Service Usage Analytics

Spatineo Monitor provides advanced usage analytics functionality for Spatial Web Services. It consists of a set of interactive drill-down tools for getting valuable insights on who is using the given Spatial Web Services, where do those users come from and what kind of information have they been looking for. The Usage analytics tools are fully integrated into the Spatineo Monitor user interface.

The usage analytics in Spatineo Monitor is based on statistical analysis of the log files originating from the web servers providing the services (so called access logs). These log files are typically produced by web servers specifically to allow analytics of the server usage. Spatineo Monitor does not have direct access to the servers to download these files, as this kind of external data access would in most cases be against the data access policies of the data providers. Instead it has a service for uploading these log files. Monitor does statistical processing of the uploaded log data enabling the interactive analytics tools in the Usage analytics view of the Monitoring & usage section to view the data in analysed form.

In a typical case a simple script is set up by the data provider's technical personnel to automatically send the new daily server log files from the customer's computing facilities to Spatineo Monitor. The upload is done in a secure fashion using SFTP protocol and client certificates for authorisation. For more detailed information about setting up the access log upload, please refer to the separate Spatineo Monitor Log Upload Guide document.

The usage analytics views are only shown to the authorised owner organisations of the monitored services, whereas the monitoring data is visible for all Spatineo Monitor users.

Request and response time analysis: the request timeline

Like with the monitoring information on the Meters & alerts view, the graphical request timeline on the top of the Usage analytics view provides an overview of the usage data for the selected service within the selected period of time. The green bars indicate the amounts of recorded user requests per hour, and the filled line graph in the background reflects the response times as recorded by the web server in the log files. As on Monitoring timeline, you can zoom in to the most interesting time periods by either by using the buttons below the graph or dragging over the timeline. For the maximum detail on the interesting phenomena, you can select just one green bar to show only the request made during that hour.

The log-file based request times can be compared to the monitored response times by changing to the Meters & alerts tab and back. These figures will show different values for two reasons: First reason is that the monitoring is done with the defined, lightweight requests, when the usage analytics shows averaged response times of all the service requests including slower and faster ones. The second source for the difference is measurement technical: the values shown in the usage analytics are recorded by the web server when the response times in the monitoring are measured by the monitoring agent client. The Internet connection and transfer times are always included in the monitoring measurements, when in the server logs the response times are sometimes based in the data transfers in the internal network, for example between the web server and the firewall or proxy server. The changes in the selected time range are automatically reflected to both views to make comparing given events easier.

As in the Monitoring & alerts view, the vertical axis scale is logarithmic to better fit in the wide variety of different response times typically experienced in problem or high load situations.

Note: The response time can only be shown in Usage Analytics if this information is provided in the original access log files. If the response times are missing from the view, please make sure that the request time logging is enabled for your access log files.

Drilling down to usage data using the Segment table

The Segment table provides a way to drill down into details about the users and their requests within the selected time period. In this table the user requests are categorised based on the different features like the country of origin, organisation/ISP, device/browser, the requested image dimensions or the coordinate reference system of the request.

On each row the number of different IP addresses, the total amount of requests and the average response time is shown. By clicking one of the lines in the table the information shown in the rest of the analytics components of the view is filtered to include only the selected requests. The filtering can be turned off by clicking same row again.

๐Ÿ“˜

Device segment update (March 13th 2015)

Analysis done after March 13th 2015 has more precise device segment categories: Desktop (including laptops), Digital media receiver, Game console, Mobile, Tablet and Wearable.

Data analysed before this change was segmented into just two categories: Desktop and Mobile. The old Desktop segment has been divided into the new segments listed above. This old segment has been renamed to "Non-mobile" so you will be able to compare analysis results before and after this change.

Further drill-down: Request content statistics table

Below the Segment table is another table showing the selected requests by the requested content. For WMS and WMTS services these are layers provided by the service. For WFS services, the table shows requests by either feature type or the name of the stored query used by the users. For each row the total amount of requests, the total transferred data amount and the average response time is shown in the table.

The selections made in the Segment table filter the visible requests in the Request content table, and both affect the graphical analytics tools on the right side of the view. The combined time range, the Segment table and the Request content filtering provides a very powerful way to quickly focus in the most interesting requests in vast amounts of logged usage data.

Request distribution by time of the day and the day of the week

On the right side of the Segment table there are three graphical analytics tools, of which the one on the first tab shows the temporal distribution of the selected requests. The top part of the graph shows the amount of requests by the time of the day in one hour slots. This graph is useful for identifying the high and low traffic times for the service. The lower part of the graph shows similar request distribution by the day of the week.

1217

User location map view

The User location map can be found in the second tab beside the Segment table. It shows the IP-address based analysis of locations where the users' devices were located at the time of making the selected requests*. The map can be zoomed and panned to get more detailed spatial distribution of the request locations up to the city level. There are additional quick zoom buttons for the whole world and the service's bounds.

Request area map view

The most advanced usage analytics tool is the Request area map on the third tab beside the Segment table. This map shows the results the statistical spatial analysis of the target areas (bounding boxes) of selected requests. This map shows you which areas the users have been most interested in when making requests for the selected service. Like the User location map, this map can also be zoomed and panned using the provided +/- and world/bounds -buttons and dragging with mouse. As with any of the two other graphical tools, the selected requests can be further filtered by clicking any rows of the Segment and Request content tables.

*) This product includes GeoLite2 data created by MaxMind, available from http://www.maxmind.com.