Figure A-4 Severity Indicating Gauge threshold examples

<font color="#e0121d"><b>Overview of OSI Systems Management Metric Objects </b></font><br/>ISO/IEC 10164-11: 1994 | ITU-T Rec X.739 INFORMATION TECHNOLOGY – OPEN SYSTEMS INTERCONNECTION – SYSTEMS MANAGEMENT: METRIC OBJECTS AND ATTRIBUTES <a href="file://http://www.itu.int/rec/T-REC-X.739-199311-I/en"><font color="#0000ff"><u>http://www.itu.int/rec/T-REC-X.739-199311-I/en</u></font></a><br/><font color="#29313b">The OSI-Management Metric Objects specification, unlike the OMG SMM, can work without an underlying metamodel, and it specifies run time objects which perform statistical calculations on a single attribute of an observed managed object over time, using a periodically invoked get operation. The concepts and mechanisms of this specification are applicable to any Cloud Computing environment which has capabilities for getting values of attributes of Cloud Computing Resources.  Any REST based protocol could be used by a Metric Object to scan the values of the observed resources.</font><br/><font color="#29313b">The underlying targets for the Metric objects are instance of Managed Objects in a running system. OSI-System management has the CMIP protocol which conveys Create, Replace, Update, and Delete operations on an agent process which reflects these management operations in a real world set of Managed Elements, represented as managed objects.   CMIP has a simple get operation to get the current value of any attribute of a managed object instance, identified by a unique distinguished name.</font><br/><font color="#29313b">Metric Objects specification was designed to define metrics for statistical calculations over time, employing the concept of a scanner.</font><br/><font color="#29313b">The data necessary to compute the observed values are extracted from the observed managed objects. The observed values are monitored by sampling at an interval specified by a granularity period.</font><br/><font color="#29313b">A Monitor Metric is a scanner that periodically gets the value of an observed attribute of a single Managed Object instance that it points at, with attributes for a managed object instance pointer, the measured attribute name, and a granularity period.</font><br/><font color="#29313b">The Metrics objects model assumes that observed attribute being scanned is either a counter (monotonically increasing integer value) or a gauge (value of a varying quantity).  There are mechanisms in the metric objects to derive a gauge from an attribute that is a counter (counter difference package), as well as well as to derive a derivative of a gauge from an observed attribute that is a gauge. </font><br/><font color="#29313b">There are also optional scanner mechanisms to specify different scheduling and control capabilities, for the times when scanning of the observed attribute are active.</font><br/><font color="#29313b">Metric objects can also have a severity indicating gauge-threshold attribute type which is used to trigger notifications for monitoring resources.  There can be multiple threshold levels (each with an associated clear threshold) applied to the derived value in a metric object.</font><br/><font color="#29313b">Figure A-4 shows an example of a severity indicating gauge model.</font><br/>