The Incommon Federation wiki has moved.

Please visit the new InCommon Federation Library wiki for updated content. Remember to update your bookmarks.

Click in the link above if you are not automatically redirected in 15 seconds.



You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 44 Next »

Deployers have multiple metadata aggregates from which to choose. This page outlines available options. Policy considerations and general configuration issues are discussed on the Metadata Consumption page. Guidance on how to configure specific Metadata Client Software is available elsewhere in this wiki.

All aggregates listed below are production-quality metadata aggregates.

Metadata Aggregates

Obtain a diff of consecutively published metadata aggregates via web or email

Late in 2013, InCommon Operations introduced three new production metadata aggregates at the following permanent HTTP locations:

All metadata aggregates are signed using the same metadata signing key and the SHA-256 digest algorithm. To verify the signature on an aggregate, a consumer must obtain an authentic copy of the InCommon Metadata Signing Certificate.

Operationally, structural changes to metadata are first introduced into the Preview Aggregate and subsequently synchronized with the Main Aggregate and the Fallback Aggregate, in that order. Time between synchronization events depends on the nature of the structural change.

Current Status of InCommon Metadata

Differences between the various metadata aggregates are recorded and archived daily:

  1. Vertical diff between two consecutive instances of the Main Aggregate

  2. Horizontal diff between the Main Aggregate and the Preview Aggregate

  3. Horizontal diff between the Fallback Aggregate and the Main Aggregate

The vertical diff captures changes to metadata over time. The horizontal diffs record the flow of metadata through the preview-main-fallback pipeline.

Metadata consumers choose exactly one of the above aggregates depending on the immediate requirements of their deployment.

 

Availability

Stability

Reliability

Affinity

Preview Aggregate

24x7

experimental

leading edge

persistent

Main Aggregate

24x7

stable

mainstream

persistent

Fallback Aggregate

24x7

legacy

trailing edge

transient

Multiple metadata aggregates allow InCommon to deploy changes to metadata more quickly, easily, and safely.

The Export Aggregate is described in a separate wiki topic.

Preview Metadata Aggregate

The Preview Metadata Aggregate helps manage the introduction of potentially breaking changes into InCommon metadata. Before such a change is deployed to the Main Aggregate, it is first introduced in preview mode. Any issues that surface are addressed before the change is synced with the Main Aggregate.

The Preview Aggregate is designed for deployments on the leading edge, such as test and dev deployments. Such deployments are strongly encouraged to consume the Preview Aggregate instead of the Main Aggregate.

Consume Main or Preview Aggregate?

An important decision point for each deployment is whether to consume the Main Aggregate or the Preview Aggregate.

Main Metadata Aggregate

In the best possible world, a deployment would configure itself to refresh its metadata store from the Main Metadata Aggregate and that would be the end of it. The problem is that metadata aggregates are brittle by their very nature, that is, a small change to metadata may have unexpected effects downstream. If this happens, a deployment can “fall back” to a previous version of metadata that is known to be backward compatible.

Fallback Metadata Aggregate

As suggested in the previous section, the Fallback Metadata Aggregate comes into play when a breaking change is inadvertently introduced into InCommon metadata. When a change is made to the Main Aggregate, and that change breaks a downstream metadata process, an affected deployment can temporarily migrate to the Fallback Aggregate. This gives the deployment time to adjust to the breaking change.

Consume the Fallback Aggregate?

A deployment should consume the Fallback Aggregate only when it has to, that is, when it is unable to consume the Main Aggregate. Consuming the Fallback Aggregate is a temporary measure while a deployment reacts to a breaking change introduced into InCommon metadata.

The Fallback Aggregate is transient in the sense that backward compatibility is provided for a limited, predetermined period of time. This forces deployments to adjust to breaking changes to metadata albeit in a controlled environment.

 

#trackbackRdf ($trackbackUtils.getContentIdentifier($page) $page.title $trackbackUtils.getPingUrl($page))
  • No labels