That's it. The core component of the large data development platform, the job scheduling system, then discusses one of the faces of the big Data development platform, the data visualization platform. Like a dispatch system, this is another system that many companies may want to build their own wheels ...
What the data visualization platform is.
But wait a second, what is the data visualization platform. We use this high-end atmospheric grade word, the object and you understand that is the same thing.
It is like a double 11 o'clock, this occupies 200 square meters of the screen, the global curve flying, colorful digital crackling, streamer overflow color, all over the pores are permeated with the spirit of the Internet to win the big screen Carnival system.
Or like a variety of positioning in the future, using three-dimensional hologram map, rotating perspective, where to play, dynamic superposition of various data suspension layer, faint flow out of a winning thousands, the temperament of XX intelligent City System.
Or, is near the naked eye 3D VR reality, far has the hacker Empire Skynet Matrix, the virtual and the reality blend, does not know is the Zhuang Zhou Dream butterfly or the butterfly Dream Village Week Ultimate Numerical materialization space.
Well, yes, it's not.
Similar is the way, are hope to use more rich visual graphic image means, the data more intuitive display. The difference is that the pursuit of visual effects, for the time being, is not within the target range of the visual system I've been referring to, but simply, cool is a plus, but not a core requirement.
So you ask, I'm talking about what this visual platform is all about. Let me change a less highbrow noun: the report system, these words, we should not unfamiliar.
So why am I so sentimental, mysterious, not early to say that the report system these few words.
This is because the traditional reporting system, mostly in the form of tables, or a limited number of legends such as line charts, static display of the underlying data snapshots, usually not too many user interaction capabilities, more is a fixed logic and form of one-way display system.
In order to distinguish the lowbrow image of traditional report system, the report system of target location and function characteristic is improved. I'm sorry to call that name again. At the very least, the title of the BI Business Intelligence system should also be crowned;
So, you see, on the market well-known report class system, not called Bi are not afraid to come out mix, if forced to a little higher, even in the periphery with a little bit of distributed computing technology, it must also be called Agile bi, to show and "old slow" traditional business intelligence system. Then everyone is "agile", what to do. That's going to go to basics, like you play hip-hop, this time to ask, you have free style it. Therefore, the visualization of such low-key and meaningful words, it is gradually popular.
So, summed up, is the report system this name represents the realm, too low. To build four modern large data platforms, we need a more modern data visualization platform than the traditional reporting system. Of course, what matters is not what it is called, but what the product form it is trying to provide behind the name.
And the wheels, the business intelligence systems, what's the problem?
There are many commercial BI products, foreign more well-known products, such as Tableau, Qlikview, Power BI, domestic claims have been agile, such as Yonghong BI, sail soft Finebi and BDP.
In addition, there are products from the Internet industry Company's new recruits, such as Aliyun's quick BI, NetEase's netease number, Amazon's quicksight is similar products. And the Aliyun of the Datav, is running a more dazzling display effect to go, such as we said before the double Xi., smart city and so on, the BI data analysis function is not its focus
From the point of view of the product itself, these commercial products do not have much problem, we build wheels, not because their products do not function well enough, such as the legend is not enough rich, user interaction is not intuitive enough to operate enough and so easy. This ability, is the basis for the survival of commercial products, others home a few Baisi People's team, after several years of more than 10 years of time to develop products, of course, we send a classmate, a short time to build their own wheels can be compared to the. Don't tell us, you see, even the father of the Penguin. In their own public cloud large data package services, provide the products are Yong Hong.
So you have to say that people do not want to pay for commercial products, so they develop it. Also not necessarily, and do not say that the price of the purchase of commercial products and the cost of their own development which high, do not have the money of open source products have a lot of ah, general-purpose, dedicated all have, for example:
Saiku/pantaho system with target positioning as commercial bi substitutes
Airbnb rental company has been playing very high and then open source superset
Kibana for log analysis in the Elk system
The Origin opentsdb for monitoring and Grafana
So, the problem is, on maturity and ease of use you probably do but commercial products, want to "save money" also has open source products, why also want to play their own, is because of idle panic mody. Upstairs of the Airbnb, there are Ali/Tencent/NetEase in the public cloud outside the sale of BI services, is also the development of their own use, we have nothing to do it.
I personally think, fundamentally or because of some application scenarios, commercial products are difficult to match the reasons, at least, for the Division I such "internet" companies are such;
The traditional business bi product, basically is very powerful, but the deployment and learning costs are also relatively high, and often the degree of customization of the process is very high, and the integration of SAP and other product systems have done more in-depth, so basically belong to a more self-consistent and closed system, their goal is to provide you with a complete set of solutions.
and the BI products on the public cloud, although the deployment and learning costs are relatively low (because the function is not mature commercial products so complex), but, from a self-consistent and closed point of view, is similar to the ability of docking external systems (or reluctantly)
For example, most products will provide a custom process from data source collection, cleaning to display, and user rights management, data storage and lifecycle management, some products even data format, are self-contained. In addition, the internal functional components of these products, data structure information, etc., usually will not be exposed to the form of service.
So, in this case, if your data-processing link can be all handed over to the corresponding system to control, or you need to query the display of the data can be completely imported into the corresponding system, or the product through the JDBC interface to query your own management of data, and there is no performance problems, then the problem is not, if not , it will be more difficult to deal with.
And it's almost impossible to integrate with your own peripheral systems on the process. Want to expand the function, such as adding a real-time graphics display capabilities, and development platform flow through and so on, also basically do not think.
As for the existing open source system, although there is no closed problem, but its own business logic is often more fixed and stereotyped, to change the cost is not low, can two development for your use, also depends on your own platform of the process and functional positioning.
Summed up, is the direct use of commercial products, or open source two development, or completely autonomous development, is basically based on your business complexity and the surrounding system you use the ecological environment to decide, usually, your business model is also complex, you need to develop the possibility of self-development is higher. However, you can minimize the overall cost by using different solutions for different scenario requirements.
Finally, you have to ask, commercial or open source products can not be mature enough to fit a variety of complex application scenarios. Theoretically, I think it's possible, but for now, at least for a few years, it's not realistic, because
First of all, in the large data field, the underlying storage and computing engine is very different, far from the standard search mode can be eminence, various business components and processes often need customization and flexible adaptation processing logic.
Secondly, the existing more mature products, its closed logical thinking to break, first of all by its business model constraints, may not be willing, and even if willing, it will take a long time to gradually complete.
Finally, for the combination of large data field application scenarios, to tell the truth, my traditional background manufacturers in this area of research and development capabilities, is to express doubt, this is not a few people's problems, but the way of thinking and product positioning problem. When most of your users have no complex requirements for such a scenario, you have no experience and are unlikely to devote your energies to the needs of a small audience of expert users. This point, the horizontal analogy to see the public cloud service providers to provide Hadoop cluster services are known, basically the most basic and simplest function to meet the needs of the vast majority of small white users, reduce service variables and risks, is their guarantee of product success, customization. Flexible. It's all free of conversation.
Ningbo Micro Whole Hospital http://www.iyestar.com/wzx/
