Performance Impacts and Anomalies

This module shows us how Dell AIOps helps identify potential performance issues in the storage environment. Dell AIOps uses time-correlated analytics to identify performance impact scenarios including an increase in latency while IOPS and/or bandwidth decreases. It also utilizes machine learning to compare current performance metrics with historical values to determine if any of the current metrics deviate from historical and expected ranges.

In this lesson, we will look at how Dell AIOps can identify and report on performance impacts and anomalies. Using machine learning, Dell AIOps analyzes historical performance data to determine the range of normal behavior and indicates when a metric is either above or below that range.

From the **Home** page scroll to the **Performance Impacts** tile. The Performance Impacts tile utilizes analytics across various performance KPIs such as IOPS, latency, bandwidth and queue length to identify systems with performance impacts and corresponding anomalies.

Looking at **Market Research**, we see that it has reported two performance impacts over the last 24 hours which are highlighted by the light pink areas in the Block Latency chart. We can also see that the second performance impact is accompanied by a performance anomaly which is identified by the blue shaded area.

Click the **"Market Research"** hyperlink

We have been directed to the **Performance** tab in the System Details page for the **Market Research** Unity array.

There are three charts that show us the highest performing storage objects on the system for the categories of **Latency, IOPS and Bandwidth**. They are sorted to show the objects with the highest averages over the last 24 hours. This allows us to immediately see the top contenders for storage resources on the system. The expectation would be that the highest-priority objects are getting the highest IOPS and Bandwidth while the lowest priority objects are experiencing the highest latency.

Now let's scroll down to the **Latency** chart to look at the performance impacts. This chart displays the measured latency in a blue line, the historic values in a shaded gray area, the time frame of the performance impact with light pink shading and the performance anomaly in a shaded blue area. The coloring is based on a time series correlation comparing 3 metrics: Latency, IOPs, and Bandwidth. Hovering the mouse over the chart will provide values for the timestamp, latency and historical seasonality. Each of the metrics is selectable so we can select or deselect the metrics based on what we want to display.

Tip! Another feature in Dell AIOps is the identification of Configuration Changes on the system. These are represented by the small blue rectangles located along the X-axis of the chart. This allows us to potentially correlate changes in the performance profile with recent configuration changes in the environment.

Select one of the configuration change boxes to bring up the Configuration Changes window.

The **Configuration Changes** window lists all of the configuration changes during that time frame. Click **"Close"**

We can now use the mouse to select a time range of interest within the chart. Dell AIOps will then show us the top contributors in that time range for the given metric.

In the following example, click and drag your mouse to select the time range around the performance impact with the latency anomaly. The left side of the chart now displays the time frame that is highlighted as well as two tabs: **Most Active and Best Match**. The **Most Active** tab displays the objects with the highest latencie during this time frame.

Select the **Best Match** tab to see those objects whose performance most closely matches the waveform to that of the selected time range on the chart.

We can see that **MR-Pool1_NAS-Datastore2** is the object that has been found to be the most closely matched to the selected time frame.

Scroll down to the **IOPS** chart and select the area in the chart where the IOPS dip and which corresponds to the second performance impact. We can again see that Dell AIOps displays the top storage objects in the left-hand side of the chart, this time, with respect to IO activity. There is again the ability to view the storage objects with the best matching waveform.

Now scroll down to the **Bandwidth** chart and again highlight the same time period in the chart. Dell AIOps supports the top object activity and best match for the Bandwidth chart as well.

Top Object Activity

This lesson shows how Dell AIOps can identify top performing virtual machine objects on shared resources.

Click **Home** Click the **search** icon.

Type **"MR\_"** in the Global Search window to find all virtual machines (and storage objects) containing this text.

Click **"MR_VM1"** from the listing of Virtual Machines.

Selecting **MR_VM1** brings us to the **Virtual Machine Details** page. The top of the window displays various properties for the Virtual Machine including its IP address, capacity information, ESXi host, and more. It also includes hyperlinks to the ESXi details page and the Dell AIOps Collector details page.

Click **"STORAGE PATHS"**

Click **"MR_Pool1_LUN1"**

We are directed to the **Properties** tab in the block object details page for **MR_Pool1_LUN1**. Click **"Performance"**

In this lesson, we will look at some additional functionality in the performance charts as it relates to VMware. We will investigate the second performance impact shown in the Block Latency chart.

Click and drag your mouse to highlight the area in the chart showing the performance anomaly that corresponds to second performance impact.

Notice how the left hand side of the chart now shows two tabs: **MOST ACTIVE** and **BEST MATCH**. The **MOST ACTIVE** column is selected by default and shows up to 5 virtual machines that have the highest latency during this time period that reside on this LUN. This allows us to identify other resources that are shared by this LUN during the performance impact. The second tab is the **BEST MATCH** tab and this tab lists the virtual machines whose performance shape most closely correlates to the shape during the selected time frame.

In the event that there are one or more performance impacts against the LUN, the Details button is available. Select the **second performance impact** from the date pull-down menu and then select the **Details** button

Click **"DETAILS"**

Select the **Possible Cause** pull down menu.

As we saw in the Performance Anomalies module, Dell AIOps identifies competing workloads on objects as the possible cause of the performance impact.

Select the **Contention** pull down menu.

For Unity systems, pools, and LUNs, Dell AIOps also identifies possible areas of resource contention. This includes analysis for SPs, Cache, Disk, and Ports.

Performance Impact by Date

Dell AIOps identifies the most likely candidates causing workload contention and the performance impact. If there are multiple performance impacts in the last 24 hours, we can use the date pull-down to choose which one to analyze.

Alert! Be aware that times and dates will differ.

Click the first date in this example. **Note:** The date will be different for you.

Click **"DETAILS"**

Selecting the Details button opens up a window showing the most likely possible competing workloads causing the performance impact as well as any identified areas of resource contention.

Let's first look at the competing workloads by selecting the pull down for **Possible Cause**.

In this example, we are reviewing the second performance impact and the potential cause of the performance impact is due to the IOPS of 3 different objects: **MR_Pool1_LUN2, MR_Pool1_NAS_Datastore1, and MR_Pool1_NAS_Datastore2**. Use the vertical scroll bar on the right-hand side of the window to see all of the potential objects.

Now we will look at possible resource contention by selecting the pull down for **Contention**.

Resource contention is currently supported for Unity systems. Dell AIOps can detect possible resource contention for SPs, Cache, Disk, and Ports. This example shows CPU Utilization on SPB and Port Errors on SPB Ethernet Port 1.

Congratulations! You have reached the end of this particular lesson. Click **Close** in the top left corner of this script to return to the demo guide.

Performance Impacts and Anomalies

Click the **"Market Research"** hyperlink

We have been directed to the **Performance** tab in the System Details page for the **Market Research** Unity array.

Select one of the configuration change boxes to bring up the Configuration Changes window.

The **Configuration Changes** window lists all of the configuration changes during that time frame. Click **"Close"**

We can now use the mouse to select a time range of interest within the chart. Dell AIOps will then show us the top contributors in that time range for the given metric.

Select the **Best Match** tab to see those objects whose performance most closely matches the waveform to that of the selected time range on the chart.

We can see that **MR-Pool1_NAS-Datastore2** is the object that has been found to be the most closely matched to the selected time frame.

Now scroll down to the **Bandwidth** chart and again highlight the same time period in the chart. Dell AIOps supports the top object activity and best match for the Bandwidth chart as well.

Top Object Activity

This lesson shows how Dell AIOps can identify top performing virtual machine objects on shared resources.

Click **Home** Click the **search** icon.

Type **"MR\_"** in the Global Search window to find all virtual machines (and storage objects) containing this text.

Click **"MR_VM1"** from the listing of Virtual Machines.

Click **"STORAGE PATHS"**

Click **"MR_Pool1_LUN1"**

We are directed to the **Properties** tab in the block object details page for **MR_Pool1_LUN1**. Click **"Performance"**

In this lesson, we will look at some additional functionality in the performance charts as it relates to VMware. We will investigate the second performance impact shown in the Block Latency chart.

Click and drag your mouse to highlight the area in the chart showing the performance anomaly that corresponds to second performance impact.

Click **"DETAILS"**

Select the **Possible Cause** pull down menu.

As we saw in the Performance Anomalies module, Dell AIOps identifies competing workloads on objects as the possible cause of the performance impact.

Select the **Contention** pull down menu.

For Unity systems, pools, and LUNs, Dell AIOps also identifies possible areas of resource contention. This includes analysis for SPs, Cache, Disk, and Ports.

Performance Impact by Date

Alert! Be aware that times and dates will differ.

Click the first date in this example. **Note:** The date will be different for you.

Click **"DETAILS"**

Selecting the Details button opens up a window showing the most likely possible competing workloads causing the performance impact as well as any identified areas of resource contention.

Let's first look at the competing workloads by selecting the pull down for **Possible Cause**.

Now we will look at possible resource contention by selecting the pull down for **Contention**.

Congratulations! You have reached the end of this particular lesson. Click **Close** in the top left corner of this script to return to the demo guide.