As mentioned earlier, the Web Service is a technique selected based on the loose application between the Internet client and the server. That makes them a natural selection that forms a grid next-generation application. However, remember that Web Service can only do limited work. In fact, the simple Web Service (now described by W3C) is not helpful to establish a grid app. However, after entering the grid service, there is a big improvement on the characteristics and services on the basis of Web Service.
Let us take a look at these improvements through a simple example. Imagine your organization has a cluster server that can perform a large-scale calculation. However, this cluster server is located in the headquarters of Chicago, and you need to use this cluster server computing power in the parties in New York, Los Angeles and Seattle. This looks a good solution for Web Service.
We can provide operations similar to SolveReallyBigsystem (), SolvefermatslastTheorem (), etc. via a mathematical WebService called MathService. First we will be able to perform a typical web service call:
Call MathService requires it to perform specific operations. MathService gives the cluster server to execute the operation of the operation. MathService returns the result of the operation.
Everything is going on so far. However, let us a little more realistic. If you plan to access a remote cluster server to perform a complex mathematical operation, you probably not perform one step, but a series of computational operations. However, Web Service is boundless and non-dispatched. "Borderless" means that Web Service does not remember what you have done from a call to another. If we want to perform a string of related operations, you must send the result of an operation as the parameter of the next operation. In addition, when we solve the boundary problem (Some Web Services Actual Work Around this Problem), the web service is still non-temporary, which means they are longer than their clients. This will indicate that when a client is used to complete the Web Service, all the information recorded by the Web Service can be accessed by the next client. In fact, another client can access Web Service when a client is using the Web Service, and potentially hinders the operation of the first client. You can affirm that this is not a very good solution.
Factory
Grid Services can solve the previous two problems by similar to Web Service. I actually use a central MathService plant instead of borderless MathService that is shared by all users, this MathService factory is responsible for managing a series of MathService instances. When a client needs to call MathService operations, it will notify this instance instead of the MathService factory. When the client needs you to create (canceled) an instance of an instance will communicate with the factory.
The figure above shows that each client does not have to have an instance. An example can be shared by two clients while accessing two instances. These instances are temporarily present because their lifetime is limited (eventually will be revoked). The life period of each instance can be different depending on the application. This works so each client has its own instance. However, there are other schemes we need to be shared by multiple users (Scenarios), and this instance will be revoked within a certain period of time when no user access is available.
Other improvements in grid services
It is the most significant improvement provided by the factory for the mesh service. However, grid services have more advantages:
