The advanced analytics and reporting features of Data Center Analytics enables you to monitor various key performance metrics of monitored resources at different time periods and enables you quickly identify the problems.
Real time analytics
Data Center Analytics supports viewing real-time data which helps you assess the status of the monitored resources in real time.
The Data Center Analytics reports display data in Near Real Time View: or Real Time View depending on the option that you select while configuring the Data Center Analytics server.
Near Real Time View: In this view, the reports show the latest data that is available for the monitored resources. This is the default view of the reports. The default refresh interval is 15 minutes.
Real Time View: In this view, the reports show data in 1-minute granularity for the monitored resources.
Flexible reporting and analysis using Data Center Analytics
In the fast-paced world of online transactions, many companies with global operations have invested in a sophisticated IT infrastructure that provides them a competitive edge. Monitoring and reporting features enable organizations to monitor applications closely and continuously to proactively identify any problems before they manifest into something more severe and requires immediate attention. Whether you are an IT manager for a bank, health care provider, or a government sector, proactive monitoring and reporting are useful in determining the performance trend of your system and addressing ways to improve customer service interactions in advance of customer feedback. To do this thoroughly requires a tool that can help track the health of you system at all hours and display the relevant metrics instantly in a report that you can share with your organization for assessment.
Hitachi Infrastructure Analytics Advisor integrates with Data Center Analytics to provide advanced reporting capability to continuously measure and analyze performance of your monitored resources. The up-to-date visual representation of your system's health enables you to share reports with others. You can create three types of reports:
- Predefined reports provide high-level details at the application level and also a granular report that shows component-level performance data.
- Ad-hoc reports enable you to combine related and unrelated metrics of any monitored resource in one report to review the overall performance impact.
- Custom reports you create with a report builder.
All reports are included in the Reports dock, and are available when you select any storage system object in the storage systems hierarchy. Predefined reports differ based on your selection of the storage system object. An interactive chart and filtering resources enable you to view every detail in any report. You can also filter reports to display the most relevant data, and can print, create a PDF, and export a report to a CSV file.
Each node in the tree has predefined reports that cover important attributes of a metric to help your analysis of the resource. If you expand and click a node, for example, 609315f7 under Pools in the tree, the performance report displays. In this case, the Pool IOPS Vs. Response Time report displays and it only shows the metrics data for the 609315f7. No data for other Pools appear on the report.
On the reports, nodes are resources such as RAID Storage 302c7d0 and RAID Storage 302c6d6, and metrics such as cache usage and write pending rate. You can do a comparison between any nodes or between metrics of a single node or different nodes. In Add Report, type the report name in the field, then add specific metrics by dragging and dropping a node from the tree to either the axis section Y/Left or Y1/Right. The left and right axis boxes display the list of available resources, for example, virtual machines and hosts.
If, for example, you want to see a pattern for a storage node between two time periods, you can compare the reports on Storage IOPS to display in one view. Each graph line is color-coded and you can zoom in reports to get a better view.
You can also compare how one metric affects the other metrics. For example, you can create an ad-hoc report that compares IOPS with Response Time. This most commonly used report shows whether an increasing load on the system (IOPS) affects the performance (response time).
To create ad-hoc reports, you can combine the related and unrelated resource metrics and drag and drop the metrics into the report from the specific instances in the tree. For example, you can see the metrics for ports and volumes in one chart at any time. Attributes that are directly related, for example, IOPS and Response Time, usually have a built-in report from the Reports dock. Sometimes, the attributes can be unrelated (or indirect) such as the storage system cache usage from the file system transfer rate on a host can consume most of the storage from the array. You can add unrelated metrics and create a comparison chart.
If the predefined charts and ad-hoc are not sufficient, you can create custom reports by building your own query. The Custom Reports feature is based on the Data Center Analytics query language. This regex-based expressive query language retrieves and filters the data in the Data Center Analytics database.
The Data Center Analytics query language allows complex analysis on the data in real time with constant run-time. The syntax makes it possible to traverse relations, identify the patterns in the data, and establish a comparison between metrics of a single component or multiple nodes.
The Data Center Analytics UI helps you build your custom query in the following three ways:
Start with a predefined query and customize it as required.
Build the query using the Build Query feature.
- Write the query directly using Data Center Analytics query language.
Strategic planning using trend analysis in Data Center Analytics
Strategic planning with trend analysis provides a repository and analytic reporting engine that enables you to identify and analyze historical performance trends necessary to optimize storage system performance and plan future capacity growth. As an IT Manager of your company, one of your primary responsibilities is to plan and set aside budget for CAPEX costs required for future growth of IT infrastructure, specifically hardware and management software, hypervisors, switches, and other network equipment. You require an easy way to predict and scale up to satisfy future needs and growth of the organization.
The Data Center Analytics management server collects and reports performance and configuration data over time. Using historical data, you can evaluate the current data usage and predict future requirements. The following report displays the storage capacity usage trend for a selected storage system over a specific time period.
The example report shows an increase in the subscription for 53086_Capacity from April 9, indicated by a blue line. This increase suggests that the pool requires additional capacity to meet the subscription commitment. As in the example report, if the consumption increases suddenly, the pool is at a greater risk of running out of disk space. Therefore, you must add more storage capacity.
Because of the short time window in which the report is created, the change in capacity is minimal, but for a longer period of time, it will be more visible. These reports are useful for you to do additional capacity planning closer to the time of actual requirement.
Trend analysis is an analytical tool to validate the effectiveness of your storage provisioning strategy over a time period. If the measure of capacity required to fulfill the subscription commitment compared with total available free capacity of a storage pool is consistently high, this indicates that your actual capacity is inadequate to meet your subscription commitment. If the measure of capacity is low, this suggests that the pool will not completely utilize the provisioned capacity beyond the current levels and you can safely move the unused capacity to another pool.
Monitoring and quick troubleshooting with Data Center Analytics
Many companies with global operations have invested in sophisticated storage infrastructure that provides them a competitive edge. Even the smallest down time in any of these critical applications has a cascading effect and results in logistical challenges. Therefore, as a Storage Administrator of your company, you must monitor these applications closely and continuously to proactively identify and stop potential problems.
Hitachi Infrastructure Analytics Advisor analyzes configuration and performance data from storage systems, hypervisors, and operating systems. It defines resource SLO thresholds based on the service agreement, and monitors the service level through customers' threshold alerts. To maximize storage performance and ensure performance is at peak efficiency, Infrastructure Analytics Advisor taps into a scalable data repository and advanced diagnostic engine to rapidly diagnose and troubleshoot storage performance bottlenecks. The most common problem is slow response time of applications.
The problem could be in any storage component such as the front-end ports, controllers, or disk drives. Infrastructure Analytics Advisor automatically sends a notification to you when a monitored metric of a storage component exceeds the defined threshold. The notification contains details of the component that exceeded the threshold to enable you to quickly identify the problem and troubleshoot it.
In the example, you navigate from the tree view of Data Center Analytics, which shows a hierarchical representation of the various storage system objects, to the highlighted storage system, and then selects an object to analyze. In this example, controller 0 exceeds the defined threshold of a monitored metric.
You quickly view the built-in component reports for historical configuration and performance metrics, and notice some unusual or unexpected behavior in the report for Transfer Rate. You see an unusual peak in activity close to midnight.
To take a closer look, you zoom in on the report.
You confirm there was a spike in activity for Write Transfers close to midnight and you must determine if this is a regular pattern or just a one-off situation. By selecting another period (day) to compare with the current values, you confirm that a similar peak occurred the day before.
You choose to review similar Configuration and Performance reports for other components, DP Pools, RAID Groups and other storage array components to analyze the affect on performance at the application level.
By focusing on the overall application instead of individual volumes or similar components in detail, you can view the performance metrics at an application level. This summary of individual resource metrics gives you a consolidated view of the overall performance. This helps you to identify and solve the problem faster than viewing individual volumes or similar component metrics.
The ability to compare related metrics enables you to quickly compare the data transfer rate and its throughput performance generated by the application, ports, and storage system to view the affect the application has on the storage system and ports. If the application is utilizing a lot of bandwidth, you can decide to provide additional bandwidth for the application or promote applications to a higher storage tier.
Monitoring of third-party resources in Data Center Analytics
Resources associated with probes that are not supported by the Infrastructure Analytics Advisor GUI are referred to as third-party resources. Such resources must be monitored using the Data Center Analytics server GUI.
Although theData Center Analytics management server collects the data from all configured probes, certain probes are not yet supported by the Infrastructure Analytics Advisor GUI. This means that non-Hitachi storage systems, certain network hardware, and hosts must be monitored from the Data Center Analytics server GUI.
The Data Center Analytics server GUI includes a separate dashboard and a set of alert thresholds that must be configured separately from those present in the Infrastructure Analytics Advisor GUI.
In this example, the dashboard shows a warning icon (!) for volumes associated with resource HM800H241001. Clicking on the icon reveals that the threshold for Total IOPS has been exceeded. Just as you would drill down within the Infrastructure Analytics Advisor GUI, clicking on this icon in turn displays the associated performance graph.