Information Technology Reference
It should be noted that the number of customers with which the system
may interact (i.e. who may require information) may be a sensitive variation
in time; for this reason, it makes sense evaluating what happens in the
examined scheduling algorithms as the 'number of requests per second'
varies. We note that customers use two independent channels to communicate
with the server node: one channel for sending and one receiving data. It has
been observed that the same object can be simultaneously requested by one
or more users: the queues of the server node that receive the requests must
take this factor into account to avoid redundancy problems; it means that the
queues that are formed in the input to the server are multi-request queues.
More precisely, the multi-request queues are formed by groups of common
and contemporary requests (we can have two or more multi-requests concern-
ing the same object, but they can be at different points of the queue, i.e. they
are present as input to the server at different times). From now on, every time
when we refer to a 'request' we will consider a 'multi-request'. Besides, let
us assume that the required objects are all of the same size. Moreover, the
assumption that the client, after submitting a request, continuously monitors
the transmission channel (which from their point of view is the receiver)
waiting for a response will be of crucial importance. This means that the
server, during the transmission phase, should not wait for the client to
connect to the channel to capture the required flow of information. Finally,
it will be assumed that there are no errors in the communication channel
and we will ignore the propagation delays on the same channel: the delays are
considered small compared with the latency that characterizes the response
by the server during transmission.
The scheduler of the server also acts on the basis of the actions of the
scheduler of the disk and the decisions of the cache manager. Considering
this situation, in the following sections we will illustrate four scheduling
algorithms for the server, with a single algorithm for the management of disk
and some of the most widely used cache management policies.
Disk and server scheduling algorithms
Scheduling algorithms on which basis the server decides the order in which
to satisfy the data requests are essentially derived by the RxW algorithm .
It is suited to satisfy the special needs of data transmission in ubiquitous
environments, even with some variations.
With reference to a disk type of memory, we consider the C-LOOK
algorithm, which collects sets of data to be considered in ascending order
with respect to their position in the cylindrical disk (i.e. about the coordinate