Home > other >  Automation of the cluster configuration deployment and update scheme
Automation of the cluster configuration deployment and update scheme

Time:09-27

First take some time to nag I do's "cloud" XX industry production demand, and then combined with a series of problems with these needs consult everybody experts,

functional requirements:
Production enterprise internal LAN to connect the office environment and related production equipment, and have connected to the Internet exports, the production process from the office environment (B/S architecture) to the device in a network to send the production order, production can the running state of real-time echo for office environment, in meet the demand of the network will need to deploy a regular Web application service site and relational database (it is not need real-time database) can be completed,


multi-tenancy requirements:
"XX industry production cloud" offer "multi-tenancy" mode, each related to XX industry production enterprises provide SaaS cloud services, is in the form of "cloud", because each factory production process in the industry and business logic basic similar, from a technical perspective, can use the same set of code logic, and the same database structure,

Stability and real-time requirements:
Due to the production enterprise to have certain real-time data and operation requirements, and the network (Internet service operators) interrupted for a short period of time can't affect production progress, so the factory internal need to deploy a server, the Web service for the office environment (B/S architecture) and connect the related production equipment,

operational requirements:
The use of the service factory hundreds, if each factory to deploy a server and provide related services for the factory, the difficulty of such operations so many servers and a lot of work, what's more, the system construction in the early service version upgrade (such as code patches, database structure change, install the software, etc.) will be more frequently, a large number of server upgrade would have a lot of problems, such as service interruptions during upgrade, upgrade version is inconsistent impact failure after the upgrade, upgrade interface to support both versions, etc., during

my plan:
In order to meet the above requirements, I put the production cloud is designed to "cloud + side" two parts, the "cloud" provide SaaS services for real-time demand is not very tall;" End "deployment at each factory within the local area network (LAN) to provide service for real-time demand is higher, the" cloud "expand, forming a" cloud "multiple" end "of the model (the" client "and" end "in the Internet of things there is a difference between the concept of), each factory" end "have the same logic code and data structures, different may be only the IP address, MAC address and data in the database,


Deployment scheme is as follows:


question:
1, this kind of similar to "cloud + end" the pattern of the industry how to implement? There is no mature and case? If so we can consider to buy or cooperate,
2, the purpose is to extract common SaaS cloud and concentrated to provide services, in order to integrate resources, centralized management, its deployment is generally done by cluster computing, and "end" is distributed, very close to grid computing, cluster computing and grid computing, the contradiction between two modes coexist?
3, "the end" whether can use the CDN technology to reduce the difficulty of management?
4, "cloud" to "end" of all the management and control, through the technical way to implement? With the cluster management tool Puppet/Linux Chef/Ansible/SaltStack/Cfengine? Or a Zookeeper? Or Google Borg? Or a more popular CoreOS now? (must be combined with the above requirements)
5, since CoreOS is lite version of Linux, so are there any restrictions on the installation of software and services? A single file updates are efficient and convenient?
6, if install CoreOS "end", its application and the database how to deploy? Main consideration upgrade of the data structure and different "end" to store data,
7, "cloud" and "end" the need for data synchronization between (asynchronous), is there any better data synchronization solution or product support? ETL tools such as open source Kettle? Or MQ products, such as supporting the MQTT protocol?

A great god, please grant instruction!

CodePudding user response:


Don't know to help
  • Related