Fusion Registry

Metadata Technology is proud to release Fusion Registry Version 7.  The Registry now supports SDMX version 2.1 and has been completely redeveloped to provide far greater scalability, faster submission processing and faster query response times. The redesign has taken advantage of Metadata Technology’s Component Architecture, a loosely coupled plug and play architecture, allowing the Registry to retrofit to existing databases – taking advantage of the GUI, web-services and validation support without having to replace an existing database.

Fusion Registry was performance tested for the time taken to process submitted DSDs. For the tests a number of DSDs were created each with the following properties:

• Each DSD had 6 dimensions.
• 3 dimensions used existing codelists from the Registry. Each of these codelists has 10 codes.
• The remaining 3 dimensions were new codelists (for each DSD) with sizes of 100, 200 and 300 codes.

The following tests were performed:

1. DSD Submission Test (batches of 10)

   DSDs were submitted to the Registry in batches of 10. Each batch consisted of 10 DSDs, 30 Codelists, 10 Concept Schemes resulting in each batch having a total number of structures of 50.

   26 batches were submitted, resulting in the Registry containing 260 DSDs with a total number of structures of 1300.

   Each batch took between 16 and 24 seconds to be processed. This time was irrespective of whether there was content already within the Registry or not. The average time for importing a batch was 20 seconds, which is 2 seconds to process each DSD and its related code lists and concepts.

   This test was repeated 20 times and the results averaged.

2. DSD Submission Test (batches of 100)

   DSDs were submitted to the Registry in batches of 100. Each batch consisted of 100 DSDs, 300 Codelists, 100 Concept Schemes resulting in each batch having a total number of structures of 500.

   10 batches were submitted, resulting in the Registry containing 1000 DSDs with a total number of structures of 5000.

   Each batch took between 171 and 177 seconds to be processed. This time was irrespective of whether there was content already within the Registry or not. The average time for importing a batch was 173 seconds, which is 1.7 seconds to process each DSD and its related code lists and concepts.

After each batch was imported a query was submitted for a random DSD and requested that in addition to the DSD all cross-referenced structures of that DSD are to be returend. Each query took between 0.1 and 0.5 of a second to process. The query time was unaffected by the number of structures within the Registry and did not increase due to an increased number of structures within the Registry

All tests were performed on an Amazon cloud instance, running Ubuntu server 10.10 with 2Gb of allocated memory.

The Videos are work in progress and more will be added over the next couple of weeks:

Set-up – Click to Watch tutorial
This tutorial shows how to set up the Registry to use a persistent database for metadata storage, and how to enable security.

Multilingual – Click to Watch tutorial
This tutorial shows how to create structures with multilingual names and descriptions in the Registry.

Subscription Notification – Click to Watch tutorial
This tutorial shows how to create subscriptions for both structure and data changes in the Fusion Registry. A demonstration of both email and HTTP Post notifications is shown, along with some sample code to consume an HTTP Post message.

To view the forum click here.

To view the online help click here.

Fusion Registry runs within a standard Apache Tomcat instance (version 5 or later). For optimal performance it is recommended that the Tomcat instance is supplied with at least 1Gb of Memory.

It runs as a Java web application on the Tomcat server with an Adobe Flash front-end.