because the user experience is different on different Native platforms, React Native doesn't require a single piece of native code to support multiple platforms, so instead of "Write once, run anywhere" (Java), it proposes "Learn Once, write anywhere" is proposed instead of "Write once, run anywhere" (Java).

Figure 4 - "Learn once, write anywhere"

This figure is a schematic diagram drawn by the author based on his understanding that, in order from the bottom up:

1. React: writing React-based code on different platforms. "Learn once, write anywhere".

2. Virtual DOM: as opposed to DOM (Document Object Model) in the Browser environment, Virtual DOM is a lightweight representation of the DOM in memory (ligheight representation of the document in the original language) that can be generated by different rendering engines UI under different platforms, JS and Native communicate with each other via Bridge (React Native communication mechanism explained ? bang's blog).

3. Web/iOS/Android: Web and iOS platforms have been realized, and Android platform is expected to be realized in October 2015 (Blog | React).

React, mentioned in several places in the previous article, is Facebook's 2013 open source web development framework, I was looking through its release, and found this paragraph:

Figure 5 - from the React release (2013)

1. Highlighted text shows that 2013 has been in the development of a prototype of React Native, and now It's kind of a thick build.

2. Another recent hot project is Flipboard/react-canvas - GitHub (see @rank for details), where the rendering layer uses Web Canvas to improve interaction fluency, similar to the first attempt above.

React itself is a huge topic, see facebook/react Wiki - GitHub for more details.

The author believes that "Write once, run anywhere" is still necessary to improve efficiency, and is similar to "Learn once, run anywhere". "Learn once, write anywhere" is not in conflict, we are internally remodeling the existing component library and HybridAPI, so that it is adapted to (complement) React Native components, so as to write a code that can be run on iOS and the Web, to be open-sourced after maturity!

Continuously update

Continuously updated...

II. Planning

The following figure shows the business and technical transformation for React Native:

Figure 6 - Business and Technical Transformation Figure 6 - Business and Technical Transformation

Bottom up:

1. React Node: React supports server-side rendering, usually used for first-screen server-side rendering; Typical scenarios are multi-page lists, first-screen server-side rendering page-flipping client-side rendering to avoid initiating 2 requests when requesting a page for the first time.

2. React Native base environment:

2.1. Framework Integration: Although React Native released Integration with Existing App document, integrated into the existing complex App will still encounter a lot of details, such as integration into the Tmall iPad client. For example, the integration into the Tmall iPad client took 2 days for the iOS students in the group.

2.2. Neorking transformation: the main thing is to re-establish the session, which is usually stored in the header cookie, React Native provides network IO fetch and XMLHttpRequest does not support rewriting cookies. fetch extension, either by the native is responsible for network IO (there is a session mechanism) and then through HybridAPI called by JS, temporarily chose the latter.

2.3. Caching/packaging solution: As long as there are resources loaded from the server side can not avoid this topic, React Native is the same, caching is used to solve the second access to the resource loading performance, packaging to solve the first access to the resource loading performance.

3. MUI is a set of component libraries, which will be transformed using the idea of complementing React Native components.

4. HybridAPI is a set of Hybrid APIs from Ali, which have been shared on multiple public occasions (e.g., sensors @ Hang JS) without further elaboration, and React Native has established its own communication mechanism, which seems to be more efficient (not verified), and the cost of transformation is not significant.

5. The fastest one will be online in mid-April, through the first few business transformation to promote the transformation of the overall system, if the effect is as expected will start a larger-scale business transformation.

More detailed planning and progress, as well as performance, stability, and scalability data will follow.

III. Risks

1. Although Facebook has 3 apps (Groups, Ads Manager, F8) that use React Native, it is not known if there will be a change in the Appstore's policy with the large-scale adoption of React Native, we can only take a step forward.

* Update:

2015.7.28 AppStore review policy adjustment: allow dynamically loaded code running in JavascriptCore, the following figure is the previous review policy, compare the highlighted part of the change.

Does qt support homegrown databases

Apps need to manipulate databases in many cases. qt supports a variety of databases, but in many cases you need to install a DLL driver. This is a bit of a pain in the ass. I think the result of trying to use MYSQL was that the driver was hard to install, and then I used SQLITE. if the requirements for the database are not very high, Sqlite should be able to meet the needs.

How to see the database technology to the inevitable development of big data technology

With the accumulation of data, some of the recorded object of the business state of the data is more and more, so the slow formation of the industry's big data, of course, some of the big database, there is a place for usability, some of the big data is a piece of garbage.

Please adopt!