Xangati Performance Engine
The cornerstone of Xangati’s Management Dashboard (XMD) suite is its patented memory-driven engine that analyzes the health of each object in your cloud infrastructure across dozens of metrics in four microseconds.
This engine architected from the ground up for cloud scale can process over a million high-resolution performance metrics in a second whereas other solutions would take minutes to tens of minutes to handle the same amount of data.
This memory driven engine architecture is a foundational distinction between Xangati and any other performance management solution designed for the cloud. All other solutions have database-driven models that process much lower resolution performance metrics minutes after the fact. The result is that these other solutions essentially miss intermittent contention storms impinging on the performance of your complete environment. And even when other solutions do eventually achieve an alert state, it is minutes after the problem has dramatically impacted the entire user community. Xangati’s architectural distinction allows our users to leverage our automated insights to triage issues. This is enabled by the alert DVR recordings capability of the system.
The DVR recordings are effectively a visual representation of the second by second processing that the performance management engine does for every object. These recordings will precisely show the key metric that is being impacted e.g. storage latency but moreover the contention points that were driving the shift. And in many customer environments with these intermittent contention storms, the problems are multi-dimensional in nature i.e. at one point the issue is CPU contention, later it is storage and even later it might be network.
In addition to framing the problem within the context of these DVR recordings they can also be shared with different team members with different expertise. For instance, if the contention has been isolated to storage capacity then the recording can be passed over to the storage administrator who can more closely study the issue and make the appropriate adjustments to his part of the infrastructure.
Xangati’s Live Processing of Disparate Data Feeds
| Data Class | Source | Data Collection Frequency | Presentation Time Resolution |
| Server/hypervisor performance | vSphere API/vCenter | Pulls 3 groups of 20-second vCenter “real-time” metrics | 20 seconds—as granular as vCenter gathers in terms of metrics |
| Storage Activity/performance | vSphere API/vCenter | Pulls 3 groups of 20-second vCenter “real-time” metrics regarding datastore/LUN | 20 seconds—as granular as vCenter gathers in terms of metrics |
| Virtual Network Traffic Activity | Virtual distributed switches | Arbitrary rates | 1 second |
| Physical Network Traffic Activity | Physical switches and routers | Arbitrary rates | 1 second |
| Guest Performance and Processes | Windows Management Interface (WMI) | Every five seconds for metrics/every 20 seconds for TopN processes. | 5 seconds for metrics/20 seconds for processes |
| Network Latency | Xangati flow summarizer latency measurements and Cisco IP SLA | Every minute | 1 minute |
| PCoIP Session Statistics | VMware View Agent | Every 5 seconds | 1 second |
| HDX/ICA Session Statistics for XenDesktop | Citrix XenDesktop Agent | Every 5 seconds | 1 second |
