1. Service Widget

1.1. Purpose

The main purpose of the widget is to collect statistics on the usage and the usability of a service.

Available services are ones created with the Finder on the Engine. Please consult the dedicated page in Finder documentation. See Services in Finder documentation

The widget will record the full history and allow full access of data in the past; It will be possible to drill-down to sources having been impacted by a disfunction in the service.

1.2. Collected Statistics

Service usage

• Sources Total number of sources having used the services
• Incoming traffic Volume of data in Byte incoming to sources from the service
• Outgoing traffic Volume of data in Byte outgoing from sources to the service
• Network availability Percentage of successful TCP connections
• Response time Average TCP connection establishment time
• Application crashes Number of application having acceded the service and experimented a crash the same day

Service usability

• Based on the usage statistics, the widget will detect the sources that have been impacted from poorly results
• Poor network availability
  • Number of sources affected by a poor availability of the network
• High response time
  • Number of sources that got a bad response time when using the service
• High number of application crashes
  • Number of sources for whom application linked to the service had crashed

1.3. Creating widget

To create a Service widget, please follow the following steps:

1. choose the Service that you want to monitor

• A select box will display the list all services defined on Engines.

2. define the criterion that will characterize an impacted source

• The minimal ratio of network availability that can be considered as acceptable
  • By default a source will be considered as impacted if it encounters less than 90% of network availability
• The acceptable network response time in relation to global average response time
  • By default a source experimenting a response time up to twice the average response time of other sources will be counted as impacted
• The maximum number of application crashes
  • By default, any application crash will turn the sources as impacted. However, if one specify a number, for example 2, it will be interpreted as the following:
  • A source will be impacted for a day if it experiment an application crash on that day, and that the number of application crashes on the past 7 days is upper than 2.

3. define gravity threshold on the number of impacted sources

• It is possible to define two gravity levels, symbolized by yellow and red bullets For each one, enter the ratio of impacted sources on total number of sources above which the threshold will be reached

4. optionally select a criterion for grouping results:

• "group by" - it can be any category on source object, results are classified by keywords
• "hierarchy" - using a hierarchy permits to structure the result as defined in the hierarchy but also to apply access rights set to users (available only if at least one hierarchy has been defined)

  
  

1.4. How the calculation is performed

1.4.1. Retrieving the statistics from Engine

• Gathering of service usage for each source
  • The Engine will be queried to get for each source the statistics of the day. Collected results will be aggregated according to grouping options
• Computing the number of application crashes
  • An application crash for the service usage is triggered when an application experimenting a crash has been in the same day linked to the service by using it.

1.4.2. Case of computing on many Engines

When computing on many Engines, the targeted service may not be defined on all Engine and can also have different definitions between Engine.

• Computation will be done only on Engines for whom the service exist
• The service definition of each Engine will be used to compute on that Engine

1.4.3. Computing the value along the time

The values computed each day for the widget will be aggregated over 4 time periods

• Day
• Week, the starting day of the week being defined by Portal configuration with Monday as a default
• Month
• Quarter, with first quarter beginning at January the 1st

For each of those periods, the operation given by widget configuration will be applied to aggregate the results: For example, the total number of sources using the service for a week will be the sum of each day value from Monday to Sunday.

The aggregation on time will be done according to grouping options, so that the full detail will be accessible in the history.

1.5. Using widget

1.5.1. Overall value

The number of impacted sources is displayed as an overall value, with a bullet indicator to situate the value through the gravity levels that have been defined.

The last change is also displayed with an arrow showing the last trend.

1.5.2. Overall chart

An overall chart of collected data is displayed. The data can be represented by period:

• By day, values for the last 60 days are displayed
• By week, values for the last year are displayed
• By month and quarter, values for the last 2 years are displayed

The chart is interactive so that the dates and values will be displayed by moving the mouse around the chart. By clicking on the chart, the widget will navigate to the selected date and display the result for that date.

When displaying result for week, month or quarter, as the data of the last day may not be complete for the period, a dotted line will symbolize the fact that last plot is an on going value.

• 
  

1.5.3. Details of displayed statistics

• For each statistic, the value is displayed and a sparkline shows the trend over the past days.
• An indicator will show the last change of each statistic with an arrow giving the meaning of that change (for example a positive move on the number of application crashes will be symbolized by a red arrow, but it will be a green one if the augmentation is about Network availability)
• If a "group by" category has been defined, each line can be expanded to see the values per group (e.g., per location). One can also select one particular group (at the bottom of the widget) and then only the values for that group is displayed for all lines.
• If a "group by" on hierarchies has been defined, each factor will display the current hierarchy node on a first level. It can be expanded to see the values of immediate child nodes. Those child nodes can be clicked to drill-down on their level, the entire widget will then be redrawn for the new level.
  • A navigation bar will appears at the top of widget to indicate the position in the hierarchy If multiple hierarchies are defined, a select box on top will allow to switch between them

  
  

1.5.3.1. Impacted sources

• For each factor leading to a source impact, the number of sources experiencing the issue is displayed. A green/yellow/red bullet is also displayed to show at a glance the level of gravity.
  • One can click on the value to access to the list of sources
• An overall entry is displayed at the top to summarize the total number of impacted sources

1.5.3.2. Usage

• A line will be created for each statistic, with value, trend chart and last change information

1.5.4. Displayed time frame

• The widget interface have a select box to choose the period to be displayed (day, week, month or quarter)
  • All displayed information will be consistent with the selected period, with the chart showing the aggregated values for each period end (each Sunday, end of month or end of quarter)
  • The evolution will be computed between the last displayed value and the preceding end of period
• Another select box will permit to choose a specific date. Be aware that the time frame covered by the chart will always end at the last computed date, and so will not vary with date selection.

1.5.5. Displaying ratios

• A select box at the top of the widget allow to display the number of impacted sources as a ratio of the total number of sources
• When displaying as ratio, the charts and last changes will also be relative to the total number of sources

•