By Ravi Prakash, Director Product Marketing
Why introduce a software integration for IBM SVC?
Figure 1 Hosts running applications access virtualized volumes exposed by an IBM SVC
- Many large customers of VirtualWisdom also use IBM SAN Volume Controller (SVC) to virtualize 3rd party storage arrays
- These customers install purpose-built SAN Performance Probes on SVC ports to monitor the response times of traffic from the hosts to the SVC versus response time from the SVC to the back-end storage array
- VirtualWisdom could monitor front-end SVC ports and monitor back-end connections and infer that if the back-end storage performance is fine but read and write performance out of the SVC was less than optimal, then the SVC controller may be busy
- If the response time for reads from hosts was acceptable but the response time for writes to the virtual disks of the SVC were delayed it could point to a cache issue on the SVC
- With this new SVC integration, we have one more piece of the puzzle to make our application centric infrastructure performance monitoring more compelling
Benefits of the new software integration for IBM SVC
- Previously if VirtualWisdom was inferring that the SVC controller was busy by using empirical data, now VirtualWisdom can pull an actual controller metric from the SVC to make a more precise determination
- Helps identify the performance impact of enabling compression on volume copies
- Helps identify the impact of compression tasks on the CPU of the SVC
- Helps identify “noisy neighbors” by pin pointing an SVC controller which may be struggling to keep up with write requests
Value of the software integration for IBM SVC to customers using VirtualWisdom for application-centric infrastructure performance monitoring:
- Expands the coverage of VirtualWisdom beyond the broad coverage that exists to date: Fibre channel SAN, wire-data on Fibre Channel and Ethernet, block storage, file storage, Software Defined Storage (VMware VSAN & Dell EMC VxFlex OS), host OS monitoring (Linux, Windows), hybrid-cloud (compute on AWS, Azure)
- The software integration for SVC introduces another set of data-points that VirtualWisdom can bring together for correlation and analysis using purpose-built analytics
- While the VirtualWisdom host OS integration can discover and provide insights on relationships between logical and physical volumes, the SVC integration shows that what the host perceives as a physical volume is in reality a virtual volume on the SVC which physically resides on these back-end storage devices.
- In addition to the host OS perspective of workload and latency we now have the SVC perspective of workload and latency. This is in addition to wire data collected by SAN performance probes from each data path.
- Click on the animated demo below to get an idea of how this integration works. You start with a pre-built report template for SVC performance, select an SVC node and navigate to its application-aware topology. Next, we drill down into one of the hosts accessing storage from the SVC node. Finally, we review the properties of one of the volumes exposed by the SVC and assigned to the host.
- With this new software integration, VirtualWisdom now monitors VMware metrics around CPU, network, disk utilization of a VM, OS metrics using host OS integration, wire data metrics from SAN performance probes, SVC metrics and VMAX metrics. VirtualWisdom now knows about queueing in the host OS, caching in the SVC, caching in the VMAX. This level of visibility when combined with wire data enables unique correlations and actionable insights by VirtualWisdom ensuring that SLAs for mission critical applications are being met.
- As part of this software integration for SVC, VirtualWisdom introduces several entities. There are physical entities (like SVC IO Group, SVC Node, SVC Node port) and logical entities (like SVC storage volume, SVC storage volume copy, SVC storage pool, SVC mdisk)
- Pre-built report templates like the one for “SVC port and cache performance” help you compare utilization of hosts connected to an SVC to the response time (both internal and external to the SVC) and cache health of the SVC.
Why manual correlation without VW’s SVC integration isn’t enough
- VirtualWisdom builds out the object model with SVC, I/O groups, nodes, mdisk, pools, volumes. In the absence of this software integration for SVC and with only a native element manager to monitor each SVC, you would have to build out their hierarchies manually starting with storage arrays, storage controllers, storage ports.
- In the absence of this integration for SVC, VirtualWisdom could monitor traffic on the wire to a LUN but would be unaware of the size of the LUN or the capacity of the volumes on which the LUN resides.
SVC integration complements the value of SAN Performance Probes for VirtualWisdom
- The software integration for SVC collects metrics at 1-minute granularity while a purpose-built SAN Performance Probe from Virtual Instruments monitors a link connected to a port on an SVC at 1 sec or 10 sec granularities. Hence the SVC integration complements the capabilities of the SAN Performance Probe which is the industry’s Gold standard for wire data monitoring.
Like what you hear and want to learn more about the new software integration for SVC? Give us a call!
Don’t forget to follow us on Twitter, LinkedIn and Facebook to stay up-to-date on the latest and greatest in app-centric infrastructure performance monitoring.