Browse Prior Art Database

Problem diagnostic technic based on multidimensional dynamic software profile

IP.com Disclosure Number: IPCOM000241231D
Publication Date: 2015-Apr-07
Document File: 3 page(s) / 50K

Publishing Venue

The IP.com Prior Art Database

Abstract

The article describes method of customer's problem categorization and diagnosis based on multidimensional runtime profile of the software instance.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 01 of 3

Problem diagnostic technic based on multidimensional dynamic software profile

Dynamic trace profile base on issues observed during software runtime can be created using the idea described at disclosure http://ip.com/IPCOM/000174090.

By disclosure description, profile manages log/trace level, so when issue appears we have logs/trace updated with data which should help identify root cause of the issue on particular machine. On the other hand, such profile is the valuable information by itself. Profile matches module or function name with dynamically created trace level. What is more, several other dimensions can be also part of this dynamically created profile, like system performance, system parameters (like network load etc).

It is possible to request customers to deliver local profile when PMR is opened. The idea is to analyze PMR database by comparing those dynamically created profiles delivered by customers.

Using such information we can extract at last 2 important information:
- identify possible root cause of the issue base on similar shape profile in different PMR's
- identify common components/functions which causes the issues in different configuration and different user scenarios base on dynamic profile configuration

What is described in this article is method for faster problem diagnostic based on multidimensional profile based on several perspective, like: dynamic tracing/diagnostic profile, performance profile etc. Said profile can be represented by multidimensional function with domain on software call tree, and co-domain in any kind of numerical space.

Problem diagnostic method is based on method of comparison profile's functions.

The main advantage of proposed approach is that it allows to quickly find similarity in problems raised by end user. Different problem may have the same root cause, which is not easy to detect when the debug information are gathered around the raised problem. What is more, proposed method allows to detect more complex interactions

within software between different dimensions - i.e. if performance is going down in particular area, but there is no errors it may be detected as something different than performance degradation caused by particular errors.

The proposed solution consists from several components which works together: 1. Dynamic profiles creation during software runtime (i.e. dynamic tracing/diagnostic level profile) (end user side)

2. Merging the profiles into one multidimensional profile (software vendor side)

3. Method for profiles comparison and providing diagnostic suggestions (software vendor side)


4. (optional) baseline profile creation (software vendor side)

Ad 1. Dynamic profiles creation during sof...