基本上你是什么尝试做的事情是DDB(分布式数据库)所做的工作.
因此，***的选择就是遵循这种方式.
#创建分割文件的算法
#具有中间服务器，该服务器存储的文件部分的索引使其他服务器无法使用.
#查询一旦到来，您就必须在侦听服务器上即时重新生成文件，但不必在此服务器上生成整个文件，为此，一种变通方法是使用一种行为类似于结点的算法并隧道化来自于此的数据其他服务器，同时保持与客户端的准确性.

我认为这不是一件很艰巨的工作，因为如今有太多可用的框架，如果您需要了解更多信息，请尝试寻找DDB的工作方式.

Basically what u are trying to do is sort of jobs a DDB(Distributed database) does.
So the best option is to follow this kind of skleton.
#Create an algorithm that splits up the file
#Have a central server that stores the index of file parts refrencing other servers.
#As soon as the query comes you have to regenerate the file on fly at listening server but its not necessary u generate the whole file at this server and a workaround for this is to use an algorithm that behaves like junction and tunnels the data coming from other servers while maintaining accuracy with client.

I think its not very tough job as so many frameworks are available these days and if u need to know more, try searching for the way a DDB works.