Shulou(Shulou.com)05/31 Report--

This article mainly introduces "what is the demand of docker cluster product form". In the daily operation, I believe that many people have doubts about the demand of docker cluster product form. The editor has consulted all kinds of materials and sorted out simple and easy-to-use operation methods. I hope it will be helpful for you to answer the doubt of "what is the demand of docker cluster product form?" Next, please follow the editor to study!

1. Phenomenon

The emergence of cloud computing is due to the high cost (time, money) for users to build a stable, available and easy-to-maintain server in their computer room. But in fact, according to my own cost calculation of building a home NAS server software, for a slightly larger enterprise user, the cost of both AWS and Aliyun is not less than the cost of building their own computer room, which is the main reason why public cloud can not replace private cloud at present. Security and performance are not really the main factors that determine users' choice of public and private clouds. For small and medium-sized enterprises (Taobao e-commerce), the cost of public cloud is low enough, for large enterprises (JD.com, banks), the cost of public cloud is not enough to offset the cost of hiring engineers to build themselves.

The popularity of docker is due to the fact that it has opened up a self-contained development to running process, reducing the overall development and operation costs. Container products prior to docker did not do this, focusing only on runtime isolation and failing to open up the entire process from development to deployment.

K8s is popular because docker solves the problem of stand-alone, while K8s solves the problem of deploying operation and maintenance containers on a machine cluster. What is solved here is deployment and operation and maintenance, not development and testing. Since it comes from Google, the application scenario of K8s is mainly aimed at distributed applications that solve stateful problems on their own.

two。 Demand

Two misunderstandings should be avoided in thinking in this field.

Henry Ford: if I had asked consumers what they wanted at first, they would have told me,'I want a faster horse!' Break through the appearance and find out what the customer really needs: "get to the destination faster and more convenient"

Since its invention, the battery life of smart phones is not as good as that of traditional phones, and it needs to be recharged once a day. The cost of tesla electric vehicles is so high that it cannot be popularized on a large scale. It is impossible for a new invention to surpass the old one in an all-round way, and there are bound to be defects. As long as we can find a way to offset this defect with advantages rather than solve it, there is no new form that will certainly solve all problems.

In fact, there are already two descriptions of this requirement in the industry, DCOS and serverless. It's the same in essence.

Let developers focus on the development of business logic to solve the problems of operation, deployment and update at the lowest possible cost. The sub-problem of this problem is to solve the problem that programs between different developers communicate with each other, that is, the problem that micro-services have to solve.

Developers need to be given enough freedom to perform all the functions of the hardware. The failure of GAE's sandbox mode illustrates this, with less castration and more abstraction, and it also proves that developers, like other users, have user habits to the language and are more difficult to change.

Developers are lazy, including unwilling to change the internal file system layout (docker) of linux, hoping to operate the whole system through shell

In order to meet the previous three points, additional hardware resources are needed to serve developers. The corollary here is that there is no living space for cloud container clusters in the fields of telecom base stations and embedded systems that need to save computing resources.

The above description considers developers, if only developers, then provide faster horses (here I use the first analogy of the full text, so this sentence may not be true), at the same time, developers can only partially influence the enterprise's IT decisions. The existing IaaS products can fully meet the needs of building some server-side programs.

As an operating system (in a broad sense), Killer-APP is the main reason that attracts developers, users and hardware manufacturers. Windows-office, linux-httpd/MySQL, IOS- Wechat, IBM mainframe-Oracle/DB2.

The definition of super application is that it can only exert its full power and wide range of applications on this platform. Corollary: if an application can solve the problem by running on a stand-alone linux, it will not be a boost to the progress of the cloud container cluster. On the contrary, the cloud container cluster will have to pay an extra cost to be compatible with these applications.

Think according to the above, PaaS in this field, can provide super applications are big data and artificial intelligence, specifically, the field of artificial intelligence is TensorFlow. Big data field, I don't know what it is. Maybe there's something else, waiting to be dug.

3. Market * seed users

Destructive technology can not succeed in the mainstream market and mainstream customers at first. Destructive technology needs to build new value networks and markets. All attempts to replace IOE in the enterprise server software stack market will fail. We need to find a market that can effectively use cloud container clusters, tolerate its shortcomings, and focus on capturing this market, rather than delusional to satisfy all users through a set of products.

The market Tesla found is the high-end sports car market. In this field, the advantage of fast acceleration of electric cars should be brought into full play, the disadvantage of high price should be fully shielded, and the disadvantage should be turned into advantages. Then counter-attack the mainstream market from this market segment, which is still on the way.

At present, the first task is to find these seed users, artificial intelligence users, big data users, or users in other fields, to observe (rather than investigate, direct questioning can only get the wrong conclusion) their behavior, pain points and iterative products. This is not a clear direction in one or two attempts, but requires resources to try several times. Before the release of the iphone, Apple released the newton handheld and the ipod, with both successes and failures.

4. Technical

Continuity technology (the technology that addresses the mainstream needs of current mainstream users) is not important (which is ironic as a technician) because today's technology has addressed mainstream needs very well.

According to mainstream thinking, the most important technology of electric vehicles is the energy storage capacity of batteries, but in fact, Tesla has improved his battery life and energy storage capacity by using BMS (battery network management technology) software. Batteries, buy Panasonic's 18650 directly, and then consider building a battery factory when you get bigger.

Storage and network are the "battery energy storage capacity" in the field of cloud computing (here I use the second analogy in this article, so it is still possible that it is not true). Traditional users value it, and their attributes are complex, so they can take advantage of it (JD.com uses K8s + neutron to build the network instead of implementing the container network himself). We need to sort out a list of technologies that we don't need to pay attention to.

It is not easy to find this technology. Apple has been able to make smartphones because of years of investment in the operating system (Mac OS X) and the purchase of capacitive screen technology. SpaceX can be simulated across ten orders of magnitude in industrial simulation software.

One principle is to find technology through user scenarios and satisfy them with the simplest and most straightforward technologies, rather than creating esoteric and complex technologies to try to capture the market.

5. Product form

The advantages and disadvantages of PaaS can be seen at present.

Advantages and disadvantages certain programs can get through development and operation (single container, complex multi-container applications have not yet been opened) without full stack IaaS capability, do not support complex network and storage operation and maintenance is simple (it is still easy for stateless applications to scale up, complex applications do not complete upgrading, etc.) only supports containers, and does not support process-type applications in large-scale situations Resource utilization exceeds the traditional virtual machine model in the case of a small scale, the resource consumption on the management side is still large, so the study on "what is the demand for docker cluster product form" is over. I hope to be able to solve everyone's doubts. The collocation of theory and practice can better help you learn, go and try it! If you want to continue to learn more related knowledge, please continue to follow the website, the editor will continue to work hard to bring you more practical articles!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.