Shulou(Shulou.com)06/01 Report--

This article introduces the knowledge of "FineBI cluster deployment under Linux". Many people will encounter this dilemma in the operation of actual cases, so let the editor lead you to learn how to deal with these situations. I hope you can read it carefully and be able to achieve something!

In enterprise applications, the configuration of a single computer is usually limited, and enterprise applications require high concurrency. At this time, the number of concurrency will be improved by the way of computer cluster, so as to improve the performance of the overall application service. Cluster is a service that takes multiple computers as a whole to provide related applications. FineBI supports cluster deployment of multi-computer services, which makes use of limited computer resources to effectively improve the concurrent performance of the whole application. This paper mainly introduces the idea of the whole FineBI cluster.

FineBI adopts the mode of load balancing cluster, which creates multiple servers as a cluster server. These problems are encountered here: 1) the storage problem of web project: FineBI in the cluster requires multiple servers to read the same web project because of its own problems. Therefore, it is necessary to achieve web project sharing. 2) system data consistency: during the operation of FineBI, there are read and write operations, and some data configuration files are written to the database. It is necessary to ensure the consistency of system data in the case of cluster. 3) load balancing: on the one hand, load balancing is used to deal with the problem of session, on the other hand, to achieve a load balancing cluster environment, the proxy server can be used to forward requests to the servers within the cluster, and load balancing and proxy server caching technology can be combined to provide useful performance. 4) FS platform cluster: if FineBI uses FS platform, various configurations of FS platform also need cluster configuration.

The following figure is a case diagram of the architecture entered by FineBI, which handles the web project through NFS file sharing.

Web Engineering Storage issu

For the storage of Web project, what we want to solve is that multiple servers guarantee to read the same web project. We can use ceph to make multiple physical hard disk components a logical hard disk, so that all nodes are accessing the same address; we can also access the same web project through the nfs file sharing service provided with linux itself. No matter which way we use, we need to make sure that:

1) visit the same web project

2) the Cube storage address is the same.

Because the storage address of cube is required to be the same under the same web project, the storage address of cube must be the same.

When it is really used, the implementation of ceph needs at least three computers, but in actual enterprise applications, three are rarely used; and nfs can and is linux itself, so the "nfs" scheme is used.

System data configuration

In the case of a single node, the use of caching and storing data through the operating system's file system is no longer appropriate in cluster mode. The main reason is the consistency of the data, multiple nodes may read and write at the same time, change the system data, which is bound to cause the overall data inconsistency. The best solution is to leave all the system configuration data to relational databases such as MySQL to manage. However, due to the large amount of work, the main reason is the lack of maintenance of a lot of code, hasty changes may bring unexpected bug. So we adopted a compromise approach. Select a few points in the cluster as the master node, referred to as M. The rest of the nodes act as child nodes, referred to as S. When all the operations related to changing the system configuration on S are sent to M for processing. M is responsible for changing the state of the system and maintaining the consistent state of the whole system. The S node gives up all the cached data, and when reading the state, it 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 M nodes, and there is no dependency.

Therefore, for the above reasons, we provide the following solutions:

1) mysql database: the configuration information of finedb in the original web project is transferred to mysql database. Because the finedb database can only have one connection and cannot be read by multiple nodes at the same time, the mysql database does not exist. Logdb also needs to be migrated

2) Master child node: we use the way of the master child node to configure the cluster. All changes to the system data are made on the master node, and the child node only reads the data on the master node.

3) Zookeeper: in order to ensure the data consistency of the master and child nodes in the case of read and write, zookeeper is also needed to communicate and act as a file lock.

Load balancing

In FineBI's clustered environment, we can use any server that supports load balancing to complete the task of rotation and keep the session sticky. Here we use the nginx reverse proxy, which uses the IP logo to send in turn to ensure that the same user is in the same session. In the case of one server and one node, the same IP ensures that the session is sticky.

FS platform cluster

Using the FS platform cluster plug-in, the FS platform can be configured to meet the cluster needs. In the FS platform cluster, all the operations of the FS platform are sent to the master node; the child nodes only act as computing servers.

This is the end of "FineBI Cluster deployment under Linux". Thank you for reading. If you want to know more about the industry, you can follow the website, the editor will output more high-quality practical articles for you!

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.