Live build Precautions: load balancing, and distributed deployment

As technical staff, we were often asked the question "Why live program to be installed in a distributed server deployment?" In fact, a small amount of users live in the platform and the amount of data, there is not much traffic, a server completely able to run from a live program, but with all the file data expansion and market growth in subscribers, applications, and databases are placed on a single server is not insured.
In particular, this is live on the server requires a relatively high APP eat more resources, so we may need to increase the server at a later stage, in order to ease the access speed getting slower and slower and gradual reduction of storage space.
Server distributed deployment of
distributed deployment need to split open the live program, set up one after another to the server, there will be a little different between servers, such as business server high CPU requirements, and database servers require more more stable fast read and write speeds and the like.
Split and distributed deployment business is the focus of a stand-alone server increases machines, applications and data services separate, concurrent processing capabilities to improve the site using an application server clustering, database read and write separation, use a reverse proxy and CDN to accelerate web site response, etc. and so on, these are the need to work carried out step by step.
Web load balancing to achieve
a live program reason to be distributed across multiple servers, the biggest reason is that you can achieve multi-server load balancing, the pressure to share a single server to multiple servers and components are protected Web server so that the server to achieve high performance while reducing the failure rate of a single point, to improve stability.
Various methods to achieve load balancing Web
. 1, the load balancing HTTP
user sends a request, Web server returns a new url modify the HTTP response header Location markers completion page redirection to achieve "load balancing." The advantage is easy to implement, the disadvantage is that poor performance in the case of large-scale access to, and increase network delay, reduce the user experience, live platform available early, even if late.

2, DNS load balancing
DNS is responsible for domain name resolution, resolution process is complete DNS domain name to IP mapping a domain name corresponding to multiple IP can make DNS load balancing services, load balancing this way, simple configuration, excellent performance . But may be a delay, a lot of trouble when the IP or machine malfunction, such as HTTP load balancing freedom is not the way.
3, GSLB load balancing
excellent performance GSLB load balancing mode, and supports a variety of configuration policies. However, involve CDN to build and maintain, so the cost will be relatively high, generally can be used to provide CDN services to constituents, and to get paid on time service.
Live program you want to run, we must carry on the server, then today Billiton magic knowledge of science and technology introduce you to Get it?

Guess you like

Origin blog.51cto.com/14528075/2458222