<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	>
<channel>
	<title>Comments on: SOA - A Shot in the Arm for Registry Fans?</title>
	<atom:link href="http://www.subbu.org/blog/2005/07/soa-a-shot-in-the-arm-for-registry-fans/feed" rel="self" type="application/rss+xml" />
	<link>http://www.subbu.org/blog/2005/07/soa-a-shot-in-the-arm-for-registry-fans</link>
	<description>HTTP, REST and some Cycling</description>
	<pubDate>Thu, 20 Nov 2008 13:53:17 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.7-beta3</generator>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
		<item>
		<title>By: Robin Mulkers</title>
		<link>http://www.subbu.org/blog/2005/07/soa-a-shot-in-the-arm-for-registry-fans/comment-page-1#comment-51</link>
		<dc:creator>Robin Mulkers</dc:creator>
		<pubDate>Wed, 03 Aug 2005 01:47:46 +0000</pubDate>
		<guid isPermaLink="false">http://wp.subbu.org/2005/07/soa-a-shot-in-the-arm-for-registry-fans/#comment-51</guid>
		<description>Hi Subbu,
I discovered your blog today and I like it, we might get into discussions over versioning soon.&lt;br /&gt;
The registry case is incredible, I remember the original goal of UDDI being "&lt;em&gt;dynamic discovery&lt;/em&gt;" and I have been thinking it was only valuable in powerpoint presentations!&lt;br /&gt;
&lt;br /&gt;
I think the registry should be used to provide application level routing, the registry should not just return a URL for a service name, the registry should return a URL for a service name, taking into consideration other variables, like who is asking for the service. The URL returned might be different for a client A and a client B. They just might ask for the same service name but have a different URL returned because for example, those service consumers are not using the same version of the service.&lt;br /&gt;
&lt;br /&gt;
Application level routing might also be used in multi-protocol SOAs, the registry deciding what protocol must be used to access a certain service in certain circumstances taking into consideration technical service level requirements like support for transactions, security context propagation or guaranteed delivery. &lt;br /&gt;
&lt;br /&gt;
Load-balancing, fail-over is more about infrastructure routing, this can be indeed managed by appliances or using software-managed clusters.&lt;br /&gt;
&lt;br /&gt;
In this context, the real alternative for a registry is a message or service broker (Hub and Spoke concept). If the routing is complex (ex:data-dependant), the only solution might be to use a broker but a broker could be a bottleneck and is definitely increased the latency of the communication because it is an additional intermediate in the communication between the service consumer and the service provider while the registry is not.&lt;br /&gt;
Robin
</description>
		<content:encoded><![CDATA[<p>Hi Subbu,<br />
I discovered your blog today and I like it, we might get into discussions over versioning soon.<br />
The registry case is incredible, I remember the original goal of UDDI being &#8220;<em>dynamic discovery</em>&#8221; and I have been thinking it was only valuable in powerpoint presentations!</p>
<p>I think the registry should be used to provide application level routing, the registry should not just return a URL for a service name, the registry should return a URL for a service name, taking into consideration other variables, like who is asking for the service. The URL returned might be different for a client A and a client B. They just might ask for the same service name but have a different URL returned because for example, those service consumers are not using the same version of the service.</p>
<p>Application level routing might also be used in multi-protocol SOAs, the registry deciding what protocol must be used to access a certain service in certain circumstances taking into consideration technical service level requirements like support for transactions, security context propagation or guaranteed delivery. </p>
<p>Load-balancing, fail-over is more about infrastructure routing, this can be indeed managed by appliances or using software-managed clusters.</p>
<p>In this context, the real alternative for a registry is a message or service broker (Hub and Spoke concept). If the routing is complex (ex:data-dependant), the only solution might be to use a broker but a broker could be a bottleneck and is definitely increased the latency of the communication because it is an additional intermediate in the communication between the service consumer and the service provider while the registry is not.<br />
Robin</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Miko Matsumura</title>
		<link>http://www.subbu.org/blog/2005/07/soa-a-shot-in-the-arm-for-registry-fans/comment-page-1#comment-50</link>
		<dc:creator>Miko Matsumura</dc:creator>
		<pubDate>Mon, 11 Jul 2005 11:36:32 +0000</pubDate>
		<guid isPermaLink="false">http://wp.subbu.org/2005/07/soa-a-shot-in-the-arm-for-registry-fans/#comment-50</guid>
		<description>Hi Subbu,

your points are good, the content of this particular piece was focused on helping people understand the distinctions between and the relationship between registry and repository. So the piece really operates at the reference architecture level, and should help people at least understand how to think and talk about these components and how they relate to one another.

The piece doesnt really go into detail on implementation strategies and standards. I've no shortage of opinions on this type of thing, but I've reserved those for a future write up... What are the "things" stored in repositories and how does that help SOA? That's a finer-grained (i.e. Blueprint level) issue. Happy to talk with folks on these topics in more detail, just send me an email...

Miko
</description>
		<content:encoded><![CDATA[<p>Hi Subbu,</p>
<p>your points are good, the content of this particular piece was focused on helping people understand the distinctions between and the relationship between registry and repository. So the piece really operates at the reference architecture level, and should help people at least understand how to think and talk about these components and how they relate to one another.</p>
<p>The piece doesnt really go into detail on implementation strategies and standards. I&#8217;ve no shortage of opinions on this type of thing, but I&#8217;ve reserved those for a future write up&#8230; What are the &#8220;things&#8221; stored in repositories and how does that help SOA? That&#8217;s a finer-grained (i.e. Blueprint level) issue. Happy to talk with folks on these topics in more detail, just send me an email&#8230;</p>
<p>Miko</p>
]]></content:encoded>
	</item>
</channel>
</rss>
