Version 10.0
User Manual
This work was sponsored by NSF under grant numbers ACI 1025159, ACI 1445806 and 2137603.
1 Browser Requirements
XDMoD supports all major desktop web browsers (with JavaScript enabled) including:
• Google Chrome 61+
• Microsoft Edge 16+
• Firefox 52+
• Safari 9+
User’s Guide: Substantial work has gone into making the XDMoD user interface easy to use. However, given its powerful analytical capabilities, it can nonetheless be challenging for users to fully exploit its capability. To help remedy this, a User’s Guide and FAQ, available through the Help Button on the top right hand side of the XDMoD Portal window have been developed. The user guide provides detailed instructions for utilizing all aspects of XDMoD from obtaining an account to explaining common user interface elements and getting the most out of powerful tools such as the Metric Explorer. Access to the guide is context sensitive providing direct access to the most relevant portions of the guide based on what the user was viewing when they requested the help. For example, if one accesses the user guide while viewing the Metric Explorer tab they will be brought directly to the section of the manual describing that tab. In addition to the user manual, a “quick start” guide is provided directly within XDMoD for several tools including the Metric Explorer, Application Kernel Explorer, Report Generator and Job Viewer tab. As shown in Figure 1-1, when a user views the Metric Explorer tab and has not created a query, visual assistance is provided to guide them in quickly getting started.
Figure 1-1: XDMoD Help. The XDMoD Quick Start Guide for the Metric Explorer Tab |
2 Introduction
XD Metrics on Demand (XDMoD) provides metrics pertaining to resource utilization and performance of high performance computing (HPC) resources, and the impact these resources have in terms of scholarship and research. While initially focused on the National Science Foundation (NSF) TeraGrid and follow-on XSEDE (XD) and ACCESS programs, XDMoD has a wide applicability to any HPC system. The goals of XDMoD framework are to
1. Provide the end-user community with a tool to more effectively and efficiently use their allocations and optimize their use of cyber-infrastructure (CI) resources.
2. Provide operational staff with the ability to monitor and tune the performance of hardware , system software, and applications to ensure optimal resource performance.
3. Provide management with a diagnostic tool to facilitate CI planning and analysis as well as monitor resource utilization and performance.
4. Provide metrics such as publications, awards, and citations to help ensure that the resources are effectively enabling research and scholarship.
The framework also provides a computationally lightweight and flexible application kernel auditing system that reflects best-in-class performance kernels to measure overall system performance with respect to existing applications that are being run by users. This allows continuous resource usage analysis and measurement of all aspects of system performance, including: global file-system performance, local processor-memory bandwidth, allocable shared memory, processing speed, and network latency and bandwidth.
2.1 XDMoD User Interface Overview
Here we provide a brief introduction to the XDMoD user interface. Greater detail is provided throughout the manual. The XDMoD user interface, shown in Figure 2-1, is organized into multiple tabs that provide functionality tied to the role of the user.
XDMoD utilizes user roles to restrict access to data and elements of the user interface such as tabs. Supported roles include a public (unauthenticated user), User, Principal Investigator, and Center Director. A detailed description of each role as it applies to data access is as follows.
Public Role: With no user account required, the public role provides non-authenticated users with access to overall utilization broken down by center or service provider (if more than one are available), resource, field of science, allocation time, etc.; and the ability to view specific time periods, and export images. Quality of service data via the Application Kernel suite is not publically available, nor is specific user data.
User: Users are able to view all data available to the Public Role as well as their personal utilization information. They are also able to view information regarding their allocations, quality of service data via the Application Kernel Explorer, and generate custom reports.
Principal Investigator (PI): A principal investigator is a user who is listed as a PI on one or more allocation or project. A PI has access to all data available to a user, as well as detailed information for any users included on their allocations or project.
Center Director/Center Staff: The director of a service provider or center will have access to detailed usage information for any user that has run jobs at their center. In addition, they will have access to detailed application kernel results for runs at their center. Directors are also able to delegate access to their center’s data to other users who assist in the operations of their center. Directors may also have access to other information for their center to ensure that the proper information is being consistently reported.
The XDMoD User Interface contains a wealth of information and has been organized into tabs to compartmentalize the data without overwhelming the user. For illustrative purposes, we will focus on the highest level role. The tabs which are described in greater detail below are: Summary tab, Usage tab, Metric Explorer tab, App Kernels tab, Job Viewer tab, and Report Generator tab
The Summary tab (Figure 2-1) provides a snapshot overview of selected data with several small summary charts visible that can be expanded to full size charts through a simple mouse click. The default is to show utilization over the previous month, but the user may select from a number of preset date ranges (week, month, quarter, year to date, etc) or choose a custom date range. The user can also customize the summary by adding charts , see the section on the Metric Explorer.
Figure 2-1: XDMoD Summary Tab. Summary tab contains a series of user configurable summary plots of usage. |
The Usage tab, shown in Figure 2-2, provides access to an expansive set of resource-wide metrics that are accessible through the tree-structure on the left-hand side of the portal window, including summaries of usage, allocations, accounts, and performance data, as well as allocations and accounts. Usage metrics provided by XDMoD include: number of jobs, total and average CPU-hours charged, number of CPUs used, wait time, wall time, minimum, maximum and average job size, average CPU used, average wall time, average wait time and user expansion factor. If installed, a suite of SUPReMM performance metrics may also be available. These metrics can be broken down by: job size, job wall time, person, principal investigator, by resource, and by organizational hierarchy (if available). Many of the plots are context sensitive and allow users to click on a data element within the plot to further analyze the data. For example, in Figure 2-2, which shows the distribution of total CPU hours by job size in 2012, one can click on any of the columns to obtain a more detailed analysis for the selected job size range. The plot can also be made available to the custom report generator by clicking the box that reads “Available For Report”. It can also be exported in either PNG (portable network graphics), PDF (portable document format) or SVG (scalable vector graphics) format. The data itself can be exported in either CSV (comma separated values) or XML (extensible markup language) format.
Figure 2-3: XDMoD Usage Tab. A Usage plot showing the total CPU hours broken out per job on all available resources. |
The Metric Explorer tab provides a powerful tool for organizing and comparing the data from a wide variety of metrics. The Metric Explorer tab, which also provides access to all of the metrics available through the Usage tab, facilitates comparison among the various metrics by allowing multi-axis plots, as shown in Figure 2-3. Displayed in the window is a plot that shows the number of jobs (left hand axis) and average core count (right hand axis) for all available resources. As shown in Figure 2-4, the data can be filtered in a variety of ways to display only a desired subset of data. For example, the plot shown in Figure 2-4 was generated from Figure 2-3 by applying a filter to display only the resource: chemistry. A notable feature is the ability to open a metric that is provided on the Usage tab directly in the Metric Explorer by clicking on the gear icon on the top right of the plot. This allows one to utilize an existing plot as a starting point and easily customize it, configure additional data series for comparison, and save it for use in a report. Taken in its entirety, the Metric Explorer provides a powerful and flexible interface to facilitate analysis of the data.
Figure 2-6: The XDMoD Metric Explorer Tab. Plot shows the number of jobs running vs Max core count on all available resources. Number of jobs running is shown on the primary axis (left-hand axis) and Max core count is shown on the secondary axis (right-hand axis). |
Figure 2-7: The XDMoD Metric Explorer Tab. A Metric Explorer generated plot showing Number of Jobs Running for vs Max core count on. Number of Jobs Running is shown on the primary axis (left-hand axis) and the max core count is shown on the secondary axis (right-hand axis). Plot was generated from Figure 2-3 by applying a filter that limited the analysis to the resource of chemistry. |
The App Kernels tab contains three sub tabs that provide information on the application kernel performance and quality of service for resources. Through this tab, users can view historical performance for all application kernels run on all resources. For example, Figure 2-8 shows the wallclock time for Enzo benchmark run on Trestles in January 2014. Note that the plot window contains a description pane that provides information on the application kernel. The data generated by the application kernels is substantial, making the exploration of the data challenging. Therefore, in order to facilitate analysis of the application kernel performance data, we developed the App Kernel Explorer subtab. Here the user can easily select a specific application kernel or suite of application kernels, a specific set of resources, and a range of job sizes for which to view performance. It allows users to directly compare application kernel performance across multiple resources.
Figure 2-8: App Kernels Tab. Plot showing wallclock time for Enzo benchmark on Trestles for June 2014. The description pane in the plot window provides more detailed information on the application kernel and the resource. |
The Report Generator tab (Figure 2-9) gives the user access to the Custom Report Builder that allows a user to create and save custom reports. For example, a user may wish to have specific plots and data summarized in a concise report that they can download for offline viewing. The user can also choose to have custom reports generated at a user specified interval (daily, weekly, quarterly, etc) and automatically sent to them via email at the specified time interval, without the need to log into the portal.
Figure 2-9: XDMoD Custom Report Generator tab. A user can create custom reports by selecting plots from the pool of plots (right hand side) that were generated previously using the XDMoD portal. |
The Job Viewer Tab shown in figure 2-10, provides the user with the capability to search for and view specific jobs or jobs that meet specified criteria. The Job Viewer displays job accounting and performance data for any job for which this information is available in the XDMoD data warehouse. There are two basic ways to search for and view jobs using the Job Viewer. If the local job id and resource are known, the quick job lookup function can be used to locate the detailed data for the job. If a job or jobs fitting a given set of criteria are desired the Advanced search function can be used to locate all jobs fitting the specified criteria.
Figure 2-10 Job Viewer Tab |
3 Obtaining an Account
Open XDMoD provides a public view that allows users without an account to view resource utilization summaries. An account is not required to view public data in Open XDMoD, however, an account is required to access many tabs including the Metric Explorer, Custom Report Generator, Job Viewer and Application Kernels.
Contact the administrator of your Open XDMoD instance for specific details about how to obtain an account.
4 Getting Started
4.1 Logging In
Open XDMoD supports two different authentication mechanisms: local login and Single Sign On. The particular mechanism is configured by the Open XDMoD site administrator. To access the login panel click the “Sign In” link on the top right of the Open XDMoD public page. An example of a login panel for an Open XDMoD instance that authenticates with Single Sign On is shown in Figure 4-1. The specific text and graphics will be different for different authentication providers. An example of the login panel for an Open XDMoD instance configured to use local login is shown in Figure 4-2.
Figure 4-1: Example Login dialog for an Open XDMoD instance configured to use Single Sign On. This example shows an Open XDMoD instance configured to use the authentication provider at the University at Buffalo. Clicking the button with the UB logo opens the official UB login page where users can use their university-wide login credentials to authenticate.
Figure 4-2: Example Login dialog for an Open XDMoD instance configured to use local XDMoD accounts.
4.2 Top Toolbar and Tabs
XDMoD content is organized into a set of tabs as shown in Figure 4-2 below and described in section 2. Common functions available to all user roles (such as My Profile, Contact, and Help) are available via buttons on a common toolbar at the upper right region of the page (Figure 4-2).
Figure 4-2: XDMoD Portal tabs and common toolbar. For description see section 2.0
5 Common User Interface Elements
Here we describe some common XDMoD user interface elements.
5.1 Duration Selection Toolbar
Figure 5-1: Duration Selection Toolbar |
The duration selection toolbar (Figure 5-1) is used for selecting the range of data displayed in a particular data set or chart. It consists of a drop down list containing commonly accessed time frames, as well as a pair of date choosers for configuring the user defined time frames start and end dates, followed by a refresh button for reloading the data and in turn refreshing the screen.
When you interact with the Start and End date fields, the Refresh button will automatically highlight to indicate that it needs to be pressed to apply the new timeframe. Pressing ENTER in either the Start or End date field will accomplish the same thing. |
Figure 5-2: Preset time frames menu The user can choose from a series of commonly selected time frames including 7, 30, and 90 days from the present or 1, 2, 3, 5, and 10 years from the present or a number of calendar years. Alternatively, a user can select their own custom date range for which to display data. The Aggregation Unit selector is located at the bottom of the Duration preset menu. The aggregation unit determines the bin granularity for the time series data. XDMoD automatically chooses a default value dependent on the date range selected. However, the user can override this default value using the aggregation unit selector. Note that for very long time date durations, greater than 2 years, the day aggregation level is not supported. |
Figure 5-3: Custom time frames - interactive date chooser The user may choose to click on the button, located to the right of the start and end date choosers, to get the interactive date chooser, shown on the left, for selecting a date value interactively via the mouse, keyboard and touch input. In order for the date from the interactive date chooser to become the effective selected date of the date chooser, the user must click a day in the calendar or click the `Today` button. This will hide the interactive date chooser and replace the value in the corresponding date chooser. If the interactive date chooser goes away due to the user clicking another part of the screen, the value of the date chooser will not change. |
5.2 Chart Configuration Toolbar
Figure 5-4: Chart Configuration Toolbar |
The chart configuration toolbar (Figure 5-4) provides options for a user to customize the appearance of individual charts (and their datasheet view), export the data in a number of output formats, and make a chart available to the Report Generator.
Figure 5-5: Dataset Filter |
When the user clicks on the filter button, the chart configuration toolbar is presented with a multi-selector that provides a list of available data sets to the left and a set of selected data sets to the right, for example, Figure 5-5 which displays the filter by resource multi-selector. The user can then select the desired filter choices by clicking on the check boxes. The search box and the paging mechanisms shown in Figure 5-5 are available to assist the user in sorting through filters having large numbers of choices. Note that for the Usage tab that the user must drill down first before the filter becomes active. The manual sections on the Usage tab and the Metric Explorer give more details on the use of the filters.
Figure 5-6: Display menu. When the user clicks on the Display button in the Usage tab, they are presented with a menu for customizing the display of the current chart or data. Options include:
| |
Figure 5-7: Export menu. Allows the user to download a chart image in a number of formats or to export the underlying data. When the user presses the Export button they will be presented with a list of possible export formats, which include:
The image export options in the Export menu provide the ability to choose the size of the file, the font size and whether to display the chart title in the exported image. The PNG and SVG options have settings for small, medium, large and poster size images and a ‘Custom’ setting that allows fine grained control over the image size in pixels. The PDF option allows the user to specify the width and height of the PDF in inches and the font size in points. | |
Figure 5-8: "Available For Report" check box. Selecting the Available for Report check box will cause the chart to be added to the list of available charts for the Report Generator. Note that the chart will retain the duration and display customizations made prior to checking the box. |
5.3 Chart Selection Tree
Figure 5-9: Chart Selection Tree. The chart selection tree is displayed in the Usage Tab. The tree nodes are grouped by Realm (e.g., Jobs, Allocations, SUPReMM, etc.) and there are three levels of data (nodes).
|
Selection of a node in the tree can be accomplished by left-clicking the node. A triangular icon to the left of the tree node will point to the right when the node is collapsed and will point to the bottom right and have a black fill when expanded. Double clicking a node will select and expand it. Selected tree nodes are indicated by a light blue background.
Collapse All Button : This will collapse all of the expanded nodes in the tree.
Collapse Selector Tree Button : This will collapse the Usage tab Chart Selection Tree to the left so that it is only visible as a thin vertical bar to the left of the screen providing more screen area for the display of the selected chart or data sheet. Clicking on the button will show the Selector Tree again.
5.4 Chart Viewer
Figure 5-10: Chart Viewer |
The chart viewer is shown in Figure 5-10 and provides an area where the user can view and customize metric data as well as export the data or make the chart available for inclusion in a custom report. At the top of the Chart Viewer is a trail of breadcrumbs that shows the origin of the chart. The chart shown in Figure 5-10 is from the view of the Usage tab. The Jobs by Resource category is expanded, followed by selection of the XD SUs Charged: Per Job metric.
Drilldowns: Most of the charts displayed in XDMoD have a drill-down capability, which will allow users to click on a data series (bar, line, pie slice, etc.) and acquire more detailed information about the relevant selection. This is accomplished by hovering the mouse over a particular data series in a chart and clicking on it, causing a new drilldown node to be displayed in the Chart Selection Tree and the new chart to be displayed. When drilldowns are available, the cursor will change from an arrow to a hand. Whenever a drilldown is not possible the user will be notified via a tooltip.
For example, if the second horizontal bar in Figure 5-10 is clicked on (NICS KRAKEN), followed by the selection of Field of Science from the menu that appears, the chart shown in Figure 5-10a will appear. This new chart will take into account all fields of science which have contributed to charging XD SUs to resource NICS KRAKEN (respective of the timeframe selected). Note that the breadcrumb trail reflects the origin of the chart.
Figure 5-10a: Example of a drilldown chart |
5.4.1 Chart Viewer Tools
Figure 5-11: Chart Viewer Tools (Thumbnail Views) |
Located on the top right of the Usage tab chart viewer when a dimension is selected, from left to right, this toolbar consists of the following buttons:
- Restoring the thumbnails to their original dimensions
- Reducing the chart size
- Increasing the chart size
6 Summary Tab
Figure 6-1: Summary Tab
The Summary tab (Figure 6-1) consists of a duration selector toolbar, a summary information bar, followed by a select set of charts representative of the usage. The refresh button can be used to reload the page.
The Summary tab provides a dashboard that presents summary statistics and selected charts that are useful to the role of the current user. Pertinent information such as the number of jobs run, average CPU time per job, and the average number of processors per job are displayed as well as selected charts that provide an overview of selected available data. For Open XDMoD users this includes information on the system as a whole, on various sub-clusters and information specific to the current user.
To access a more detailed and configurable version of each chart in the Summary tab, click on the gear icon to the right of the chart title (see Figure 6-2 below).
Consult the Quick Filters drop-down menu in the upper-left corner to see summary content tailored to specific roles (Figure 6-1).
Based on the user's role, the Summary page may show your jobs (for user role), those of your group (for PI role) or those of your entire center (for center director role).
Figure 6-2: Summary Portlet Tools. From left to right, this toolbar provides a button for collapsing the portlet, a button to configure the chart, and a button that when hovered over will present a tooltip to describe the data in the chart. |
7 Dashboard Tab
The dashboard tab is an optional replacement for the summary tab. The dashboard must be enabled by the XDMoD system administrator and is only available for logged in users. It displays various information in separate display windows called components. The information displayed in the dashboard varies according to the assigned role of each user account. Table 7-1 lists the default settings for each role.
Table 7-1: List of dashboard components by user role. Some of the components are only available if the (optional) XDMoD module is installed.
Role | Components Displayed | XDMoD module |
Center Director | Center summary Saved charts and reports Center Report Card | - - Application Kernels |
Center Staff | Center summary Saved charts and reports Job Efficiency Report Recent jobs Center Report Card | - - SUPReMM - Application Kernels |
Principal Investigator | Recent jobs Job Efficiency Report Summary Charts | - SUPReMM - |
User | Recent jobs Job Efficiency Report Summary Charts | - SUPReMM - |
The various dashboard components are described below.
7.1 Dashboard Components Overview
There are two categories of dashboard components: full-width components and half-width components. Full width components are in a fixed location at the top of the dashboard tab and the half-width components are arranged in columns. All half-width components have a title bar such as the example shown in Figure 7-1. The title bar includes the name of the component and, for components that show data limited by time. show the date range of the data. Clicking the question mark icon on the right of the title bar will show a help window that explains how to use the component.
Figure 7-1: Example title bar of a dashboard component. Typically the title includes information about the date range of the data. The question mark button on the top right is used to open a help window. Some components also have additional icons for component-specific features.
The layout of the half-width dashboard components can be customized by dragging the title bar of a component (Figure 7-2). The layout settings are saved automatically and persist between logins. The layout can be reset to default using the “Reset to default” button in the Settings tab of the “My Profile” dialog (Figure 7-3). The My Profile dialog is accessed via the “My Profile” button on the top left of the XDMoD portal. Full-width dashboard components cannot be moved.
Figure 7-2: Example of rearranging the layout of half-width dashboard components. A component can be moved by selecting the title bar and dragging into the new position.
Figure 7-3: The dashboard layout can be reset to the default value using the button available in the Settings tab of the Profile dialog.
7.2 Chart Component
The Chart Component shows charts from the Metric Explorer. An example of the chart component is shown in Figure 7-4. The time range of the chart is shown in the title bar. The wheel icon on the top right of the component will open the chart in the Metric Explorer.
Figure 7-4: Example of a chart component. The time range of the chart is shown in the title bar. The wheel icon on the top right will open the chart in the Metric Explorer.
The default chart configuration settings are controlled by the Open XDMoD system administrator. Extra charts can be added to the dashboard via the Metric Explorer. Charts are added via the “Show in Summary tab” checkbox in the Metric Explorer basic chart options (see Section 9.2 for more information).
It is also possible to edit the default charts via the Metric Explorer. Click the wheel icon on the top right of the chart component to open the chart in the Metric Explorer. You can then edit the chart in the Metric Explorer and save the chart without changing the chart name field or deselecting the “Show in Summary tab” checkbox. To reset a system chart to the default just delete it from the list of save charts in the Metric Explorer.
7.3 Saved Charts and Reports Component
Figure 7-5 shows the saved charts and reports component. This component lists a user’s charts that were saved in the Metric Explorer and their reports from the Report Generator. The list can be sorted by the name, type (chart or report), and last modification date. By default, the list is sorted by the last modification date in descending order. The user can filter the list by name by using the search box.
Figure 7-5: Shows the saved charts and reports components.
7.4 Center Summary Component
Figure 7-6 shows the center summary component. This component displays the thumbnails of charts from the center summary report that is tailored based on user role. Clicking on a thumbnail opens a modal containing an interactive version of the chart. The user can open the chart in Metric Explorer to modify it. Note that charts modified in Metric Explorer have to be saved and added to the center summary report in order to be displayed as a thumbnail in the center summary component. The user can control the time period of the charts by using the time range menu. The user can also download the charts as a PDF by clicking the Download Report button.
Figure 7-6: The center summary report shows thumbnails of charts from the center summary report that is tailored based on user role.
7.5 Center Report Card
The center report card component is provided by the optional XDMoD AppKernels module. The component shows a summary of the results of recent application kernel jobs. Figure 7-7 shows an annotated example of the center report card.
Figure 7-7: An annotated example of the Center Report Card component that is available in the optional AppKernels module.
7.6 Jobs Component
The Jobs component displays information about recent HPC jobs. An annotated example is shown in Figure 7-8. The job information is based on the Job Performance (SUPReMM) realm if the Job Performance module is installed. If the Job Performance module is not installed then the data are obtained from the Jobs realm and the component will display ‘N/A’ in the CPU column indicating that the cpu usage is not available. The information displayed in the Jobs component is restricted by XDMoD’s role based access controls. A summary of the data available is shown in Table 7-2 below. The “Filter by Person” widget is not displayed for the user role since it is superfluous.
Figure 7-8: Annotated example of the Jobs component. The User role does not have permission to list other users jobs and does not have the “Filter by Person” widget. The Job overall CPU usage information is only available if the optional Job Performance (SUPReMM) XDMoD module is installed.
Table 7-2: Summary of the role based access control restrictions for data shown in the job component.
XDMoD account role | Information available |
User | HPC jobs that were run by the user. |
Principal Investigator | HPC jobs that were run by the user and HPC jobs that were run on the PIs projects. |
Center Staff | All HPC job data. |
Center Director | All HPC job data. |
7.7 Job Efficiency Component
The Job Efficiency component is only available if the Job Performance module is installed. The Job efficiency in XDMoD is defined based on a comparison of the resources (CPU, Memory, etc.) requested to those used by an HPC job. The information in the Job Efficiency component is subject to the same role-based access controls as the Jobs component as listed in Table 7-2. An annotated example of the component view for Center Staff and Center Director roles is shown in Figure 7-9. The Principal Investigator role does not have a toolbar that switches the data grouping between user and PI. The User role example is shown in Figure 7-10.
Figure 7-9: Annotated example of the Job Efficiency component. The total number of jobs and core hours are shown broken down by efficiency categorization. Click on a row in the table to open a window that lists the jobs that were classified as inefficient.
Figure 7-10: Annotated example of the Job Efficiency component for XDMoD accounts with User role. The total number of jobs and core hours are shown broken down by efficiency categorization. Click on the bar chart to show the information about the jobs that are summarized.
8 Usage Tab
Figure 8-1: Usage Tab |
The Usage tab (Figure 8-1) provides a convenient way to browse XDMoD’s realms. It consists of a chart selection tree on the left, and a chart viewer to the right of the page. You can interact with the chart selection tree to explore the realms and view individual metric charts.
Figure 8-2: Usage Tab Metric |
To view a metric in the Usage tab (Figure 8-2), expand the selection tree and select an item. The CPU Hours Per Job metric from the Jobs realm is shown. You can use the Duration and Chart Configuration controls to customize the chart (Refer to section “Common User Interface Elements”). For example, you can:
- Change the data duration
- Change the plot’s display
Note that any selection in the Duration control applies to all charts in the Usage tab.
Figure 8-3: Usage Tab Drilldown |
You can drill down on a metric in the Usage tab (Figure 8-3) by left-clicking on the plotted data and selecting a dimension. Drilldown of CPU Hours Per Job by Resource is shown. To further customize the plot, you can use the configuration controls, or export the plot to Metric Explorer by clicking the Gear icon at upper right of the plot window. |
9 Metric Explorer
The Metric Explorer allows you to create complex plots containing multiple metrics. It has many point-and-click features so you can easily add and filter plotted data and customize plot appearance.
Figure 9-1: Metric Explorer |
The Quick Filter drop-down menu in the top toolbar allows you to rapidly focus the information source used by the Metric Explorer. Based on the quick filter selected, the values that are displayed will vary. For example, if you are a PI, selecting the PI quick filter provides access to the information from all of your students’ jobs; however, if you select User, you see only the information from jobs that you submitted yourself.
Other features of the top toolbar include:
Duration Selectors: The duration selection toolbar is provided for selection of the desired time frame. To define a custom timeframe, specify a start and end date in the date fields to the right of the Duration drop-down menu.
Export Menu: Allows you to download the currently visible chart in graphical or datasheet form, and in various formats.
Print: Prepares a version of the chart for printing, and automatically presents you with a print dialog.
Available For Report: when checked, the exact chart will be included in the report generator’s list of available charts.
For more information about the top toolbar, refer to the “Common User Interface Elements” section in this document.
The Chart Options region above your chart gives you ultimate flexibility in the amount and type of statistics you want to view at any time. The Chart Options region comprises five sections: Charts, Data, Global Filters, and Options. Any changes you make in each section will be immediately reflected in the chart.
The Chart Viewer is the area where the chart is generated by using the chart settings. The chart viewer also uses a paged view. When not all of the data series fit on one chart, multiple charts will be generated. Buttons are provided above the chart for navigation through the charts, when needed. The maximum number of data series in the chart, or page size, can be configured in the chart limit input box above the chart.
The sections that follow discuss the Chart Options region in more detail.
9.1 Managing Charts
Figure 9-2: Charts tools | The Charts section (Figure 9-2a) allows you to manage chart presets. Chart presets are mappings between all the components involved in the chart (e.g. data series, filters, etc.) and the textual identifier you assign to each query. Once you click on to create a new chart, you will be presented with the dialog reference in Figure 9-2b. Here, you have the opportunity to name your chart for easy reference. Similarly, you can click on to remove a selected chart preset. In earlier versions of XDMoD, charts were automatically saved. However, the current version gives you more control over your charts. You must manually save your charts by clicking on either the Save or the Save As button, see Figure 9-2a. Charts or changes to charts that you do not explicitly save will be lost when you exit XDMoD. An Undo button allows you to cancel the last change, and a Redo button lets you cancel the effect of the last Undo. To access your saved charts, simply select the preset from the Load Chart list (Figure 9-2c). If you have lots of charts, you can utilize the Search field located in the top left of the Load Chart area to target specific charts. |
9.2 Basic Chart Options
The Basic Chart Options button can be found at the top right of the chart viewer (Figure 9-1). It allows you to customize general chart settings, see Figure 9-3.
| |
Figure 9-3: Chart Options |
In the present version of the Metric Explorer, you can easily change plots by simply clicking on the data series in the chart. A dialog box opens (Figure 9-4) allowing a wide range of changes, including adding another metric, drilling downs, and options that alter the display format.
Figure 9-4: In the Metric Explorer Plots can easily be changed directly by clicking on them. |
99.3 Managing Data Series
Figure 9-5: Adding additional data to a chart is made easier through the menu on the left-hand-side. Here we show how to add CPU Hours Per Job to an existing plot of NUs charged Per job and Number of jobs Ended (Selecting CPU Hours Per Job is shown in the upper left corner of the window). |
Figure 9-6: Adding data to the plot is as simple as clicking on the chart.
Adding another data series to a chart can be accomplished in several ways. Figure 9-5 shows that clicking on a metric in the chart tree brings up a dialog box. This dialog allows you to choose what data will be displayed and how it will be grouped. In addition, clicking in the chart view area will bring up a chart options box (Figure 9-6) that allows you to choose to add a new metric. Clicking on a particular data series also brings up a dialog box that has an option to select a new metric.
An alternative way to introduce data into a chart is to click on the Add Data button as highlighted in Figure 9-7. Doing so will present you with a collection of categories referred to as realms. Click on a realm to reveal all the respective metrics. In Figure 9-7 the allocations realm has been selected showing 6 possible metrics. Clicking on the XD SU’s Allocated metric brings up the Data Series Definition dialog box. The Data Series Definition dialog box allows you to control how the metric is to be represented and displayed on the chart. In addition, filters can be applied to the data series for even more precision. For example, you may be interested in the number of jobs run by a particular user at an institution for a particular time frame. Using filters allows you to acquire this information.
Figure 9-7: Data Series workflow of Metric Explorer In addition to adding new metrics to a plot, the Metric Explorer also allows the user to drill down for more detailed information on a particular metric. Figure 9-8 shows the dialog box that opens when you click on a data series within the chart viewer. Selecting “Drilldown” will display the data analyzed by the chosen dimension or grouping, as described by Section 5.3 (the Chart Viewer) previously. |
Figure 9-8 Changing how the data for a given data series is displayed can now be achieved simply by clicking on the data series to bring up a dialogue box. |
9.4 Managing Filters
As mentioned in the previous section, filters allow you to focus on a subset of the data. To apply filters to your data, you can either consult the Local Filters section of the Data Series Definition dialog (see Figure 9-10), or make use of Global Filters (see Figure 9-11). Local Filters are solely tailored to a single data series, whereas Global Filters are applied to all the data series you have introduced into the Data section of the Metric Explorer. To create a filter, click on the Add Filter button (as highlighted in Figure 9-11). Doing so will provide you with a list of dimensions (or parameters) you can filter with. Once you have selected a dimension, you will be presented with a dialog allowing you to navigate, search, and select parameter values. Click on OK to apply any selected filters. | |
Figure 9-10: Filters section of Metric Explorer |
Filters in the Metric Explorer are applied as follows: Dimensions are logically and-ed together, and values associated with the same dimension are logically or-ed together. An example of filter logic is illustrated in Figure 9-11. | |
Figure 9-11: Understanding filter logic |
10 Job Viewer
The Job Viewer Tab displays information about individual HPC jobs and includes a search interface that allows jobs to be selected based on a wide range of filters. An example of the job viewer tab is shown in Figure 10-1. The main components of the job viewer are the search history pane and the main viewer pane. The search history pane uses a tree menu that shows all saved searches as nodes in the tree. The jobs that were selected in each search are listed under their respective search node. The main viewer pane has a tabbed user interface that shows the information for a selected job with one job per tab. The different types of information within a job are shown in sub-tabs.
Note that the information available for a given job depends on the data collection software that is used by the resource provider.
Figure 10-1: Job Viewer Tab |
9.5 Searching for jobs
There are two ways to search for jobs. If the job id is known, the quick job lookup function can be used to locate the detailed data for the job. Alternatively, the Advanced search function can be used to find jobs based on one or more search parameters.
To activate the Job Viewer click on the Job Viewer tab, then on the Search button in the upper left corner. This will bring up the Search window (shown in Figure 9-2). The search window has three main components, the Quick Job Lookup form, the Advanced Search form and the Results panel. Use the Quick Job Lookup form to find a known job based on its job id (i.e. the identifier that the job was assigned by the job scheduler). First choose the Resource on which the job ran from the drop-down list. Next enter the desired job number in the Job Number box. Finally click on the Search button in the lower right corner of the form. If there is information for this job in XDMoD then a record will show in the results pane. You may optionally choose a descriptive name for the job search in the Search Name entry box and then click Save Results to add the job to the search history for viewing.
Figure 9-2: Search Window |
Alternatively, if you are looking for jobs that satisfy a general set of criteria, you can use the Advanced Search form (see Figure 9-3). You must specify (1) the time frame for the search as well as (2) one or more Filters. The filter field is selected from the drop down list, the value for this field is then selected from the box to its right and then click on the Add button to add the filter to the search. Multiple filters can be added to the search and the same filter field may be added multiple times with different values. You can optionally enter a descriptive name for the search in the Search Name entry box (3).
After the inputs have been completed, click on Search (4). If there are any jobs that fit the given search criteria, a list of jobs will be displayed. You can select specific jobs that you would like to view by checking the box next to the job ID, (5) then click Save Results (6) to add the selected jobs to the search history for viewing.
Figure 9-3: Using the Advanced Search Form in the Search dialog. |
9.6 Viewing job data
Job information may only be viewed for jobs that are in the search history. The search history tree on the left hand side of the interface shows the saved searches. An annotated example of a search history tree is shown in Figure 9-4. The tree shows all saved searches, all selected jobs for each search and the list of data available for each job. Selecting a data node for a job will result in the job data loading in the main viewer pane.
Figure 9-4: Job viewer search history tree. The tree shows every saved search (1), all selected jobs for a search (2) and the data available for each job (3)
The different types of data available for a job are listed below:
- At-a-glance analytics
- Accounting data
- Job script
- Executable information
- Summary metrics
- Detailed metrics
- Timeseries plots
Note that not all data may be available for all jobs. Availability of data depends on the collection software installed on the HPC resource as well as the hardware capabilities on the compute nodes. The only information that is guaranteed to be present is the Accounting data.
9.6.1 At-a-glance analytics
The at-a-glance analytics are shown at the top of the job tab and are automatically displayed if they are available (there is no data node displayed in the search history for the at-a-glance analytics). There are three plots that show a summary of the values of selected performance metrics of the specific job. The metrics that are displayed here are normalized with values near 1 (displayed in green) indicating a good value and values near 0 (displayed in red) indicating a poor value for the metric. We note that not all jobs should be expected to have good values for all metrics. The default metrics that are displayed are CPU User Balance, Homogeneity, Memory Headroom and CPU User (see Figure 9-5). CPU User Balance indicates how well balanced the usage is between the various cores in the job based on the fraction of time each core spends in CPU user mode. CPU User Balance is defined as
where 
is the CPU usage for the core with the highest average usage and 
is the CPU usage for the core with the lowest average usage. A value near 1 is ideal in that all cores are working equally; a value near zero indicates that one or more cores are nearly idle. Homogeneity is a measure of how uniform the L1D cache load rate is over the lifetime of the job. The L1D load rate is a useful metric measuring if the job is accomplishing useful data processing. High values near 1 are good; if the value is low near 0, the job should be investigated to determine if data processing terminated prematurely. CPU User is the ratio of CPU clock ticks spent in the CPU user mode summed over all of the cores divided by the total clock ticks for the job for all of the cores. Values near 1 are ideal indicating that all cores are spending most of their time in CPU user mode. Values near 0 indicate a job where the cores are spending only a small fraction of their time in CPU user mode.
9.6.2 Accounting data
This tab shows the information about the job that was obtained from the resource manager . This includes timing information such as the start and end time of the job as well as administrative information such as the user that submitted the job and the account that was charged.
9.6.3 Job script
This tab shows the job batch script that was passed to the resource manager when the job was submitted. The script is displayed verbatim.
9.6.4 Executable information
This tab shows information about the processes that were run on the compute nodes during the job. This information includes the names of the various processes and may contain information about the linked libraries, loaded modules and process environment.
9.6.5 Summary metrics
This tab shows a table with the performance metrics collected during the job. These are typically average values over the job. The label for each row has a tooltip that describes the metric. The data are grouped into the following categories:
- CPU Statistics: information about the cores on which the job was assigned, such as CPU usage, FLOPs, CPI
- File I/O Statistics: information about the data read from and written to block devices and file system mount points.
- Memory Statistics: information about the memory usage on the nodes on which the job ran.
- Network I/O Statistics: information about the data transmitted and received over the network devices.
9.6.6 Detailed metrics
This tab shows the data generated by the job summarization software. Unlike the summary metrics listed above, this data is not normalized and the format and content is specific to the job summarization software used. Please consult the relevant job summarization software documentation for details about these metrics.
9.6.7 Timeseries plots
The final aspect of the Job Viewer tab is the ability to view timeseries plots of several different metrics. The available plots for a job are listed under the Timeseries tree node in the search history (see Figure 9-6). The metrics that are available depend on the collection software that runs on the compute nodes as well as the hardware on the nodes themselves. A brief description of the metrics follow:
- CPU User: The ratio of time spent in user mode to total CPU time for the cores that were assigned to the job.
- L1D loads: The ratio of L1D cache loads to reference CPU clock ticks for the cores that were assigned to the job.
- Memory bandwidth: The rate of data transferred to DRAM.
- Memory Usage: The memory usage reported by the OS for the nodes that were assigned to the job.
- Interconnect MPI traffic: The rate of data transferred over the high-speed interconnect
- Parallel filesystem traffic: The rate of data read from and written to the parallel filesystem
- SIMD instructions: The rate of SIMD instructions (this is correlated to the number of floating point operations).
- Process memory: The amount of memory used by the processes that were run by the job
- NFS Filesystem traffic: The rate of data read from and written to NFS mounted filesystems.
It is possible to drill down on the timeseries data by clicking on the data series on the chart or by using the search history tree. The degree to which it is possible to drill-down varies according to the metric. For example, the CPU metric shows the compute node-level average values and it is possible to drill down to the per-core values. The memory metric shows the compute node-level average value and it is possible to drill down to the individual NUMA nodes (for supported collection software).
Figure 9-5: Job Viewer Information |
Figure 9-6: Chart vs Time |
10 App Kernels
Figure 10-1: Application Kernels |
The Application Kernels Tab consists of three subsections, each has a specific goal in order to make viewing Application Kernels simple and intuitive. The three sections consist of the Application Kernels Viewer, Application Kernels Explorer and the Reports subsidiary tabs. The section below will describe each of these subsidiary tabs in much greater detail
10.1 App Kernels Viewer
The Application Kernel Viewer selector tree (Figure 10-1) functions similarly to the chart selection tree of the Usage tab. Data is displayed hierarchically where the top level nodes are the names of the available application kernels followed by the resources on which that kernel was run, followed by the metrics that were generated on that resource, followed by the number of processing units used as shown in Figure 10-1. Clicking on the tree nodes will display a set of thumbnails in the chart viewer that reflects the node's location in the hierarchy - the farther down in the hierarchy that a node lies, the more specific the set of data displayed. The available node types are:
- App Kernel: Selecting an application kernel name will load all thumbnails associated with that kernel (all metrics for all resources).
- Resource: Selecting a resource loads the thumbnails for all the metrics collected for this resource and application kernel.
- Metric : Selecting a metric will display the chart with a dataset for each number of processing units.
- Processing Unit: Selecting the number of processing units will display data for the specific number of processing units for that metric, resource, and application kernel.
When navigating the App Kernels Viewer subtab, targeting solely a specific number of node configuration will yield a comprehensive plot:
Figure 10-2: Application Kernel Viewer Plot |
The following table helps to analyze the plot more effectively:
Change Indicators | An exclamation point icon is shown if and whenever a change occurs to the execution environment of the app kernel (library version, compiler version, etc). |
Running Average | Shown as a dashed line on the chart. The running average is the linear average of the last five values. |
Control Band | A green band is shown, representing the values of the running average considered "In Control" at any given time. A control region is picked to be the first few points in a data set and updated whenever an execution environment change is detected by the app kernel system. The control band then is calculated by clustering the control region into two sets based on the median and then finding the average of each set. The two averages define the control band. |
Control Zones | A red interval is shown on the plot when the control value falls below -0.5, indicating an out of control (worse than expected) running average, and a yellow interval when the control value is greater than 0, indicating a better than control (better than expected) running average. Other running average values are considered "In Control". |
Control Plot | Represented as a dotted line. The control is calculated as the distance of the running average to the nearest boundary of the control band, normalized over the range of the control band. |
Discrete Controls | Convert the control values from real numbers to discrete values of -1, 0, 1. Values less than zero become -1 and values greater than zero become 1. |
10.2 App Kernel Explorer
Figure 10-3: App Kernel Explorer |
The App Kernel Explorer (Figure 10-3) allows you to conveniently compare performance between two or more resources, provided the selection of processing units and metrics. Any changes in selections made on the left-hand section of the Data Explorer will immediately be reflected in the right-hand section (Data Comparator).
It is important to note that changes made in the Metrics Section affect the availability of options in the Resource Selector.
Selections made in the Processing Units section are considered global in the sense that they are applied to every single metric selected. Any selections (and de-selections) made in Processing Units override those made for any metric. Processing units can be selected by clicking the check boxes.
Alternatively, all possible values for either Metrics, Resources or Processing Units can be toggled on or off using the global selection boxes as shown in Figure 10-3a.
Figure 10-3a: Reset buttons for App Kernel Explorer sections |
10.3 Reports tab
The Reports tab (Figure 10-4) is used for sending reports based on a specific Resource and App Kernel. The first section of the Reports tab is for sending reports periodically. The options include daily report, weekly report and monthly report. Each of these time frames contains a pull down menu which gives a set of options indicating under what circumstances a report will be sent. The options include always, on any error, on any major error, on pattern errors or on major pattern errors.
The weekly report and monthly report also have another drop down menu which gives options for day of the week for weekly report (e.g. Monday, Tuesday, etc.) and which day of the month for monthly report (e.g 1st, 2nd, 3rd, etc.). Once all settings have been selected, you can click save settings in order to start sending reports.
There is also an option to send reports covering specific and customizable date ranges via email; this can be found directly below the first section.
Figure 10-4 Reports Subtab |
11 Data Export
To manage data export requests, click on the Data Export tab (Figure 11-1).
The Data Export functionality allows you to request batch exports of non-aggregate records from the data warehouse. These requests are limited to a one year duration, but there is no limit to the number of requests. The exported data can be formatted as either CSV or JSON. Submitted requests are typically fulfilled within 24 hours and an email is sent to notify you that your data is ready for download.
The Data Export tab appears as shown below with a form used to create requests (Figure 11-2) on the left and a list of existing requests (Figure 11-3) on the right.
Figure 11-1 Data Export Tab
11.1 Create Data Export Request
To create a new request fill out the form and the left and click the button. When a realm is selected the list of fields that will be exported are displayed below the form. A start date, end date and format must be selected. The maximum duration is one year. The export formats are CSV and JSON. The JSON export is structured as an array of arrays.
Figure 11-2 Data Export Create Request Panel
11.2 Manage Existing Export Requests
After submitting a request it will be added to the list in the "Submitted" state. The request may be deleted before it is fulfilled.
Figure 11-3 Data Export Status Panel
Requests are fulfilled on a set schedule and emails are sent to notify you that your data is ready for download. The email will include a link that can be used to download the file and the file will also be available to download through the web portal. The file is retained for a set amount of time, typically 30 days. The expiration date will be displayed in both the email and the web portal. After this date the file will no longer be available for download, but the request may be resubmitted.
12 Report Generator
To manage reports, click on the Report Generator tab.
The default view of the Report Generator UI is shown below. The left region provides a listing of any reports you have created. The right region displays any charts you have chosen to make available for building a report (see Available for Report check box section).
Figure 13-1: Reports Overview window |
Each report entry present under the Report Generator tab displays the title of the chart, whether it was generated manually or from a template, the delivery schedule, the delivery format and the number of charts. In the Available charts section each chart has its title, the timeframe of interest, and a thumbnail view. The time frame defaults to the one chosen prior to making the chart available for reporting.
If you create a chart by a process that includes at least one filtering step and choose to make that chart available for reporting, information on the filtering and drilldown processes can be viewed in by moving your mouse over the information icon to the left the chart title in the chart entry as shown in Figure 12-2.
Figure 12-2: Viewing additional details for a chart entry |
The top toolbar serves as the starting point for managing a report.
Figure 13-3: Report Generator toolbar |
New: Creates a new report. A default (and unique) name will be chosen, and no charts will be included.
New Based On: Creates a new report which contains all the charts and settings of an existing report. A default (and unique) name will be chosen and assigned to this report.
Edit: Opens an existing report for editing.
NOTE: Another way to open a report for editing is to double-click on one of the entries in My Reports:
Figure 12-4: Editing a report |
- Preview: Previews the (selected) report prior to sending it.
- Send Now: Builds and sends the (selected) report to the email address associated with the user.
- Download: Builds the (selected) report and presents it to the user as a downloadable file.
1312.1 Selection Model
All the lists / grids under the Report Generator tab have been designed such that multiple selections can be made for more efficient workflows.
Clicking on one entry, followed by holding shift while clicking another entry results in the selection of both entries along with all the entries that fall in between them:
Figure 13-5: Selecting charts for addition in a report |
Clicking on one entry, followed by holding the control key (or command key on a Mac) while clicking another entry results in individual selections being added (appended) to what is currently selected.
Figure 13-6: Another method to select charts for addition in a report |
Note: Deselections can be made by clicking on selected items in the grids while holding down the control key (or command key on a Mac).
Whether you choose to create a new report or edit an existing report, you will be directed to the Report Editor:
Figure 12-7: Selecting reports and charts via the Select menus |
An alternative way to select reports and charts in the Report Generator is to make use of the Select menus (as highlighted in Figures 13-7a and 13-7b). Using these select menus is especially helpful if you are dealing with a substantial number of reports or charts. The select menus also gives you the ability to invert selections made in the report and chart grids. In other words, when you click on Invert Selection, any items previously selected will become deselected, and vice-versa (see Figure 12-7c).
Figure 13-8: Report editor |
General Information for this report is displayed on the left side of the Report Editor. The only required field is the report File Name, which must be unique. When creating a new report (including a new report based on another report), a default unique file name will be generated and assigned to the File Name field. The Chart Layout section allows you to control how many charts are to appear on a single page. The Scheduling section allows you to customize the frequency in which reports are to be sent to your email address (associated with your account). Schedule options include Daily, Weekly, Monthly, Quarterly, Semi-annually, and Annually. To disable automatic delivery of reports, set the Schedule option to Once. Currently available formats for reports are PDF and Word Document. |
Figure 13-9: Information to be entered to customize a report
To add charts to a report, simply drag an entry from the list of Available Charts (on the right-hand side) into the Included Charts list. See the section entitled Selection Model for adding multiple charts at once.
The order in which the charts appear in the Included Charts section of the Report Editor dictates the order in which they will appear in the report. Re-ordering of charts in the report can be accomplished by simply dragging the charts into the desired order:
Figure 13-10: Reordering charts in a report |
1312.2 Changing Chart Timeframes
Should you decide (during the report building process) that you want to change the timeframe of a given chart, you can easily accomplish this task by either clicking on the timeframe or the edit icon to the left of the time frame in the respective chart entry. As a result, you will be presented with a dialog for conveniently updating the timeframe to your liking.
Figure 12-11: Changing the timeframe of a chart entry in the report editor |
Note: If you want to edit timeframes for multiple charts in a report, select two or more charts, then click on the Edit Timeframe of Selected Charts button in the top menu. Doing so will bring up the same Edit Chart Timeframe dialog as mentioned above. When you click on Update, each of the charts you had previously selected will update to reflect the new timeframe you specified.
12.3 Previewing a Report
At any time during the report editing process, you may preview the current state of the report by first saving and then pressing the Preview button (previewing reports is also possible from the My Reports view):
Figure 13-12: Previewing a report |
12.4 Building, Exporting Reports
To build the report (as a PDF or MS Word document), you can make use of the Send Now and Download options available from any of the following sections: My Reports, Report Editor, and Report Preview. The Download feature will allow you to immediately access or download the resulting document, whereas the Send Now feature will email the constructed report to the address associated with your profile (see the My Profile section).
Figure 13-13: Sending a report with Send Now or Download |
13 My Profile, About, Roadmap, Contact and Help
On the upper right side of the XDMoD page, there are a series of five buttons that allow the user to set his profile, get more information about XDMoD, view the XDMoD Roadmap, contact the XDMoD developers and seek help.
Figure 15-1: My Profile, About, Roadmap, Contact and Help functional buttons |
13.1 My Profile Button
My profile allows you to view and update general settings pertaining to your account. To access your profile, click on the My Profile button located in the upper right section of the portal.
Your current role will be displayed in the title bar of the My Profile window. In the figure above, for example, it shows the user is a Program Officer.
The first section in the My Profile window contains User Information such as: First Name, Last Name, and Email Address. All of the aforementioned fields must be specified in order to update your profile.
The last section in My Profile allows you to optionally update your account password. In order to update your password, click the Update radio button and specify the new password twice into the fields that follow.
Figure 15-2: Password change window |
Also in the My Profile window is the option to choose your Role Delegation by clicking the tab at the top of the window (Figure 13-2a). This tab allows you to choose a specific staff member connected to your account to delegate special privileges.
Figure 13-2a: Role Delegation Menu |
13.2 About Button
Clicking the about button will display a page of information about XDMoD.
13.3 Contact Button
Selecting the contact button will bring up a window that allows the user to send a message to the XDMoD team. The user must fill in his name, email address and his message.
Figure 15-3: Contact window |
13.4 Help Menu
Figure 15-4: Help menu |
Clicking on the Help menu will provide you with the following options: User Manual, FAQ and YouTube Channel (Figure 13-4). Clicking on User Manual will direct the user to the XDMoD User Manual (the document you are viewing at this very moment). If help is available for the section of XDMoD you are currently visiting, the User Manual will automatically navigate to the respective section when it loads. Clicking on FAQ will take you to a page containing Frequently Asked Questions. Clicking on YouTube Channel takes you to the XDMoD YouTube Channel.
13.5 Roadmap
Clicking on the RoadMap button brings up a screen which displays an information page giving details on how to navigate to the XDMoD development roadmap (Figure 13-5). Click on any feature in this page and you will be taken off of the XDMoD site to the Trello.com site where you will be able to view and interact with the XDMoD roadmap (Figure 13-6). To return to XDMoD, simply close the browser tab/window. If you simply want to view the Roadmap you can select the various Trello cards shown in Figure 13-6 for further details about each roadmap task. You can also “vote” for a feature in the roadmap that you would particularly like to use. To do this Sign Up or Log In to Trello (upper right corner Figure 13-6). Then clicking on any roadmap feature such as X-Y plot capacity, will give you access to the Vote box, as shown in Figure 13-7.
Figure 13-5: Roadmap information page |
Figure 15-6: XDMD Roadmap on Trello.com |
Figure 15-7: Voting for a feature in the XDMD Roadmap on Trello.com |
14 Design Philosophy
The XDMoD portal provides end users with a rich web-based user interface as shown in Figure 2-1. Developed using the Model-View-Controller design pattern, the view is provided via a thin client written entirely in JavaScript and built on the ExtJS user interface toolkit while the Controllers and access to the Model are provided by a set of web-based RESTful services. Information in the portal is organized using a set of tabs that allows related information to be grouped together while providing the user an easy way to navigate the interface and make the most of the available screen real estate. The RESTful services provide access to data stored in the warehouse while maintaining a clear separation between the user interface and the application logic. This also allows 3rd party tools to access the services directly and ingest or present the information as they see fit.
15 Appendix I-XML Document Type Definition
<!DOCTYPE xdmod-xml-dataset [
<!ELEMENT header (title, parameters?, start, end, columns)>
<!ELEMENT rows (row?)>
<!ELEMENT parameters (parameter)>
<!ELEMENT parameter (name, value)>
<!ELEMENT row (cell)>
<!ELEMENT cell (column, value)>
<!ELEMENT title (#PCDATA)>
<!ELEMENT name (#PCDATA)>
<!ELEMENT value (#PCDATA)>
<!ELEMENT column (#PCDATA)>
]>
XDMoD User Manual v10.0 CCR @ UB