vROPS multisite design using Federation MP
I have been working with on of my clients in UK where they are having classic multi site environment. Each site’s SDDC stack leverages capacity reports and efficiency parameters beautifully put up by vROPS but again the issue was to either go for a centralized vROPS setup with remote collectors af far sites (with respective challenges) or to go with local vROPS deployments (simpler but again respective challenges).
Recently, VMware had announced vRealize Operations Federation Management Pack you unify multi-site vRealize Operations Manager deployments into a single pane of glass. that will basically go with a simpler approach, vROPS local setups but with the capability of receiving key metrics for specified objects from other vRealize Operations Manager(s) deployment in the environment.
The vRealize Operations Federation Management Pack follows a simple server-client architecture. This management pack is installed on the vRealize Operations Manager instance, which has been identified as unified single pane of glass instance. It is highly recommended that this instance is only used for federation (not for data collection).
Once vRealize Operations Manager is deployed, the management pack configuration can be tweaked to add existing vRealize Operations Management instances to point to this federated instance.
Sizing Best Practices:
Out-of-the-box, vRealize Operations Federation Management Pack collects data for objects such as ESXi hosts, clusters, datacenters, datastores, and so on. Virtual machine data is not collected. Based on this, most of the environments can be monitored with one Extra Large Node.
The federation instance of vRealize Operations Manager should be used only for collecting data from the federation MP. No other MP on this vRealize Operations Manager instance should be configured to collect from any other data sources.
If you add new objects and metrics to an existing deployment of your federation vRealize Operations Manager instance, ensure to run the sizing calculator to meet the sizing requirements.
With vRealize Operations Federation Management Pack, you can either collect data directly from child vRealize Operations Manager instances or use a remote collector to optimize the collection process. The default collection cycle of vRealize Operations Federation Management Pack is 5 minutes. This time
interval works accurately for a centralized vRealize Operations Manager collecting data from other child vRealize Operations Manager instances that are located within a short distance. You can also use a remote collector for data collection. Otherwise, when centralized vRealize Operations Manager collects
data directly without a remote collector and latency is long, you can set the collection interval to 15 minutes. This ensures data collection without any gaps.
As most of the federation use cases are met by collecting data for objects other than virtual machines from your environment, it is recommended to avoid the collection of virtual machine data as this drastically increases the sizing requirement for the federation instance in a large-scale environment.