Foglight for Storage Management Shared 4.6.5

Exploring Connectivity with SAN Topology Diagrams

SAN Topology diagrams map the connections through the virtual environment (VMs-to-virtualization storage-to-extents) to the resources in the SAN (LUNs and NASVolumes). Topology diagrams hide port-level details to focus on connectivity. Using this view, you can see which resources have an alarm status along each connection path, and begin to form hypotheses about whether or not the SAN environment is contributing to performance issues in the virtual environment. You can then drill down on resources to test those hypotheses.

In a topology diagram, you can view individual port connections by clicking one or more cloud icons. Details are shown in a separate window.


	TIP: If your primary goal is to assess the performance of the ports that are used by the datastores and disk extents connecting to the SAN, use the SAN Data Paths tab instead (when available). For more information, see Exploring I/O Performance with SAN Data Paths.

The SAN Topology view displays as a graph for a VM, Datastore or CSV, and for a storage LUN or NASVolume.

The SAN Topology view for a Cluster or server or host will offer the following choices for viewing the topology:

•

Show Datastore or CSV resources used by VMs as graph

•

Show RDM or pass-through disk resources used by VMs as graph

•

Show storage resources used by VMs as table

•

Show all storage resources as table

The option to show all storage resources as a table will include Datastores or CSVs that are not used by VMs.

The following workflow explains how you can verify connectivity and the status of entities and storage devices in your infrastructure using a topology diagram. This procedure assumes that you navigated to a topology view from a Virtualization Explorer dashboard (see Introducing the Virtualization Dashboards) or from a Storage Explorer component dashboard (see Investigating a LUN or Investigating a NASVolume).

To use a topology diagram graph:

Click on a Datastore or CSV icon to see the VMs that use this storage. Click on a cloud icon to see the extents that map to the LUN.


	NOTE: When viewing the graph for a Cluster, the VMs and disk extents are not displayed on the main graph. When viewing the graph for a server or host, if it supports large numbers of VMs, as in a VDI environment, the VMs will not display on the main graph.

The following image shows virtual machines connected to separate datastores. The disk extents are identified and linked through the network to LUNs. For descriptions of icons, see Summary of Icons Used in Topology Diagrams.

When a topology diagram has a very large number of elements to be displayed, the graphical representation may be changed to a table for clarity. If the graph is too large to display in the viewing area, you can use the browser’s scroll bars and the Navigator to view the hidden areas. The Navigator enables you to scale the topology diagram and, in large diagrams, drag the viewing area to a different part of the diagram.

Click a Datastore or CSV icon to see the VMs that use this storage.

Click any icon to review key details and metrics about the entity.

Click a Cloud

icon to display the ports and SAN paths from the logical storage to the LUN, through the host or server.

In this example, you can see that the LUN’s controller has a Critical status.

To investigate an entity, click its icon. For example:

Click the controller

icon.

To view the alarm message, click the alarm count (opens an Alarm Summary window) or the controller name (opens the Controller dashboard).

The Alarm Summary reports the cause.

In some topology diagrams, you may see an inferred host (indicated with an orange line). On these diagrams, if an inferred-host-to-port assignment is incorrect, you can click the host port icon and click the appropriate action to correct it.

To use a topology diagram table:

Click on a VM cell to display the VMs that use this Datastore or CSV, or RDM. Click on a cloud icon to display the ports and SAN paths from the storage to the LUN, through the hosts or servers.


	NOTE: The table view provides the same information as the graphical view.

Exploring I/O Performance with SAN Data Paths

The SAN Data Paths tab focuses on input/output performance of the ports used in the data paths connecting the virtual environment to the SAN. This view combines an I/O Data Paths table with a topology diagram that includes port details and I/O performance metrics. The table displays the worst-performing path segment of the possible data paths between each disk extent and the LUN during the time period, helping you to identify bottlenecks resulting in high latency. The diagram enables you see the entities that are capable of doing I/O through each path segment, corresponding to the table rows.

Because the metric values in the SAN Data Paths tab represent the average values over the time period, consider selecting a shorter time range (one hour or less) to help investigate performance spikes. For example, if a customer is complaining about latency issues in the last 15 minutes, you may want to set the dashboard time range to the last 15-30 minutes to focus on the problem period.


	TIP: If your primary goal is to discover which SAN resources are used by entities in your virtual environment, use the SAN Topology tab instead. For more information, see Exploring Connectivity with SAN Topology Diagrams.

This workflow walks you through using the SAN Data Paths tab from the VMware Explorer’s ESX Host dashboard. The content of the SAN Data Paths tab may be slightly different on the Virtual Machine, Datastore, and LUN dashboards, but the flow is the same. The workflow for Hyper-V servers, VMs, and CSVs is similar, but uses the Hyper-V terminology. This workflow continues from Introducing the Virtualization Dashboards.

To review data paths:

On the SAN Data Paths tab, review the In Use I/O Paths table to see the worst-performing path segments in terms of Latency.

The color of the bars (in the table and in the diagram) highlights data values that fall within various thresholds. Metrics without thresholds are displayed in dark blue.

The columns display the following details and metrics. On the Virtual Machine, Datastore, and LUN dashboards, the details and metrics may be slightly different.

•

Datastore/Disk Extent. List of datastores and the disks that they use, ordered so that the datastores with high-latency disk extents appear at the top. Datastores configured from a NASVolume show only the associated volume; no other data is available. If an RDM or Other node is displayed, the disk extents under this node are RDMs providing storage directly to the virtual machine.

•

Latency. Average latency per operation.


	NOTE: When physical hosts are inferred, latency values are unavailable because the disk extents that map to the selected LUN are unknown. For more information, see Inferring Physical-Host-to-Storage Relationships.

•

Data Rate. Average data rate for I/O from the ESX or VM to the LUN.

•

ESX FC Ports --> SAN Util. Displays the busiest link (read or write utilization) in the possible paths between the ESX and the FC switches. Click the cell to display all the port links. Review the topology diagram to see the ports and link utilization. Data is not available for IP ports.

•

SAN --> A/F Ports Util. Displays the busiest link (read or write utilization) in the possible paths between the FC switches and the array/filer ports. Click the cell to display all the port links. Review the topology diagram to see the ports and link utilization. Data is not available for IP ports.

•

A/F Ctrl Busy. Displays the CPU % Busy metric for the busiest controller in the data path for a storage array or filer. % Busy values are not available on some devices.

•

LUN / NASVolume / Dir. Displays the LUN that is mapped to the extent, or displays the NASVolume (filers) or directory (Isilon arrays) providing the storage for a datastore.

•

% Competing I/O at LUN. Displays the percentage of I/O being experienced by this LUN for all VMs accessing the Datastore, not just those in this ESX. Click the cell to display the top five VMs doing I/O to this LUN.

•

LUN State. Reports on the state of the LUN as follows:

Table 6. LUN states
	Indicates that the LUN is reporting activity that gives an indication of performance problems, it is currently degraded or rebuilding, or the % Busy or Latency metrics are over their thresholds for the time period. Dwell on the cell for details.
	Indicates that the vendor does not provide % Busy or Latency metrics.
	Indicates that either the % Busy metric or Latency metric is within normal range during the time period, and the LUN is not reporting that it is currently degraded or rebuilding. Dwell on the cell for details.

•

Latency (ms). Average latency per operation to the LUN during the time period.

In the diagram, ensure that Select Desired View is set to Show full data paths.

The diagram shows the data paths between datastores and the LUNs providing the physical storage within an ESX host. The bars shown below the icons summarize average latency for hosts (if available), link utilization for FC ports, and percentage busy for controllers. For descriptions of icons, see Summary of Icons Used in Topology Diagrams.

This diagram highlights how ports are often shared by multiple disk extents. You may be able to trace the cause of a poorly-performing disk extent to poor link utilization values or to some resource issue further down the data path in the SAN. Or if all SAN ports and resources are in a Normal status, you may be able to rule out a storage issue.

Review the data paths for the worst performing entities to identify bottlenecks that may be contributing to the high latency values.

To see more of the diagram, use the Navigator to shrink the diagram or move the viewing area. For more information, see Exploring Connectivity with SAN Topology Diagrams.


	TIP: You can also close the I/O Paths table using the arrow on the far right of the table title bar.

To investigate an entity, click its icon or name. Use the breadcrumbs to return to this view.

If you find that the colors used in some metric bars do not reflect acceptable thresholds in your environment, you can ask your Foglight for Storage Management Administrator to edit the threshold values in the following registry variables:

•

Latency: VMW:diskTotalLatency.[warning|critical|fatal]

•

Latency: HPV:diskTotalLatency.[Warning, Critical, Fatal]

•

Port utilization: StSAN.FCSwitchPort.Utilization.[Warning|Critical|Fatal]

•

A/F Ctrl Busy: StSAN.Controller.PctBusyThreshold.[Warning|Critical]


	NOTE: Registry variables are global variables that are often referenced by rules. Before editing a registry variable, ensure that the edit will not cause unintended changes to how the affected rules trigger alarms. For more information, search for “Registry Variable” in the online help.