Project: about 2000 meters in every 2 seconds send a data to the server, the server computing and then returned to the instrument, that is to say, for 2 seconds to complete the data of 2000 times to accept - operation - send,
Before a project is 4000 tables 60 seconds a back and forth, the ActiveMq accumulate data, this time for a backlog of solution

CodePudding user response:

Well, no plan, squeeze the data just said you not enough consumption

So all you need is to speed up the consumption, rather than other options

CodePudding user response:

Hundreds of ten lines of code, you can write their own implementation one master server and n worker task scheduling server communications, which are the worker get a task, then the results back to the worker, such as 3 or 4 orders,

CodePudding user response:

Then put the results back to the worker - like & gt; Then put the results back to master such as

If feel the generic framework is too complex, too trival, too slow, a small percentage itself with its code can write a more suitable for their own,

CodePudding user response:

Maybe the ans say not so good, but you just need to think about actual everyday we all see the actual scene

Every subway hall, high-speed hall, a bunch of people -- -- -- -- -- -- -- -- -- -- -- -- that is what you said "backlog"
Now it rains, each reservoir full, some roads submerged half a meter -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- is what you said "backlog"

So what is the problem?
1. It is not enough capacity scheduling, subway trains is not enough, not enough drainage, water is not enough, so not to say that in the above scheme, but to speed up below scheduling
2. Don't be afraid of "backlog", is the purpose of building the reservoir regulation, the subway hall is also a relay control, road drainage and regulation, we are not asking the reservoir is empty "backlog", also does not require the subway hall not a wooden man with a "backlog", also not request city there is no water in the gutters "backlog", they will not overflow the design requirements are required, if the overflow, it opened, diversion, flood, have to flush (human means are the same, control the upstream, interpretation of the downstream, a release path is not enough, just add a few more free path)


Always ActiveMq things here is actually reservoir, there are capacity is right, regulating peak was right, he is to protect downstream, if you feel too much capacity, think humans do? Obviously not find alternative to reservoir, but current limiting, diversion, flood discharge

CodePudding user response:

Don't know whether you slow application or database,

Scheme,
1 load balancing, the same set of programs, you can deploy multiple sites, and then let the request evenly dispersed to each webapi,
2 as sp, create a database of primary and secondary synchronization, a master and slave N, are generally belongs to the category of read/write separation,

CodePudding user response:

Please tell your details,

CodePudding user response:

If it is a server, that nothing;
If it is multiple servers, distribute, perform, statistics, to return