Linux Finebi cluster deployment

In enterprise applications, the configuration of a single computer is usually limited, and enterprise applications are high concurrency, which increases the number of concurrent computers in a computer cluster, thus improving the performance of the overall application service. A cluster is a service that uses multiple computers as a whole to provide relevant applications. Finebi supports the cluster deployment of multi-computer services, which utilizes limited computer resources to effectively improve the concurrency performance of the overall application through cluster deployment. This paper mainly introduces the idea of the whole Finebi cluster.

As a case of an architecture that finebi in, this approach works through NFS file sharing.

650) this.width=650; "Src=" http://dl2.iteye.com/upload/attachment/0123/8094/ 9de956b9-9cb9-317c-8ace-1d79cd5e9898.png "style=" border:0px; "/>

Web Engineering Storage Issues

The storage of Web engineering, we have to solve is that multiple servers are guaranteed to read the same Web project. We can make multiple physical hard disk components through Ceph to a logical hard disk, so that all nodes are accessing the same address, or you can access the same Web project through the NFS shared file service with Linux itself. Whichever way you use it, we guarantee that:

1) access to the same Web project

2) The cube storage address is consistent

Because the same Web project requires the storage address of cube to be consistent, it is required that the cube storage address must be the same.

When used, the implementation of CEPH requires at least three computers to be implemented, while the actual enterprise application is less than three, and NFS can be Linux itself, so use the "NFS" scenario.

system data configuration

in the case of a single node, using caching and the way the file system of the operating system is used to save data, No longer appropriate in cluster mode. The main reason is the consistency of data, multiple nodes may be read and write simultaneously, change the system data, will eventually cause the overall data inconsistency. The best solution is that the system configuration data is all given to a relational database such as MySQL to manage. But because of such a large amount of engineering, the main reason is because many code is lack of maintenance, hasty changes may bring unexpected bugs. So we adopt a compromise approach. Select a few points in the cluster as the primary node, or M for short. The remaining nodes serve as child nodes, referred to as S. All actions related to changing the system configuration on s are sent to M for processing. M is responsible for changing the system state and maintaining the consistent state of the entire system. The S node discards all cached data, reads the state, no longer reads its own data, but sends a read request to M to get the data on M. The M node itself can have cached data. Other data s nodes are equivalent to the M node, and there is no dependency.

650) this.width=650; "Src=" http://dl2.iteye.com/upload/attachment/0123/8096/ Baa30683-3821-3573-9922-4c79a0ec6d48.jpg "style=" border:0px; "/>

Therefore, according to the above, we provide the following solutions:

1)MySQL database : The existing Finedb configuration information in the original Web project is dumped into the MySQL database. The MySQL database does not exist because the Finedb database can only have one connection and cannot read multiple nodes at the same time. Logdb also need to be migrated;

2) Master node : We use the parent node to configure the cluster, the system data changes are on the primary node, the child nodes only read the data on the master node;

3)Zookeeper: In order to ensure the read-write situation, the master node to ensure data consistency, but also need to Zookeeper communication, as a file lock function.

Load Balancing

In the Finebi cluster environment, we can use any server that supports load balancing to complete the task of rotation and ensure that the session is sticky. Here we use the Nginx reverse proxy, using IP identity round-off, to ensure that the same user in the same session. (in the case of a single server node, the same IP will ensure that the session is sticky).

FS Platform Cluster

Using the FS platform cluster plug-in, the FS platform configuration can meet the cluster requirements. In the FS platform cluster, all operations of the FS platform are sent to the master node, and the child nodes are just the compute servers.

Linux Finebi cluster deployment