John, on the AIS Software Catalog...

pbourgeau — Fri, 16 Nov 2007 17:42:00 GMT

Hello,

The following is a contribution by John Ellis, one of our senior program managers. He is responsible for one of the most important aspects of our service: the Software Catalog. This exerpt describes in detail how our Catalog operates, and how it can lend value to your repoting experience.

=======

The software catalog is comprised of “signatures” and “annotation”. The signatures are the raw properties that have been inventoried from computer systems (e.g. Publisher, Title, Version etc.) for a specific piece of software. The annotation is data that has been supplied by researchers to provide more information about the piece of software. Initially, this annotation is limited to specifying a category for the application, but in the future this will be expanded to provide more information and value.

In addition to adding information, the annotation supplied by researchers also addresses inconsistencies in the signature data. For example, in an ideal world, a software company would always use the same name in the publisher field of the software signatures it produces. However, in the real world this tends not to be the case. Even Microsoft uses “Microsoft”, “Microsoft Corporation” and, sometimes in the case of updates & patches, supply no name at all in the publisher field. The annotation provided by the researchers operates on a case-by-case basis and supplies an appropriate name for the publisher which leads to vastly increased consistency, readability and organization of the data in reports etc.

In fact, the benefit of this consistency is so great that systems and processes have been evolved to “alias” publisher names and add the signature to the software catalog even if the signature itself has not been categorized. This explains where certain entries will appear in reports as “uncategorized” even if the publisher name has been aliased.

Another key role of the researchers is to identity public/private signatures. Users of the service will always see the data they have submitted – whether nominally public (e.g. a commercially available software application) or nominally private (a line-of-business application) – as a set. Great care is taken to ensure that customer data is restricted to being viewed by that customer only. However, there are major benefits in extracting the public signatures from the collected inventory – if only so that that the researchers can supply the desired annotation. As an integral part of this activity, the researchers indicate which signatures are public to ensure that when the list of public signatures is used for other aggregate reporting and systems management purposes, the others are excluded.

The distribution of signatures over customers has a classic “long tail”. There are certain signatures that are reported by many/most customers, but these represent a tiny percentage of the total. Conversely, there are huge numbers of signatures that are reported by only one, two or three customers ….. viewed as a graph, these would appear as a “long tail”. From the service perspective, the return on investment from focusing on signatures in this “long tail” is very low, and in many cases these signatures are restricted to a particular customer. However, the return on investment from focusing on the popular signatures is very high since these are almost guaranteed to be public commercial software or freeware. As a result of this distribution of software signatures, the number of customers reporting a particular signature is a major factor in establishing the priority to add further annotation.

When a signature is displayed as “uncategorized” it is generally in a queue waiting to be categorized. If it is considered to be restricted to a particular customer, it will likely never be categorized. If it is popular, it will be categorized as soon as possible. Over time additional systems and processes will evolve to accelerate the rate at which software signatures are annotated. In the first release of the service, there will a significant latency as these processes are developed. It is likely that the catalog will not undergo substantial update for 2 or 3 months while new signatures are being received & processes evolve. After that time, updates on a weekly basis are a reasonable expectation.

“This posting is provided “As Is” with no warranties, and confers no rights. “

System Center Online : catalog

John, on the AIS Software Catalog...