1. First of all, redis can be deployed on any node, depending on the specific application of the author;
If you want to deploy on a stand-alone machine and pursue faster query efficiency, it is fastest to deploy the business logic application on one node (probably the java server you are talking about). Because it is a stand-alone access, use the Unix domain socket protocol. The interaction speed will be faster;
2. Figure out your needs. Do you need distribution? Can a single redis node meet the demand? Is the local memory sufficient?
If the local redis is sufficient, do not use any high-end distributed system. The more complex things are, the more likely it is to cause problems. If distributed is added, the performance will be damaged, and it is generally not as good as a single server (a single server can meet the conditions. premise);
2. After determining that distributed deployment is required, it is not recommended to use the distributed solution that comes with the redis cluster. The redis version released in 3.0 is currently not used in many real productions; the most widely used one is to use Twitter's Twemproxy as a proxy to connect A bunch of redis nodes;
Refer to this article: http://www.infoq.com/cn/news/2014/11/open-source-redis-cache?utm_sourc...
Which server should redis be installed on? The leader said some time ago that it would be deployed on a java server, and recently it was said that it would be deployed on a database server
Any one will do, it mainly depends on your deployment ideas. How do you plan to do it? And you said there are only two machines? Still want to do distributed?
The main thing to look at is the memory. Distribution will write data separately on multiple machines, so the memory and data capacity must be considered.
![[Web front-end] Node.js quick start](https://img.php.cn/upload/course/000/000/067/662b5d34ba7c0227.png)
1. First of all, redis can be deployed on any node, depending on the specific application of the author;
If you want to deploy on a stand-alone machine and pursue faster query efficiency, it is fastest to deploy the business logic application on one node (probably the java server you are talking about). Because it is a stand-alone access, use the Unix domain socket protocol. The interaction speed will be faster;
2. Figure out your needs. Do you need distribution? Can a single redis node meet the demand? Is the local memory sufficient?
If the local redis is sufficient, do not use any high-end distributed system. The more complex things are, the more likely it is to cause problems. If distributed is added, the performance will be damaged, and it is generally not as good as a single server (a single server can meet the conditions. premise);
2. After determining that distributed deployment is required, it is not recommended to use the distributed solution that comes with the redis cluster. The redis version released in 3.0 is currently not used in many real productions; the most widely used one is to use Twitter's Twemproxy as a proxy to connect A bunch of redis nodes;
Refer to this article:
http://www.infoq.com/cn/news/2014/11/open-source-redis-cache?utm_sourc...
Redis already supports Cluster, please read the documentation directly.
http://redis.io/topics/cluster-tutorial
Any one will do, it mainly depends on your deployment ideas. How do you plan to do it? And you said there are only two machines? Still want to do distributed?
The main thing to look at is the memory. Distribution will write data separately on multiple machines, so the memory and data capacity must be considered.