|Table of Contents|
Figure 1 presents an overview of the components that compose the architecture of CSGrid. The CSGrid Server (Central Server) is accessible to users through a browser. In order to execute the programs (refered to as Algorithms), another type of service is needed, which is called SGA. The SGA is responsible for monitoring execution hosts and for the execution of algorithms on such resources. The CSGrid Server is a Java application running on background (i.e. a daemon), which communicates with SGAs, through CORBA. Each user can execute a CSGrid Client through a browser. The client is also implemented in Java and communicates with the server through RMI.
- number of users (current and projected).
- average number of simultaneous accesses (current and projected)
- average usage of disk space
- availability requirements (e.g. 24x7 availabilidy implies the redundance of some nodes, configuration of RAID for storage, etc).
Table 1 show the profile of each component regarding the usage o memory, CPU, disk space and network. The definition of the terms, high, average and low, are dependent of the context,so Table 2 can be used as reference.
Table 3: Operating Systems
The execution servers (SGAs) can and normally are hosted in multiple servers (which can use different operating systems and hardware configurations). In order to make the system aware of new servers it is only necessary to update configuration files. In the case of the Data Repository, normally the changes are completely seamless to the system.