Teradata Viewpoint 14.10 Release
This article is the official release announcement of Teradata Viewpoint 14.10 with an effective release date of May 6th 2013. With new enhancements in Alerting, Workload Management and Monitoring areas, this release of Viewpoint 14.10 continues to expand its scope and provide ability to monitor Hadoop systems along with Aster and Teradata systems.
The primary themes of the Viewpoint 14.10 release are to provide front end and visualization for new Teradata Database 14.10 features and Hadoop system monitoring. There are enhancements in Alerting, Monitoring and Management areas. Following are the highlights of Viewpoint 14.10:
- Stats Manager
- Hadoop System Monitoring
- Workload Management enhancements (Group throttle, New classifications, ability to unlock rulesets, etc..)
- Reports in the Query Monitor portlet
- Alerting Enhancement
Browser support has also been updated to reflect support for Firefox 18, Chrome 24, Safari 5.1, IE 8.x and 9.x.
The Stats Manager portlet complements the Auto Stats feature of Teradata Database 14.10 and will work with relaese 14.10 and later. Stats Manager allows DBAs/Users to efficiently manage their stats collection process. It is a new Tool option in Add Content | Tools menu.
Before we go into details of this new feature, let’s discuss why this is needed. Accurate cardinality and cost helps Teradata optimizer to decide an optimal plan. Statistics provides cardinality information to Teradata optimizer. Cardinality changes significantly with bulk load jobs making stats stale and inaccurate. Some times it is even challenging for an experienced DBA to understand which object stats would be beneficial which can result in collecting extra stats or missing collections of critical stats. Collect stats jobs usually are resource intensive jobs as they have many collect stats statements; it is always good to know what is needed and what is not and save some CPU cycles. Due to scheduling issues the user may not have enough time to complete the collect stats job and there is a need to prioritize and run collect stats for important or stale stats first. Stats Manager tool simplifies some of these tasks and help users automate the stats collection process. The Stats Manager portlet can be used to:
- View statistics on a system
- Schedule statistic collection jobs
- Identify missing stats
- Detect and refresh stale statistics
- Identify and discontinue collecting unused statistics
- View when statistics were last collected and are scheduled for collection again
- Set priority of collect stats statement with regards to to other collect stats statement
- Shows CPU Utilization of collect stats jobs allowing the user to analyse if a particular job consumes more than anticipated amount of CPU.
There are two main tabs in Stats Manager – Statistics and Job.
Statistics tab shows all objects (e.g. databases and tables) on the system, that have at least one statistic or that has at least one outstanding recommendation. The user can drill down on the data grid to navigate between the database, tables and Column. Figure 1 is example of Statistics by Database view.
Actions has three options -
Automate enables statistics to be collected by collect jobs.
Deautomate stops statistics from being collected by collect jobs.
Edit Collect Settings allows the user to edit thresholds, sampling, and histogram settings. The information bar displays the percentage of statistics that are approved for automation, allowing the user to determine if more statistics need to be approved for automation. Percentage of automated stats have collect jobs allows the user to determine if additional collect jobs are needed. Recommendations display a list of the recommendations by an analyze job. By clicking the link the user has an option to approve or reject recommendations given by analyze job. Statistics Table displays all objects with at least one statistic, or one recommendation that has not been approved or rejected. The table is configured using Configure Columns from the Table Actions menu. The user can automate any objects for stats collection process in this tab. This allows the user to approve statistics for collection by collect jobs. The user can also view Statistics detail reports by drilling down to stats object, see Figure 2.
Job tab displays the list of user-defined collect and analyze job definitions. From this view, the user can create collect stats and analyze jobs, manage existing jobs, and review job reports. Figure 3 represents the top Job tab layout.
Actions has three option -
New Collect Job enables user to define a job to collect statistics,
New Analyse Job enables user to define a job to evaluate statistic use and make recommendations and
View History lists the run status and reports for collect and analyze jobs over time.
Job Definitions Table displays summary information about jobs and allows drill down to show the details. Job Schedule displays a nine-day view of jobs that are running, scheduled to run, or have already run. Mouse over a date to have it show the list of jobs.
A Collect job generates and submits COLLECT STATISTICS statements to the Teradata Database for objects that were approved for automation in Statistic Tab. The user can assign a priority to individual COLLECT STATISTICS statements. see Figure 4.
The user can schedule a job to run for limited time and then have a new schedule to resume the job at a different time of the day (See Figure 5).
An Analyze jobs option allows the user to evaluate statistics status and get statistic-related recommendations. Analyzing objects enables the user to determine where additional statistics might be useful and identify existing statistics that are used frequently or are stale. Once the recommendation are generated the user can review and automate the object for stats collection process in Tab. See Figure 6 for various functions that Analyze job can perform.
The Viewpoint Log Table Clean Up feature can be used to cleanup job results stored in DBS TDStats database.
Hadoop System Monitoring
Teradata Viewpoint 14.10 supports Hadoop system monitoring for Hortonworks provided Hadoop solutions packaged as part of Aster 3 Big Analytic Appliance. A new Hadoop Services portlet allows users to monitor status of various services running on the Hadoop systems. Using expandable service view on MapReduce, HDFS and HBase users can view key metrics details for the selected services (See Figure 7).
The Aster Node Monitor portlet has been renamed to Node Monitor as it now monitors both Aster and Hadoop systems. Using the Node monitor portlet for Hadoop systems, users can view node level metrics, available Hadoop services, and the status of services for each node on the system. User can also view hardware statistics details such as CPU usage, memory usage and network activity. Navigating through the Hadoop system topology, users can also view detailed service component and JVM metrics for the HDFS and MapReduce services. (See Figure 8)
Like Aster system monitoring, Hadoop systems monitoring was also integrated with existing portlets. The usability, look and feel of the portlets were maintained but underlying data and metrics corresponded to the monitored Hadoop system. Below are the existing portlets that were modified to support Hadoop system monitoring:
- Alert Viewer – View all the Alerts logged for Hadoop systems.
- Capacity Heatmap – Displays trends for key metric usage related to system, HDFS and MapReduce.
- Metrics Analysis - Displays and compares trends for key metric usage related to system, HDFS and MapReduce in a graphical format across different Hadoop systems.
- Metrics Graph – Displays trends for key metric usage related to system, HDFS and MapReduce in a graphical format.
- Space Usage – Monitors space usage on a Node such as total space, current space, percent in use and available space.
- Admin – Provides the ability to add Hadoop systems and define Alerts for Hadoop systems.
- System Health- Hadoop systems can be identified a “H” in the system's icon and drill down shows all the key metrics related to Hadoop system. See Figure 9
Reports in Query Monitor
In Viewpoint 14.10 we added three new reports in Query Monitor.
Multi-Session report: New option in Query Monitor
By Utility|By Jobwas added to display all the utility jobs that are running with drill down capabilities for individual sessions logged on by a particular Utility Job and the ability to further drill down to see session details. (See Figure 10)
Hot AMP report: A new option
By Vproc|By Skewed AMPdisplays AMPs with most skewed sessions that exceeded the CPU skew threshold set in the
PREFERENCESview. (See Fig 10)
By PE report: A new option
By Vproc|By PEdisplays total number of sessions logged on to the PE and CPU value for the PE. (See Fig 10)
Teradata Workload Management enhancements
Teradata Viewpoint 14.10 introduced group throttles where a user can define throttle on a group of workloads. We also added new classifications by UDF, UDM, memory usage and collect stats. These features are dependent on Teradata 14.10. In Teradata Viewpoint 14.10 user can now unlock any ruleset if they have the appropriate permissions. Users can now also model a system Ruleset this is useful for comparing the Workload management features for different platforms (Appliance v/s EDW) or for different versions of Teradata.
Various new Alert options and Alert type were added in this release of Viewpoint.
- An option to send an alert for Teradata Database restart was added.
- In Session alert include or Exclude users option was added. If user wants to define a session alert for small set of users they need not add other users to the exclude user list instead include user option can be used. It also supports splat wildcard. (See Figure 11)
- Users can now send an alert for long running sessions using newly added Active time alert option in Session Alert type.
- Spool space (MB) alert option was added in session alert to send an alert if a session uses more than anticipated amount of spool space.
- Delta I/O (logical I/Os) alert option was added to send an alert for a session consuming excessive logical I/O during the last collection interval.
- In Database Space alert type users can now specify threshold for Current Spool Space (%) and Peak Spool Space (%) to send an alert when Current Spool Space and Peak Spool Space exceeds the threshold. Splat wildcard support was added to Database space include/exclude user list
- A new Alert type Table space was added late in the Viewpoint 14.01 release with a new alert option on DBC.TransientJournal table with ability to specify current perm and skew threshold.
In Viewpoint 14.10 we modified the Lock Logger architecture for Teradata Database14.10 and follow-on releases. When Viewpoint 14.10 is used with Teradata Database 14.10 the Lock Info collector uses the data written to the DBQL Lock Log table to capture lock information therefore DBQL query logging must be enabled with the “WITH LOCK” option.
We continue to have a voluminous release with copious features across a number of strategic areas. Hope you avail the new additions and improvements in Teradata Viewpoint 14.10. We always look forward to your thoughts and comments.