"Enabling the Australian data grid"
Data federation
Data federation refers to combining geographically dispersed data
from different sources into one virtual filesystem. The advantages of
this approach is that the user does not need to know where the data are
and can access files from different collections.
SRB zones
The SRB is scalable and can work in a stand-alone configuration or
in a
federated way with several SRB zones and associated MCATs. In a
federation, each SRB zone is part of a larger network and several zones
interact with each other. Each SRB zone has its local MCAT, SRB servers
and its own resources and users. From the user’s point of view, a zone
is represented as a folder within a logical file system. The user then
can move from one folder to the other without being aware that the two
‘folders’ are different systems and geographically in different places.
Figure 2 shows how two zones can be established in two locations to
form a data grid.
SRB Zone B comprises Server D and MCAT-SRB Server C which maintains Data Resource 2.
Let's suppose that an application or a user in Zone B requests a file located on the Data Resource 1 in Zone A (the arrows show the typical flow of data).
Server B redirects the request to Server A (2).
Server A retrieves the file from Data Resource 1 (3)
Server A services Server D directly (4).
In this setup all requests are made to local SRB servers however the challenge resides in efficiently synchronising all MCATs.
