简体繁体 English

如何管理负载平衡

[英]How to manage load balancing

原文 2012-07-24 11:43:41 2 2 load-balancing

There are 10 servers of a particular organization . 特定组织有10 台服务器 。 Give some load-balancing algorithm for it. 为它提供一些负载平衡算法 。 Also consider a case where 1 of the server takes 10 sec to process a request and other takes only 1 sec. 还要考虑这样一种情况，其中1个服务器需要10秒来处理请求而其他服务器只需要1秒。 So, how will you manage this bottleneck ? 那么，您将如何管理这个瓶颈？

2 个解决方案

There are several ways of doing so but one of the best way is LOAD BALANCING BY COST . 有几种方法可以这样做，但最好的方法之一是按成本加载平衡 。 In this type of balancing we assign a cost to each connection from client to server on the basis of the distance between them, policies of sub-networks between them, medium of data transmission between them etc. and after doing so we try to relocate/refer the connections to that server which gives the overall minimum cost of the connections. 在这种类型的平衡中，我们根据客户端到服务器之间的distance ，它们之间的子网络策略，它们之间的数据传输介质等为每个连接分配成本，并且在这样做之后我们尝试重新定位/引用与该服务器的连接，它给出了连接的总体最低成本。

May be a simple solution is: Out of every 10 tasks, assign the 1st one to the server that takes 10 sec. 可能是一个简单的解决方案：在每10个任务中，将第1个任务分配给需要10秒的服务器。 The next 9 tasks goes to the server that takes 1 sec. 接下来的9个任务进入服务器需要1秒。

Of course, here I assume tasks are coming in at 1 sec intervals as an assumption. 当然，在这里我假设任务以1秒的间隔进行作为假设。 With that assumption and above approach both the servers are fully utilized. 通过该假设和上述方法，两个服务器都得到了充分利用。