Anomaly diagnosis for root cause analysis

Kusto Query Language (KQL) has built-in anomaly detection and forecasting functions to check for anomalous behavior. Once such a pattern is detected, a Root Cause Analysis (RCA) can be run to mitigate or resolve the anomaly.

The diagnosis process is complex and lengthy, and done by domain experts. The process includes:

• Fetching and joining more data from different sources for the same time frame
• Looking for changes in the distribution of values on multiple dimensions
• Charting more variables
• Other techniques based on domain knowledge and intuition

Since these diagnosis scenarios are common, machine learning plugins are available to make the diagnosis phase easier, and shorten the duration of the RCA.

All three of the following Machine Learning plugins implement clustering algorithms: autocluster, basket, and diffpatterns. The autocluster and basket plugins cluster a single record set, and the diffpatterns plugin clusters the differences between two record sets.

Clustering a single record set

A common scenario includes a dataset selected by a specific criteria such as:

• Time window that shows anomalous behavior
• High temperature device readings
• Long duration commands
• Top spending users You want a fast and easy way to find common patterns (segments) in the data. Patterns are a subset of the dataset whose records share the same values over multiple dimensions (categorical columns).

The following query builds and shows a time series of service exceptions over the period of a week, in ten-minute bins:

Run the query

let min_t = toscalar(demo_clustering1 | summarize min(PreciseTimeStamp));  
let max_t = toscalar(demo_clustering1 | summarize max(PreciseTimeStamp));  
demo_clustering1
| make-series num=count() on PreciseTimeStamp from min_t to max_t step 10m
| render timechart with(title="Service exceptions over a week, 10 minutes resolution")

The service exception count correlates with the overall service traffic. You can clearly see the daily pattern for business days, Monday to Friday. There's a rise in service exception counts at mid-day, and drops in counts during the night. Flat low counts are visible over the weekend. Exception spikes can be detected using time series anomaly detection.

The second spike in the data occurs on Tuesday afternoon. The following query is used to further diagnose and verify whether it's a sharp spike. The query redraws the chart around the spike in a higher resolution of eight hours in one-minute bins. You can then study its borders.

Run the query

let min_t=datetime(2016-08-23 11:00);
demo_clustering1
| make-series num=count() on PreciseTimeStamp from min_t to min_t+8h step 1m
| render timechart with(title="Zoom on the 2nd spike, 1 minute resolution")

You see a narrow two-minute spike from 15:00 to 15:02. In the following query, count the exceptions in this two-minute window:

Run the query

let min_peak_t=datetime(2016-08-23 15:00);
let max_peak_t=datetime(2016-08-23 15:02);
demo_clustering1
| where PreciseTimeStamp between(min_peak_t..max_peak_t)
| count

Count
972

In the following query, sample 20 exceptions out of 972:

Run the query

let min_peak_t=datetime(2016-08-23 15:00);
let max_peak_t=datetime(2016-08-23 15:02);
demo_clustering1
| where PreciseTimeStamp between(min_peak_t..max_peak_t)
| take 20

PreciseTimeStamp	Region	ScaleUnit	DeploymentId	Tracepoint	ServiceHost
2016-08-23 15:00:08.7302460	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6	100005	00000000-0000-0000-0000-000000000000
2016-08-23 15:00:09.9496584	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6	10007006	8d257da1-7a1c-44f5-9acd-f9e02ff507fd
2016-08-23 15:00:10.5911748	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6	100005	00000000-0000-0000-0000-000000000000
2016-08-23 15:00:12.2957912	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6	10007007	f855fcef-ebfe-405d-aaf8-9c5e2e43d862
2016-08-23 15:00:18.5955357	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6	10007006	9d390e07-417d-42eb-bebd-793965189a28
2016-08-23 15:00:20.7444854	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6	10007006	6e54c1c8-42d3-4e4e-8b79-9bb076ca71f1
2016-08-23 15:00:23.8694999	eus2	su2	89e2f62a73bb4efd8f545aeae40d7e51	36109	19422243-19b9-4d85-9ca6-bc961861d287
2016-08-23 15:00:26.4271786	ncus	su1	e24ef436e02b4823ac5d5b1465a9401e	36109	3271bae4-1c5b-4f73-98ef-cc117e9be914
2016-08-23 15:00:27.8958124	scus	su3	90d3d2fc7ecc430c9621ece335651a01	904498	8cf38575-fca9-48ca-bd7c-21196f6d6765
2016-08-23 15:00:32.9884969	scus	su3	90d3d2fc7ecc430c9621ece335651a01	10007007	d5c7c825-9d46-4ab7-a0c1-8e2ac1d83ddb
2016-08-23 15:00:34.5061623	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6	1002110	55a71811-5ec4-497a-a058-140fb0d611ad
2016-08-23 15:00:37.4490273	scus	su3	90d3d2fc7ecc430c9621ece335651a01	10007006	f2ee8254-173c-477d-a1de-4902150ea50d
2016-08-23 15:00:41.2431223	scus	su3	90d3d2fc7ecc430c9621ece335651a01	103200	8cf38575-fca9-48ca-bd7c-21196f6d6765
2016-08-23 15:00:47.2983975	ncus	su1	e24ef436e02b4823ac5d5b1465a9401e	423690590	00000000-0000-0000-0000-000000000000
2016-08-23 15:00:50.5932834	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6	10007006	2a41b552-aa19-4987-8cdd-410a3af016ac
2016-08-23 15:00:50.8259021	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6	1002110	0d56b8e3-470d-4213-91da-97405f8d005e
2016-08-23 15:00:53.2490731	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6	36109	55a71811-5ec4-497a-a058-140fb0d611ad
2016-08-23 15:00:57.0000946	eus2	su2	89e2f62a73bb4efd8f545aeae40d7e51	64038	cb55739e-4afe-46a3-970f-1b49d8ee7564
2016-08-23 15:00:58.2222707	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6	10007007	8215dcf6-2de0-42bd-9c90-181c70486c9c
2016-08-23 15:00:59.9382620	scus	su3	90d3d2fc7ecc430c9621ece335651a01	10007006	451e3c4c-0808-4566-a64d-84d85cf30978

Use autocluster() for single record set clustering

Even though there are less than a thousand exceptions, it's still hard to find common segments, since there are multiple values in each column. You can use the autocluster() plugin to instantly extract a short list of common segments and find the interesting clusters within the spike's two minutes, as seen in the following query:

Run the query

let min_peak_t=datetime(2016-08-23 15:00);
let max_peak_t=datetime(2016-08-23 15:02);
demo_clustering1
| where PreciseTimeStamp between(min_peak_t..max_peak_t)
| evaluate autocluster()

SegmentId	Count	Percent	Region	ScaleUnit	DeploymentId	ServiceHost
0	639	65.7407407407407	eau	su7	b5d1d4df547d4a04ac15885617edba57	e7f60c5d-4944-42b3-922a-92e98a8e7dec
1	94	9.67078189300411	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6
2	82	8.43621399176955	ncus	su1	e24ef436e02b4823ac5d5b1465a9401e
3	68	6.99588477366255	scus	su3	90d3d2fc7ecc430c9621ece335651a01
4	55	5.65843621399177	weu	su4	be1d6d7ac9574cbc9a22cb8ee20f16fc

You can see from the results above that the most dominant segment contains 65.74% of the total exception records and shares four dimensions. The next segment is much less common. It contains only 9.67% of the records, and shares three dimensions. The other segments are even less common.

Autocluster uses a proprietary algorithm for mining multiple dimensions and extracting interesting segments. "Interesting" means that each segment has significant coverage of both the records set and the features set. The segments are also diverged, meaning that each one is different from the others. One or more of these segments might be relevant for the RCA process. To minimize segment review and assessment, autocluster extracts only a small segment list.

Use basket() for single record set clustering

You can also use the basket() plugin as seen in the following query:

Run the query

let min_peak_t=datetime(2016-08-23 15:00);
let max_peak_t=datetime(2016-08-23 15:02);
demo_clustering1
| where PreciseTimeStamp between(min_peak_t..max_peak_t)
| evaluate basket()

SegmentId	Count	Percent	Region	ScaleUnit	DeploymentId	Tracepoint	ServiceHost
0	639	65.7407407407407	eau	su7	b5d1d4df547d4a04ac15885617edba57		e7f60c5d-4944-42b3-922a-92e98a8e7dec
1	642	66.0493827160494	eau	su7	b5d1d4df547d4a04ac15885617edba57
2	324	33.3333333333333	eau	su7	b5d1d4df547d4a04ac15885617edba57	0	e7f60c5d-4944-42b3-922a-92e98a8e7dec
3	315	32.4074074074074	eau	su7	b5d1d4df547d4a04ac15885617edba57	16108	e7f60c5d-4944-42b3-922a-92e98a8e7dec
4	328	33.7448559670782				0
5	94	9.67078189300411	scus	su5	9dbd1b161d5b4779a73cf19a7836ebd6
6	82	8.43621399176955	ncus	su1	e24ef436e02b4823ac5d5b1465a9401e
7	68	6.99588477366255	scus	su3	90d3d2fc7ecc430c9621ece335651a01
8	167	17.1810699588477	scus
9	55	5.65843621399177	weu	su4	be1d6d7ac9574cbc9a22cb8ee20f16fc
10	92	9.46502057613169				10007007
11	90	9.25925925925926				10007006
12	57	5.8641975308642					00000000-0000-0000-0000-000000000000

Basket implements the "Apriori" algorithm for item set mining. It extracts all segments whose coverage of the record set is above a threshold (default 5%). You can see that more segments were extracted with similar ones, such as segments 0, 1 or 2, 3.

Both plugins are powerful and easy to use. Their limitation is that they cluster a single record set in an unsupervised manner with no labels. It's unclear whether the extracted patterns characterize the selected record set, anomalous records, or the global record set.

Clustering the difference between two records sets

The diffpatterns() plugin overcomes the limitation of autocluster and basket. Diffpatterns takes two record sets and extracts the main segments that are different. One set usually contains the anomalous record set being investigated. One is analyzed by autocluster and basket. The other set contains the reference record set, the baseline.

In the following query, diffpatterns finds interesting clusters within the spike's two minutes, which are different from the clusters within the baseline. The baseline window is defined as the eight minutes before 15:00, when the spike started. You extend by a binary column (AB), and specify whether a specific record belongs to the baseline or to the anomalous set. Diffpatterns implements a supervised learning algorithm, where the two class labels were generated by the anomalous versus the baseline flag (AB).

Run the query

let min_peak_t=datetime(2016-08-23 15:00);
let max_peak_t=datetime(2016-08-23 15:02);
let min_baseline_t=datetime(2016-08-23 14:50);
let max_baseline_t=datetime(2016-08-23 14:58); // Leave a gap between the baseline and the spike to avoid the transition zone.
let splitime=(max_baseline_t+min_peak_t)/2.0;
demo_clustering1
| where (PreciseTimeStamp between(min_baseline_t..max_baseline_t)) or
        (PreciseTimeStamp between(min_peak_t..max_peak_t))
| extend AB=iff(PreciseTimeStamp > splitime, 'Anomaly', 'Baseline')
| evaluate diffpatterns(AB, 'Anomaly', 'Baseline')

SegmentId	CountA	CountB	PercentA	PercentB	PercentDiffAB	Region	ScaleUnit	DeploymentId	Tracepoint
0	639	21	65.74	1.7	64.04	eau	su7	b5d1d4df547d4a04ac15885617edba57
1	167	544	17.18	44.16	26.97	scus
2	92	356	9.47	28.9	19.43				10007007
3	90	336	9.26	27.27	18.01				10007006
4	82	318	8.44	25.81	17.38	ncus	su1	e24ef436e02b4823ac5d5b1465a9401e
5	55	252	5.66	20.45	14.8	weu	su4	be1d6d7ac9574cbc9a22cb8ee20f16fc
6	57	204	5.86	16.56	10.69

The most dominant segment is the same segment that was extracted by autocluster. Its coverage on the two-minute anomalous window is also 65.74%. However, its coverage on the eight-minute baseline window is only 1.7%. The difference is 64.04%. This difference seems to be related to the anomalous spike. To verify this assumption, the following query splits the original chart into the records that belong to this problematic segment, and records from the other segments.

Run the query

let min_t = toscalar(demo_clustering1 | summarize min(PreciseTimeStamp));  
let max_t = toscalar(demo_clustering1 | summarize max(PreciseTimeStamp));  
demo_clustering1
| extend seg = iff(Region == "eau" and ScaleUnit == "su7" and DeploymentId == "b5d1d4df547d4a04ac15885617edba57"
and ServiceHost == "e7f60c5d-4944-42b3-922a-92e98a8e7dec", "Problem", "Normal")
| make-series num=count() on PreciseTimeStamp from min_t to max_t step 10m by seg
| render timechart

This chart allows us to see that the spike on Tuesday afternoon was because of exceptions from this specific segment, discovered by using the diffpatterns plugin.

Summary

The Machine Learning plugins are helpful for many scenarios. The autocluster and basket implement an unsupervised learning algorithm and are easy to use. Diffpatterns implements a supervised learning algorithm and, although more complex, it's more powerful for extracting differentiation segments for RCA.

These plugins are used interactively in ad-hoc scenarios and in automatic near real-time monitoring services. Time series anomaly detection is followed by a diagnosis process. The process is highly optimized to meet necessary performance standards.