wsdm message
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]
Subject: Public demo/interop scenario discussion
- From: David Melgar <dmelgar@us.ibm.com>
- To: <wsdm@lists.oasis-open.org>
- Date: Mon, 2 May 2005 13:00:38 -0400
During last week's Thursday call, there
was discussion regarding the scenario for a public interop/demo to occur
in the June timeframe. The venue has not yet been settled. I volunteered
to take the pen on writing up a scenario.
Step one, we need to settle on the outline
for the scenario.
Goals:
- Provide a compelling scenario which
illustrates WSDM's usefulness to analysts and the press.
- Show that having an interoperable
standard supported by multiple vendors can accomplish more than proprietary
interfaces.
Inventory of resource types that we
can work with.
- Blackberry provided by IBM
- Server provided by Dell
- Web services provided by several vendors
including HP, CA, IBM, others etc. Sorry I don't have the complete list.
The demo could consist of isolated scenarios
illustrating particular functions, or it could be a single scenario tied
together through multiple vendors, or a combination of both. The consensus
on the call seemed to lean towards a single scenario tying in multiple
vendors. This is more complex to implement because of the coordination
required, but is a more compelling illustration of the value of interoperability.
The discussed scenario is as follows:
- Premise is that the Blackberry
typically invokes a web service (Weather station) to obtain information.
Multiple servers are available which host multiple weather station web
service instances.
- A manager is monitoring the
response time of various weather station instances.
- A manager (maybe the same manager
or another one) is monitoring the CPU utilization of one or more servers.
The metric being monitored must be something which indicates a slowdown
of the server, not a failure of the server. Failure of the server would
likely mean that metrics are not available from it.
- The Blackberry has subscribed
to events from a management provider(?) which indicate which web service
instance to use.
- The server currently being used
to access the weather station web service becomes heavily loaded with high
CPU utilization.
- A manager detects the high CPU
utilization situation. It obtains response time metrics from other weather
station instances. Based on the response times received, it sends a notification
to the Blackberry informing it of the weather station instance with the
fastest response time.
- As a corollary, one or more
additional weather station instances may come on-line during the scenario.
A management provider sends out an advertisement that this new resource
is available. The manager is able to dynamically add knowledge of this
new resource and add it to the mix when determining the replacement weather
station instance to use.
This demo consists of WSDM messages
exchanges between multiple vendors and disparate resource types, providing
a solution that could not have been accomplished with a proprietary protocol.
This is not much different from load balancing, just using WSDM to do it.
Issues/notes/questions:
- Dell indicated during the call that
they would be unlikely to be able to bring a second server. Therefore some
other vendor needs to support server management capabilities. At a minimum
this is whatever metric is the trigger to indicate server slowdown, ie
CPU utilization. Could be a laptop, although a "server" maybe
more impressive.
- Initial discovery. How will the manager(s)
know what resources are available?
- Who does the Blackberry subscribe
to for notification of the server slowdown? There are multiple management
providers and no coordination point other than the managers.
- Are there other capabilities that
are important to show? The demo should not be too long or complicated,
but should highlight the more important WSDM features.
- The scenario assumes that one of the
web service resource providers is managing a web service resource on the
Dell server. This may require coordination if using the same infrastructure
(.Net), or multiple stacks should be able to coexist on the same server,
ie .Net, Java as long as the ports do not conflict.
- WSDM event format seems like an important
feature to show. Not clear if this scenario uses it sufficiently.
David Melgar
Web Services Toolkit Development
Emerging Technologies
dmelgar@us.ibm.com
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]