manual.opvizor
Multi-Tenant Environments
Performance Analyzer supports different architectures to monitor and separate multi-tenant environments. Typically, a customer wants to check the real performance of their own systems in an individual dashboard. As the responsible system administrator, you typically want all the details, while the customer should see the most important data – to not raise more questions than answered.
A tenant dashboard
Typical Multi-Tenant Setups
Depending on the goals you want to achieve, there are different kinds of multi-tenant setups. Sometimes, the customer owns the whole environment including his own VMware vCenter but outsourced the management; sometimes the vCenter is centralized and all tenants just use their part of the resources. Another important factor is the isolation of data and what data should be seen in what environments and by whom.
Many Tenants, Each Tenant Owns a VMware vCenter
In this case, each tenant runs a complete environment, including a VMware vCenter. The setup can look as follows.
Performance Analyzer Appliance 1 – Central Location
In the central location, the largest Performance Analyzer appliance will be deployed. All data of the central location as well as Tenant 1 and Tenant 2 location will be collected here.
This appliance is only configured to collect data from the central vCenter system. There is also a receiver listening for the data coming from the tenant appliances.
The dashboard that is accessible in the central location can be used to view all locations at once or separately, depending on the selection.
Performance Analyzer Appliance 2 – Tenant 1 Location
This appliance is configured to collect data of the local tenants VMware vCenter system, but sends the data to the local receiver as well as the central Performance Appliance receiver. Only one TCP port is used for communication and can be configured to the network needs.
Performance Analyzer Appliance 2 – Tenant 2 Location
Similar setup as Tenant 2. If there are different storage systems at the different locations or different applications running, the appliance configuration can differ. That is, Tenant 2 should also gather NetApp filer data as well as MS SQL data.
Many Tenants, One VMware vCenter Manages All Tenant Environments
One very common setup is, to have only central VMware vCenter server that manage the environments of different tenants and locations. The separation is done either by vCloud Director or simply by using roles and permissions.
In this case, our Performance Analyzer appliance can be configured in 3 different ways:
1) Use different Performance Analyzer appliances for each tenant using the same permissions as the tenant itself. That is the simplest possible setup and it ensures that the tenant user only sees its own data. The downside is, to run multiple Performance Analyzer appliances. This setup makes also sense, when no merged data should be accessible by the master operator.
2) Performance Analyzer as a multi-tenant appliance. Multiple data collectors are configured to collect data using different credentials (permissions) and send the data into different time series databases. Every tenant has its own user and organization. That ensure that the data is separated but can also be merged if needed (master operator).
3) Data is completely collected by one data collector, but the dashboards are configured to show separate data by tenant or user. This setup is pretty straightforward, but the data separation is only protected by the dashboards configuration and not by the time series database.
Different Performance Analyzer Components
Before we dig into details, it’s important to know that Performance Analyzer consists of different services (components) that can be completely separated, based on the architecture. All components can be separated to run on different appliances. That enables us to scale as well as to build efficient multi-tenant scenarios. We currently focus on time series data, but we can handle unstructured and configuration data as well. Just ask us, if you´re interested in learning more.
The data collector connects to the data source, for example, VMware vCenter API, NetApp API or MS SQL Systemdb, to gather all performance and statistical data in the configured crawling interval.
The data receiver listens to data and decides where to write the data
The Time Series Database gets feeded by the data receiver and handles the data retention as well as the possible queries on the data itself.
The Query Engine allows to query the time series data using simple and complex functions (math, transform, aso.)
The Graphical front-end, in our case Grafana, allows us to visualize the data based on the query engine that sits on top of the data.
All components need to work hand in hand and know about the data stream, metrics format and collected interval to make the visualization as great as you´ll like it.