Client request to send smaller amounts of data, each request within the 200 bytes, the server returns the amount of data, too small, the server USES the Linux + tomcat, every second will be received about 200 from different client's request, can a VPS host? Or need to separate the server? VPS host need hundreds of dollars a year, and independent server requires a few hundred dollars a month, his play can province province, want to ask everybody's experience,

CodePudding user response:

This to the actual pressure test, if the server to access the file or database, it is easy to read and write in the file or database server bottlenecks,
According to my guess, general enterprise Web applications, a server is not a second processing 200 HTTP requests

CodePudding user response:

You should play have a workstation computer can support

CodePudding user response:

Due to ajax polling mode takes the server, so I decided to use HTML 5 WebSocket, but WebSocket need to solve the problem of heavy even without Internet, online articles is to use the heart reconnection mechanism, but I think this plan is not perfect, because the heartbeat interval is short, the server pressure will still is bigger, it is better to use ajax polling, and longer intervals, real-time and will discount, I think the server is generally stable, broken network is mainly on the client side, so I want to a method, the client code timing to query the network is connected, but requests are not sent to my server, because my server is too weak, but to a certain well-known website, to apply for a relatively small volume of resources, if the application is successful, indicate that the network unobstructed, WebSocket are basic sure won't hang, otherwise indicate that the network is broken, the WebSocket will certainly die, so to my server initiate reconnection, such concurrent let some famous website is equal to 200 times per second to bear, you said this way?

CodePudding user response:

High concurrency core personal think, is to reasonable use of caching, reduce unnecessary IO.
So, depending on the kind of service you provide, and in the right layer Settings cache. If this all the same 200 requests that you directly after the first request will result in a cache, behind the 199 requests is very fast, server stress is small.
But if the 200 requests is different, to calculate returns results, that is about to consider distributed, message queue technology. Such as still depends on the specific business.
This blog is quite good, can see: https://blog.csdn.net/DreamWeaver_zhou/article/details/78587580

CodePudding user response:

refer to 3rd floor big dog's reply:

since ajax polling mode takes the server, so I decided to use HTML 5 WebSocket, but WebSocket need to solve the problem of heavy even without Internet, online articles is to use the heart reconnection mechanism, but I think this plan is not perfect, because the heartbeat interval is short, the pressure will still is bigger, it is better to use ajax polling, and longer intervals, real-time and will discount, I think the server is generally stable, broken network is mainly on the client side, so I want to a method, the client code timing to query the network is connected, but requests are not sent to my server, because my server is too weak, but to a certain well-known website, to apply for a relatively small volume of resources, if the application is successful, indicate that the network unobstructed, WebSocket are basic sure won't hang, otherwise indicate that the network is broken, the WebSocket will certainly die, so to my server initiate reconnection, such concurrent let some famous website is equal to 200 times per second to bear, you said this way?

Or the type of concrete analysis under the 200 requests set server after receipt of a request to do, suit the remedy to the case.
Heartbeat mechanism itself, moreover, no problem, but the advice is not to do so, back to deal with a lot of problem, client management, break line reconnection, data is broken package, glue bag and so on, to use ready-made communication protocol framework, such as the Mqtt protocol (there are various language ready-made server and client implementation library).

CodePudding user response:

What server you use? Tomcat?
The configuration of asynchronous?
You of the request processing need to access the database? Or memory cache

If the configuration asynchronous, but heart treatment, no database access, single core is sufficient to support, you can use apach ab test

CodePudding user response:

Professional reply thank you friends, I am doing an online chess game, design allows a maximum of 200 people online at the same time to the game, if the heartbeat reconnection mechanism, interval of 1 second, then the server will need to handle 200 clients in a second time request, and I have no economy ability has a powerful server, let alone what cluster and distribution, board games communication are frequent, but a small amount of data and the few hundred bytes each time, most of the time don't need the IO and database operation,