To achieve interface stability through design,
the following SOA structure requirements apply:
a. Service
Level Agreements (SLA)s or equivalently named contractual instruments shall
contain Explicit Operational Agreements.
b. Explicit
Operational Agreements shall contain interface definitions.
c. Explicit
Operational Agreements should contain a governance policy
d. The
authentication and authoritization process for who can examine published
contracts shall be standards based..
e. Interface
artifacts shall be published in a registry/repository or other object, e.g.,
ESB.
f. Interface
Definitions shall be standards based, e.g., WSDL, XSD, etc.
g. The
SLA shall contain performance characteristics associated with each interface.
h. SLAs
shall be maintained between the Provider and individual Consumers or classes of
Consumer
i. Versioning
processes may be contained in a separately defined governance process. Governance principles are described in the
NCSF. A NCOIC governance pattern is under
consideration.
To my taste, copied content has almost nothing to do with "Interface Stability". I think that stability of service interface is about how the interface can work in the changing environment, changing behaviour model and related messages. For example, Interface Definitions shall be standards based, e.g., WSDL, XSD, etc
" does not contribute into the stability, IMO, because someone may (should not be restricted from) publish (ing) a non-standardised but immutable (100% stable) interface. Moreover, common (not thought through) use of WSDL leads to constant changes if the interface, i.e. minimal stability. In the essence, it is not a standardisation that important, but the usability pattern is important. And the latter has escaped aforementioned list.
An example of interface stability: 'adding or removing data elements of the exchange messages should not result in the change of interface'. I can say the same thing regarding the operations and use WSDL (in a smart way) to implement this.
Anyway, thank you very much for such interesting material.
- Michael
-----Original Message-----
From: Rex Brooks <
rexb@starbourne.com>
To:
mpoulin@usa.com; 'Ken Laskey' <
klaskey@mitre.org>
Sent: Wed, Mar 23, 2011 3:30 pm
Subject: Re: [soa-rm-ra] meeting time conflict
Hi Guys,
I didn't send the redline version and waited for this latest clean
version of the NCOIC Service Interface Pattern. I'm not suggesting
you go through this in detail to ensure that the principles espoused
in Sections 4 and 5 are actually taken up in this document, but you
could if you wanted to do that. However, I think this provides a
fairly good example of what can be expected of "Testing" with
"Management" in mind. Please do not redistribute despite the fact
that it is not strictly disallowed. I send it to you two because I
think it bears on finishing up your work. Sorry if it's a bit late.
Reminder: This is still a work in progress and this is not yet
approved, but it is likely to be approved.
Cheers,
Rex
On 3/23/11 4:05 AM,
mpoulin@usa.com wrote:
Hi Folks,
I will have a conflict of meetings today: I will be able
(now) to participate in the first and the
last 30 min time-windows
only. Please, plan my presentation on the Management Model
accordingly.
Sorry,
- Michael
--
Rex Brooks
President, CEO
Starbourne Communications Design
GeoAddress: 1361-A Addison
Berkeley, CA 94702
Tel: 510-898-0670