Health trend analysis method on utilization of network resources

• US 7,254,750 B1
• Filed: 03/30/2004
• Issued: 08/07/2007
• Est. Priority Date: 03/30/2004
• Status: Active Grant

• Current Assignee: Unisys Corporation
• Original Assignee: Unisys Corporation
• Inventors: Okun, Justin Azriel, Maw, Patricia Lynn
• Primary Examiner(s): Baderman; Scott
• Assistant Examiner(s): BONURA, TIMOTHY M
• Application Number:
  US10/812,688
• Time in Patent Office: 1,225 Days
• Field of Search: 714/47, 702/179, 702/180
• US Class Current: 714/47.2
• CPC Class Codes: G06F 11/3409 for performance assessment
• CPC Class Codes: G06F 11/3452 Performance evaluation by statistical analysis
• CPC Class Codes: G06F 2201/86 Event-based monitoring
• CPC Class Codes: G06F 2201/88 Monitoring involving counting

Abstract

An embodiment of the present invention is a method for monitoring various data sources for events indicative of a current or potential system health problem, thereby detecting positive or negative trends in system health data. At the core of this module is an algorithm that analyzes a series of data points that have been collected into sample sets over time.

Claims

1. In a network of multi-processors having a series of local systems operated by Client-Users wherein a series of defined operating policies (P) are stored along with acceptable parameter limits for each policy, a method for developing a health trend analysis of future possible problems in network resources comprising the steps of:

(a) accessing a health counter value for the allowable upper boundary “

U”

for a selected operating policy (P);

(b) accessing a health counter value for the allowable lower boundary “

L”

for said selected operating policy (P);

(c) establishing, via a trend counter, a trend of upward or downward increases or decreases in the availability of system resources which includes the steps of;

(c1) collecting data points, in a sample set “

N”

where “

N”

is the number of collected data points;

(c2) verifying that the number of collected data points is=“

N”

,

(c3) calculating the average percent difference “

A”

which represents the cumulative difference between each data point and the one just before it for all points in the set;

(c4) multiplying the average difference “

A”

by the number “

N”

of data points to get a result A*N;

(c5) calculating a common sense value “

C”

by finding the percent difference between the first value and the last value of the sample set;

(c6) establishing whether the product A*N is greater than said lower boundary “

L” and

less than said upper boundary “

U”

;

(c7) preparing a summary of step (c6);

(d) acquiring a series of data points associated with said selected policy (P);

(e) setting a counter limit “

TCL”

(Trend Counter Limit) to a value specified for the selected policy (P).

2. The method of claim 1 wherein step (c7) includes the steps of:

(8c7a) determining that the value of A*N falls between “

L” and

“

U”

(YES);

(8c7b) incrementing a counter designated as “

not-a-trend”

counter;

(8c7c) querying whether said “

not-a-trend”

counter value is greater than the value of TCL (Trend Counter Limit);

(8c7d) if “

not-a-trend”

counter is greater than the value of TCL (YES), then deleting all collected samples;

(8c7e) resuming operations at step (c1) of claim 1.

3. The method of claim 1 wherein step (c7) includes the steps of:

(9c7a) determining that the A*N value does not fall between “

L” and

“

U”

;

(9c7b) querying whether the trend type is upward as pre-ordained in health policy specifications;

(9c7c) if the trend is upward, then querying whether A*N value is greater than the upper boundary “

U”

; and

if so;

(9c7d) determining whether the common sense value “

C”

is greater than the upper boundary value “

U”

; and

if so,

(9c7e) calculating the size of the increase (Delta I) which indicates the average value of the sample set multiplied by the number of data points in the set multiplied by the first value in the set;

(9c7f) incrementing a trend counter and zeroing out said “

not-a-trend”

counter;

(9c7g) checking to see if said trend counter value is greater than the trend counter limit value (TCL); and

if so,

(9c7h) reporting an upward trend to said Client-User;

(9c7i) additionally smoothing out the data collected by using reported values below the lower boundary “

L”

to offset reported values above the upper boundary “

U”

to eliminate spikes and false alerts.

4. The method of claim 1 wherein step (c7) includes the steps of:

(10c7a) determining that the A*N value does not fall between “

L” and

“

U”

;

(10c7b) querying whether the trend type is downward as specified in health policy specifications;

(10c7c) if the trend is downward, then querying whether A*N value is less than the lower boundary “

L”

; and

if so,

(10c7d) determining whether the common sense value “

C”

is less than the lower boundary value “

L”

, and if so,

(10c7e) calculating the size of the decrease (Delta D) which indicates the absolute average value of the sample set multiplied by the number of data points in the set multiplied by the first value in the set;

(10c7f) incrementing a trend counter and zeroing out said “

not-a-trend”

counter;

(10c7g) checking to see if said trend counter value is greater than the trend counter limit value (TCL); and

if so,

(10c7h) reporting a downward trend to said Client-User;

(10c7i) additionally smoothing out the data collected by using reported values above the upper boundary “

U”

to offset reported values below the lower boundary value “

L”

to eliminate spikes and false alerts.

5. In a multi-processor network holding multiple numbers of “

local systems”

which monitor themselves to create a collection of health events and predictive events and which include operating policies (P) to be effectuated a system for sensing future trends which predict future problems which may occur in system resources comprising;

(a) means to initialize an algorithm for a specified health trend policy;

(b) means to collect a sample set of data points using a counter at X points in a time interval sampling period;

(c) means for calculating a value which represents the general increase or decrease in the allocation-utilization of a monitored resource;

(d) means for calculating the average “

A”

percent difference of the values in the current sample set;

(e) means to determine that the value “

A”

falls between a low boundary value “

L” and

an upper boundary value “

U”

for the selected policy involved;

(f) means to utilize a trend counter limit (TCL) to determine if the number of collected sample sets exceeds the TCL value, and, if so;

(g) means to indicate that the value of the monitored health index for that resource is in a “

steady state” and

there is no trend involved.

6. The system of claim 5 wherein said means for calculating (d) includes:

(13d1) means to recognize that said value “

A”

indicates that the value “

A”

is outside the upper “

U” and

lower “

L”

boundary set for the selected policy (P) and said policy is monitoring for an upward trend;

(13d2) means to recognize that said value “

A”

is above the upper boundary value “

U” and

also a common sense value indicates the first data point is below the last data point (upward trend).

7. The system of claim 5 wherein said means for calculating (d) includes:

(14d1) means for determining if the average percent difference “

A”

is equal to or less than both the upper boundary “

U” and

the lower boundary “

L”

for the selected policy (P); and

, if so;

(14d2) means for calculating the downward movement (Delta D) by multiplying “

A”

by the first value in the sample set to get an estimated data point;

(14d3) means for indicating a downward trend.

8. The system of claim 7 wherein each means (14d3) for indicating a trend includes:

(14d3a) means for smoothing out data points in said sample set which involve inadvertent spikes in value.