This specification details the procedure required to "engage" (that is, to initialise the use of) a cloud service.
Once engaged, the consumer may then, as required, repeatedly send a "work request" (see below) to that service.
This specification, code examples uses the following default namespace: "http://www.w3.org/ns/cloud"
and RIF, the Rule interchange Format, used in the selection of services, uses the following namespace: "http://www.w3.org/2007/rif#"
A service Query is sent by a cloud consumer in an attempt to find
an appropriate cloud service to which future work requests will be sent
This query consists of an "all routers" multicast request, to which
listening services send a response consisting of an XML document
representing a rule-set, drawn from the Core sub-set of RIF [
RIF], which
will, when implemented, yield the charging regime of the service,
expressed in units of US cents per GHz-minute
In this way, a consumer is able, in knowing the required amount of
computation, to estimate the cost imposed by each service for the
desired work
the consumer, after receiving all responses from "interested" cloud
services, may then choose from among those services using whatever
selection policy is desired.
Note that the charging regime provided by a cloud service is
purely a guide: consumers are not required to implement a RIF processor
in order to make a service selection
Once a cloud service has been engaged, the intention of a consumer
to have that service perform some computational work is communicated
in the form of a "work request",
This request consists of an XML document (as shown in
example 1), itself consisting of a single element consisting of an
empty element with the tag "work_request", having the attributes named
"params", containing a reference to the requested computation parameters,
"style", containing the computational "style" of the task and "content"
containing a reference to a model of the computational task to execute
(all these attributes will be explained further below)
The "style" with which a computation takes place consists
of the minimum "availability" of the computation which, itself,
consisting of a proportion of some period, expressed as a
percentage, over which a computation is expected to be running
and responsive.
The availability of a computation is calculated as follows,
where "MTTF" is Mean Time to Failure and "MTTR" is Mean Time to Repair:
Example 2
This example lists the "style.css" from Example 1,
which itself specifies that those elements (of the model given in that
example) having a class of "abc" have a minimum availability of 99%,
and the model element with ID "xyz" should have a minimum availability
of 99.5%
.abc {
availability: 99%
}
#xyz {
availability: 99.5%
}
4.2 Computation parameters
the parameters to effect the model's execution appear
in the form of a "params" element, the element children
of which, each representing a parameter, containing a "name"
attribute, the value of which contains the name of a
parameter, and a "value" attribute, the value of which
contains a reference to the value (the meaning of which
is outside this specification) of the parameter
value
The following example illustrates the parameter list,
containing two parameters named "x" and "y", referenced
by the above work request
<params xml:id="params">
<a name="x" value="#a">
<b name="y" value="#b">
</params>
References
[RIF]
Michael Kifer; Harold Boley RIF Overview. 5 February 2013. W3C Working Group Note. URL: