Databases Reference
In-Depth Information
Figure 1: The problem of broadcast temporal information
powered on. Thus, some versions may not be able to be received by some clients.
This leads to the possibility of a temporal inconsistency among stored contents at
client sides. That is to say, the problem of maintaining the temporal consistency of
contents in data broadcasting systems is caused by the fact that some clients cannot
always receive all the versions.
Figure 1 shows an example, in which some valid times of versions stored at a
client become inconsistent because the client could not receive all the versions. In
this example, each rectangle and each time interval (e.g. [6/1,6/30]) associated
with it denote a data unit and the data unit's valid time, respectively. Here, the
data unit [The schedule on June] was first created and transmitted to a client on
June 1st. The valid time of this data unit was originally assumed to be [6/1,6/30].
At the server side, assume that a certain event (on June 15th) in the data unit was
cancelled on June 10th. Since this data unit is updated, the valid time of the previous
version (say version 1) of the data unit is changed to [6/1,6/10]. Assume that the
client equipment was, however, not powered on at the time, and that the client
could receive neither the updated data unit (version 2) nor the document (version
1) whose valid time was changed. Then, if the data unit [The schedule on July]
whose version number is 3 is created and transmitted on June 30, the valid time of
the data unit (version 1) stored at the client becomes to contain wrong information
and leads to cause a temporal inconsistency.
2.2
Version Trees for Updates of Contents and Valid Times
Here, we propose a version control mechanism
5)
, in which for each data unit, both
of updates of it's content and valid time are represented by a binary tree at the
