Shulou(Shulou.com)06/02 Report--

[toc]

Virtualization KVM

The more mainstream virtualization technology, but in the relatively poor level of operation and maintenance companies, you can also see pirated VMware products. If the test environment is slightly larger, consider using a cloud platform, such as various popular Stack. If you have the ability, you can use the API provided by libvrit, and you can write your own management platform. Suitable for internal use, a variety of open source Stack are relatively heavy, there are some costs to use.

Docker

Unifies the development and online environment, suitable for micro-services.

Public cloud

The public cloud has everything you need. In a company that uses the public cloud, there is no need for operation, maintenance and testing to help enterprises quickly complete the DevOps transformation. The company does Dev and the public cloud does Ops.

Public cloud has corresponding products at the IaaS, PaaS and SaaS levels. If you really understand what you need, the company can choose the right product according to its own situation.

For security reasons, you can use sub-accounts to do some API-related operations.

Storage service

If the amount of data is large, you can consider using open source distributed storage, which will be used later on private cloud platforms or Docker or K8S.

WEB Container OpenResty

OpenResty is not a new thing, but a combination of Nginx and Lua. Zhang Yichun, founder of OpenResty, embeds Lua script into Nginx. Combined with Lua, the functions of Nginx can be flexibly extended, such as implementing WAF and so on. And do much more than that.

Database section

It mainly includes relational database and non-relational database. Basic SQL skills are still needed.

MySQL

Known as the most popular database in the world, it is still widely used. It seems that the title of "most popular" is not bragged. Operation and maintenance staff are most concerned about how to ensure the security and high availability of data.

What needs to be done at the operation and maintenance level is:

Data backup data recovery master-slave synchronization

There are still many tools on the market for backup and recovery. Just choose what you are familiar with.

PostgreSQL

Known as the most advanced database in the world, there are still a lot of users, but most of these users are abroad and are not as widely used as MySQL at home. If you don't think you have a suitable database to choose from, choose PostgreSQL.

Redis

The mainstream key value storage database. It is generally used to cache relational databases.

MongoDB monitoring article Zabbix

The more famous open source monitoring system has powerful functions. If you want to customize or secondary development, it is relatively difficult (we need to have C language (back-end) and PHP (front-end) skills), the difficulties mentioned here are for the operation and maintenance staff.

Open-Falcon

Open source monitoring system by Xiaomi, written in GE language. If you want to customize or secondary development is relatively easy, support scripting languages (Bash, Perl, Python, etc.) to extend or secondary development. These scripting languages are no stranger to operators, but proficiency in using them is another matter.

A good-looking UI

The above monitoring software, their UI interface is not very Nice, you can use the more popular Grafana.

ZMON

Can do monitoring and index monitoring.

Prometheus

Metrics monitoring of the application. Four golden indicators:

Latency: time spent on service requests; for example, average delayed traffic / throughput of HTTP requests: measure service capacity requirements; for example, errors in processing HTTP requests per second: measure the occurrence of errors; e.g. HTTP 500error saturation: measure resource usage; e.g. CPU/ memory / disk usage

With metrics monitoring, coupled with service call chain tracking, the effect will be better.

Service chain call tracking

Problems solved by service chain call tracking:

Provide call chain performance and latency analysis between services Zipkin analysis of service dependency analysis

Service call tracking system launched by Twitter

Jaeger

Service call tracking system launched by Uber

Skywalking

Service call tracking system launched by Huawei

Configuration management section Ansible

Compare lightweight (relative to Puppet and SaltStack) automated configuration management tools. Let the operators say goodbye to the task of using scripts to loop to complete batch operations, but really use parallelism to do operation management.

Programming language chapter

As a member of the IT world, no matter what kind of work it is and can't speak a programming language, I'm embarrassed to say hello to others.

Python

Python's slogan is: "Life is short, use Python."

Golang

Golang has the performance of C and the posture of Python. Its slogan is: "Life is short, let's Go."

Lua

OpenResty needs to be used in Lua, so it's a good idea to master it.

JavaScript

No matter what position it is now, there must be some JS, otherwise there is still some distance from the whole stack. If not, want to do a page, but also to find front-end colleagues, how inconvenient.

Code hosting

With the programming language and the corresponding development environment, there should be a place for code hosting. Gitlab is recommended as an internal code hosting service here, and I believe many companies have chosen the same way.

Development framework JAVA related framework

When it comes to Java development, there are a few components:

JDKMaven (Ant or Gradle) Nexus code managed database (relational database, non-relational database)

Spring family buckets need to know what they have and how they work. Such as:

Service registration and discovery of load balancing circuit breaker what is going on? configuration center Spring BootSpring CloudPython related framework Django

A large and comprehensive WEB framework in the Python world.

Flask

A small and beautiful WEB framework in the Python world.

CI/CD

This has a large proportion in the daily work. If this piece is not done well, the operation and maintenance will be more painful. There are many CI/CD tools and choices in the market, so it is recommended to use Jenkins or self-research.

Operation and maintenance gadget

Usually, you can develop some gadgets, which can help yourself or developers to get some information quickly, such as:

By parsing the registration information of Eureka, you can know where your service is deployed and the port you listen to copy files from one machine to another, from one database instance to another, and synchronize data to another database instance. There are many summaries.

After the operation and maintenance staff want to do the basic work well, they can use more time and energy to do more valuable things. The basic work is mainly as follows:

Continuous integration and continuous deployment infrastructure monitoring service performance monitoring

After the above work is done, we can make some internal operation and maintenance platforms to facilitate colleagues to develop more efficiently.

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.