Information Technology Reference
In-Depth Information
We consider the two types of boot behaviour: normal boot and soft boot. In normal
boot a node completely suspends all its operations on media, media signalling and
community communication; it drops all previously accommodated soft state and starts
anew. On soft boot a node suspends only etiquette defined behaviour and drops only
etiquette related state. Immediately after a normal boot a network element does not
have any active workflow; it needs to receive either media or media signalling to
launch one. However, etiquette behaviours are fully specified by the refinement proc-
ess above and can be used right after any type of boot, for example to collect commu-
nity context. On soft boot a network element has no memory on trust establishment,
all even on-going workflows are considered as fresh, i.e. the e-rules within a policy
set consider these workflows as just being started. In both boot cases we are interested
in booting of the etiquette behaviour only, thus we no longer distinguish between the
two types of bootstrapping. We introduced soft boot to reflect the process of a new
node joining a community, thus soft boot can also be regarded as booting to a com-
munity.
Community communication that can be done for a number of concerns must be
performant and scalable. Our approach is to use progressive communication patterns
that will gradually evolve to serve more complex tasks by morphing under already
achieved progress within the concerns of interest. Trust establishment is a primary
concern on boot, should follow easy to discover, or standard patterns like those
sketched in E0 - E3. However, after initial trust is established, community peers
could launch etiquette communication for other concerns (QoS, interoperability, auto-
configuration, etc.) using trusted peers — this substitution of etiquette messages (7) -
(9) by messages of another concern we dub etiquette progression. Progression can be
continued with deeper or wider trust establishment itself using fitness function events
to modify etiquette messages as in the next section.
3.4 Evolving to Fit the Community
Situation outlined by (10) will be noted by each community member as stability of
their message boxes and existing etiquette will be perceived as no longer productive.
While trust is the permanent concern the peers will attempt to use progression as
situative re-refinement of e-rules using basic etiquette and fitness function. Figure 6
summarises a possible internal organisation of an autonomic node.
After initial trust as in (10) is established between pairs of community peers the
rule E4 — progression of the etiquette — will be triggered by the local fitness func-
tion event
F
ϕ
=
, meaning that node's fitness is unchanged during time
interval
τ
. Natural etiquette progression suggested here is the exchange between
trusted peers of summaries of established trust relations together with summaries per
workflows. This way, trusted peers can compute [sub-]community fitness and distrib-
ute back to peers their trust connectivity. There is no need to make any additional
computation for this, sufficient will be to distribute to peers residuals as in (10).
The re-refined etiquette is the exchange of residuals from cross operation on local
message boxes between trusted nodes that results in a new state, e.g. (15)
()
const
τ
