diff options
Diffstat (limited to 'docs/cpta/methodology')
| -rw-r--r-- | docs/cpta/methodology/index.rst | 176 |
1 files changed, 88 insertions, 88 deletions
diff --git a/docs/cpta/methodology/index.rst b/docs/cpta/methodology/index.rst index 349778999e..612f6b32db 100644 --- a/docs/cpta/methodology/index.rst +++ b/docs/cpta/methodology/index.rst @@ -60,88 +60,94 @@ Trend Analysis -------------- All measured performance trend data is treated as time-series data that -can be modelled using normal distribution. After trimming the outliers, -the median and deviations from median are used for detecting performance -change anomalies following the three-sigma rule of thumb (a.k.a. -68-95-99.7 rule). - -Metrics -```````````````` - -Following statistical metrics are used as performance trend indicators -over the rolling window of last <N> sets of historical measurement data: - -- **Q1**, **Q2**, **Q3** : **Quartiles**, three points dividing a ranked - data set of <N> values into four equal parts, Q2 is the median of the - data. -- **IQR** = Q3 - Q1 : **Inter Quartile Range**, measure of variability, - used here to calculate and eliminate outliers. -- **Outliers** : extreme values that are at least (1.5 * IQR) below Q1. - - - Note: extreme values that are at least (1.5 * IQR) above Q3 are not - considered outliers, and are likely to be classified as - progressions. - -- **TMA** : **Trimmed Moving Average**, average across the data set of - <N> values without the outliers. Used here to calculate TMSD. -- **TMSD** : **Trimmed Moving Standard Deviation**, standard deviation - over the data set of <N> values without the outliers, requires - calculating TMA. Used for anomaly detection. -- **TMM** : **Trimmed Moving Median**, median across the data set of <N> - values excluding the outliers. Used as a trending value and as a - reference for anomaly detection. - -Outlier Detection -````````````````` - -Outlier evaluation of test result of value *X* follows the definition -from previous section: - -+----------------------------+----------------------+ -| Outlier Evaluation Formula | Evaluation Result | -+============================+======================+ -| X < (Q1 - 1.5 * IQR) | Outlier | -+----------------------------+----------------------+ -| X >= (Q1 - 1.5 * IQR) | Valid (For Trending) | -+----------------------------+----------------------+ +can be modelled as concatenation of groups, each group modelled +using normal distribution. While sometimes the samples within a group +are far from being distributed normally, we do not have a better tractable model. + +The group boundaries are selected based on `Minimum Description Length`_. + +Minimum Description Length +-------------------------- + +`Minimum Description Length`_ (MDL) is a particular formalization +of `Occam's razor`_ principle. + +The general formulation mandates to evaluate a large set of models, +but for anomaly detection purposes, it is usefuls to consider +a smaller set of models, so that scoring and comparing them is easier. + +For each candidate model, the data should be compressed losslessly, +which includes model definitions, encoded model parameters, +and the raw data encoded based on probabilities computed by the model. +The model resulting in shortest compressed message is the "the" correct model. + +For our model set (groups of normally distributed samples), +we need to encode group length (which penalizes too many groups), +group average (more on that later), group stdev and then all the samples. + +Luckily, the "all the samples" part turns out to be quite easy to compute. +If sample values are considered as coordinates in (multi-dimensional) +Euclidean space, fixing stdev means the point with allowed coordinates +lays on a sphere. Fixing average intersects the sphere with a (hyper)-plane, +and Gaussian probability density on the resulting sphere is constant. +So the only contribution is the "area" of the sphere, which only depends +on the number of samples and stdev. + +A somehow ambiguous part is in choosing which encoding +is used for group size, average and stdev. +Diferent encodings cause different biases to large or small values. +In our implementation we have chosen probability density +corresponding to uniform distribution (from zero to maximal sample value) +for stdev and average of the first group, +but for averages of subsequent groups we have chosen a distribution +which disourages deliminating groups with averages close together. + +One part of our implementation which is not precise enough +is handling of measurement precision. +The minimal difference in MRR values is currently 0.1 pps +(the difference of one packet over 10 second trial), +but the code assumes the precision is 1.0. +Also, all the calculations assume 1.0 is totally negligible, +compared to stdev value. + +The group selection algorithm currently has no parameters, +all the aforementioned encodings and handling of precision is hardcoded. +In principle, every group selection is examined, and the one encodable +with least amount of bits is selected. +As the bit amount for a selection is just sum of bits for every group, +finding the best selection takes number of comparisons +quadratically increasing with the size of data, +the overall time complexity being probably cubic. + +The resulting group distribution looks good +if samples are distributed normally enough within a group. +But for obviously different distributions (for example `bimodal distribution`_) +the groups tend to focus on less relevant factors (such as "outlier" density). Anomaly Detection ````````````````` -To verify compliance of test result of valid value <X> against defined -trend metrics and detect anomalies, three simple evaluation formulas are -used: - -+-------------------------------------------+-----------------------------+-------------------+ -| Anomaly Evaluation Formula | Compliance Confidence Level | Evaluation Result | -+===========================================+=============================+===================+ -| (TMM - 3 * TMSD) <= X <= (TMM + 3 * TMSD) | 99.73% | Normal | -+-------------------------------------------+-----------------------------+-------------------+ -| X < (TMM - 3 * TMSD) | Anomaly | Regression | -+-------------------------------------------+-----------------------------+-------------------+ -| X > (TMM + 3 * TMSD) | Anomaly | Progression | -+-------------------------------------------+-----------------------------+-------------------+ - -TMM is used for the central trend reference point instead of TMA as it -is more robust to anomalies. +Once the trend data is divided into groups, each group has its population average. +The start of the following group is marked as a regression (or progression) +if the new group's average is lower (higher) then the previous group's. Trend Compliance ```````````````` Trend compliance metrics are targeted to provide an indication of trend changes over a short-term (i.e. weekly) and a long-term (i.e. -quarterly), comparing the last trend value, TMM[last], to one from week -ago, TMM[last - 1week] and to the maximum of trend values over last -quarter except last week, max(TMM[(last - 3mths)..(last - 1week)]), +quarterly), comparing the last group average AVG[last], to the one from week +ago, AVG[last - 1week] and to the maximum of trend values over last +quarter except last week, max(AVG[last - 3mths]..ANV[last - 1week]), respectively. This results in following trend compliance calculations: -+-------------------------+---------------------------------+-----------+------------------------------------------+ -| Trend Compliance Metric | Trend Change Formula | Value | Reference | -+=========================+=================================+===========+==========================================+ -| Short-Term Change | (Value - Reference) / Reference | TMM[last] | TMM[last - 1week] | -+-------------------------+---------------------------------+-----------+------------------------------------------+ -| Long-Term Change | (Value - Reference) / Reference | TMM[last] | max(TMM[(last - 3mths)..(last - 1week)]) | -+-------------------------+---------------------------------+-----------+------------------------------------------+ ++-------------------------+---------------------------------+-----------+-------------------------------------------+ +| Trend Compliance Metric | Trend Change Formula | Value | Reference | ++=========================+=================================+===========+===========================================+ +| Short-Term Change | (Value - Reference) / Reference | AVG[last] | AVG[last - 1week] | ++-------------------------+---------------------------------+-----------+-------------------------------------------+ +| Long-Term Change | (Value - Reference) / Reference | AVG[last] | max(AVG[last - 3mths]..AVG[last - 1week]) | ++-------------------------+---------------------------------+-----------+-------------------------------------------+ Trend Presentation ------------------ @@ -160,17 +166,18 @@ Trendline Graphs ```````````````` Trendline graphs show per test case measured MRR throughput values with -associated trendlines. The graphs are constructed as follows: +associated gruop averages. The graphs are constructed as follows: - X-axis represents performance trend job build Id (csit-vpp-perf-mrr- daily-master-build). - Y-axis represents MRR throughput in Mpps. - Markers to indicate anomaly classification: - - Outlier - gray circle around MRR value point. - Regression - red circle. - Progression - green circle. +- The line shows average of each group. + In addition the graphs show dynamic labels while hovering over graph data points, representing (trend job build Id, MRR value) and the actual vpp build number (b<XXX>) tested. @@ -217,26 +224,15 @@ PA is defined as follows: archived data. b) Parse out the data filtering test cases listed in PA specification (part of CSIT PAL specification file). - c) Evalute new data from latest PT job against the rolling window of - <N> sets of historical data for trendline calculation, anomaly - detection and short-term trend compliance. And against long-term - trendline metrics for long-term trend compliance. -3. Calculate trend metrics for the rolling window of <N> sets of - historical data: +3. Re-calculate new groups and their averages. - a) Calculate quartiles Q1, Q2, Q3. - b) Trim outliers using IQR. - c) Calculate TMA and TMSD. - d) Calculate normal trending range per test case based on TMM and - TMSD. +4. Evaluate new test data: -4. Evaluate new test data against trend metrics: - - a) If within the range of (TMA +/- 3*TMSD) => Result = Pass, + a) If the existing group is prolonged => Result = Pass, Reason = Normal. (to be updated base on the final Jenkins code). - b) If below the range => Result = Fail, Reason = Regression. - c) If above the range => Result = Pass, Reason = Progression. + b) If a new group is detected with lower average => Result = Fail, Reason = Regression. + c) If a new group is detected with higher average => Result = Pass, Reason = Progression. 5. Generate and publish results @@ -251,3 +247,7 @@ Testbed HW configuration The testbed HW configuration is described on `this FD.IO wiki page <https://wiki.fd.io/view/CSIT/CSIT_LF_testbed#FD.IO_CSIT_testbed_-_Server_HW_Configuration>`_. + +.. _Minimum Description Length: https://en.wikipedia.org/wiki/Minimum_description_length +.. _Occam's razor: https://en.wikipedia.org/wiki/Occam%27s_razor +.. _bimodal distribution: https://en.wikipedia.org/wiki/Bimodal_distribution |