Database Reference
In-Depth Information
der control, we can clearly identify the boundaries of
AuditLogs
,
Er-
rorLogs
, project control data, and the Service Repository itself. This considera-
tion requires a precise definition of Inventory Endpoints' interface(s), suitable for
all types of Governance actions.
• To continue with Governance, a services metadata harvesting tool is a really ef-
fective feature. However, ask yourself: if you, as an architect, have been devising
an enterprise Service Inventory for a while now and still need to perform reverse
engineering to reveal hidden dependencies, then how valid is your service tax-
onomy? Could it be that you overlooked some services/entities during the design
phase? No, we are not saying that a harvester is unnecessary; that's not the point.
Indeed, you should run it regularly (it's an Ant task in JDeveloper), and if in OER
Asset Editor's search report you see anything new or something that you haven't
seen in the
Unsubmitted
/
Unfilled
folder, you can proudly ask for a raise.
• Whatever technical realization you choose for our Service Inventory (building
blocks such as services or components; languages such as Java, C, PL, and SQL;
and so on), your primary concern is the availability of your Service Registry/Re-
pository for the flawless support of Discoverability, along with other utilities from
the first table in this chapter. Please take a second look at the design rules in this
table.
What would be the most logical approach to address availability and performance issues
expressed in the preceding bullet points? Correct, to position SR as close to the composi-
tion controller as possible. But how would this be possible? Only by the segregation of the
service taxonomy model from its physical realization. Doing so, we should be able to re-
implement it on any technical platform, easily accessible by the concrete composition
controller. Remember, Secure Gateway in DMZ is an ESB too, and Service Broker is
common to both. Thus, service metadata lookup is not an extraordinary feature. Will you
query your production OER or OSR from DMZ? Think twice. You could have the discov-
ery node in DMZ, but what about the physical storage? Oracle DB? Again, think twice
when it comes to security. You should be quite close to the iron. How will you securely
synchronize your Production and DMZ discovery nodes?
Runtime lookup
One of the practical ways to classify services and artifacts' taxonomy is to detect the type
of data we see on every service layer. For vertical infrastructure layering, we suggest that
you use the Oracle AIA service layer notation. This leaves us with three main layers: Ad-
apters, Enterprise Services (usually hosted or available through ESB), and task-orches-
trated services in the Enterprise Business Flow layer. Note that vertical stratification for
Search WWH ::
Custom Search