BlueXP is now NetApp Console
Monitor and run hybrid cloud data services
Managing performance capacity and measuring performance utilization in storage systems is difficult. This has resulted in customers and storage administrators applying large margins of safety to ensure acceptable performance. Oncomand performance manager with its rich set of performance analytics for netapp storage systems delivers a new feature to aid in performance capacity management. The performance capacity used metric helps a user identify whether the performance capacity of a node or an aggregate is reaching a point where the performance will degrade if the workload increases. A customer or storage administrator can now refer to performance capacity metric to optimally utilize this NetApp storage system without having to apply large margins of safety to ensure acceptable performance. So let's look at a demonstration of how customers and storage administrators can leverage performance capacity metric to optimally utilize their NetApp storage system. We are at the node performance explorer page. Uh we have selected a particular time range and then we are looking at the events for this particular time range and then the latency and the utilization charts for this particular node in the cluster. We could observe that the latency which is at 6 milliseconds increases to around.9 milliseconds. Similarly the utilization also goes up say from 20% to actually around 40 50 and 60% um in this time range. Just by looking at the latency and utilization a user cannot determine if the current workload on this node can be increased or if new workloads can be added or to kind of find out if the node is being optimally utilized. So in order to help a user answer these questions, a performance manager provides a new feature which is called performance capacity. So performance capacity metric incorporates latency and utilization in a storage resource into a single performance metric. So this metric helps a user identify whether the performance capacity of a node is reaching a point where the performance will degrade if the workload increases. Okay. The zoom view will enable us to view the zoom view will enable us to zoom in to a granular time range. We can zoom into a granular time range and also we can apply the event as well as if we have defined any thresholds even those thresholds can be applied here. Okay. The performance capacity used chart includes color bands. If the performance capacity values are in the green band this indicate that the node is currently underutilized. If the values in the yellow band indicate that the performance capacity or the node utilization is in the optimal range and red indicates that the node is actually being overutilized where the workload will now start experiencing increase in latencies. Okay. So let's go back to the performance capacity and the latency charts.So in this example I've actually increased uh the workload to kind of show that as the usage as a performance capacity on the node gets used. You can also see that there is an increase in latency from this particular node standpoint over the selected time range. This node is operating in the green band which means that this node can actually accommodate an increase in current workloads or you can also add a new workloadto further analyze this. Performance manager provides us with a performance capacity used advanced chart.This chart breaks down the performance capacity usage based on user protocol workload and then the background processes. So what do I mean by background system processes? Rate reconstruct, disc rubbing, dduplication, snap mirror, these are examples of background processes. So the advanced chart breaks down the performance capacity used on this node into user based workload and then the background. It also gives you a sense of how much performance capacity is free. You can see that as uh the workload increases the capacity which is free for you to add new workloads or extend your current workload also goes down. that the performance capacity usage on this node is actually driven by user protocol workload. Okay. So in order to help you understand and how many IOPS are available on this node uh we also provide the available IOPS chart. The available IOPS is an absolute metric. It indicates how many IOPS are left in the resource for use. this example for this time range uh based on the usage this node has IOPS available ranging from almost 200 and going up uh to almost 15,000. So do note that as the performance capacity used uh on this node goes down the available IOPS to add new workload or increase the current workload kind of goes up. A user can actually select different time ranges to understand the storage resource utilization over a month, week, etc. So using the performance capacity chart and using the available IOPS chart, a user can assertain the performance capacity available on this node. Performance manager actually goes beyond and allows a user to add and compare performance statistics either from other nodes in the cluster or compare them with aggregates. For this example, let's understand if workloads are equally distributed across both the nodes in the cluster. So I can just click on add here and let's go down for the selected time range. The performance capacity used on both nodes in the cluster can be viewed from the performance capacity used chart. So here itappears that the workload is actually distributed equally across both the nodes in the cluster. In a scenario where uh where the performance capacity used on both the nodes varies significantly, we can investigate further to if aggregates or volumes or workloads can be moved non-disruptively.Also looking at the available IOPS chart, we can find that the node one is a better candidate to place a new workload compared to node two given that the IOPS available are high. So performance capacity used values are also displayed on the dashboard and from the node or the aggregate inventory page. Average values uh which are based on hourly samples from the previous 72 hours are also displayed. The top performance page displays node and aggregates with the highest or the lowest performance capacity used values. So in this demonstration we looked at how customers or storage administrators can leverage performance capacity metric to optimally utilize the NetApp storage systems. Thank you for watching this video. Please log on to netapp.com and navigate to oncommand performance manager for more
ONTAP 9 and OnCommand Performance Manager 7.0 deliver a new feature to assist with performance capacity management. This demo shows how performance capacity information is used to help make decisions about provisioning and move operations.