In my previous blog, I covered recent changes to NetApp® OnCommand® Unified Manager 7.2 that provide flexibility and alignment with security best practices. We made things simpler by having one server do the work of two. But what happens when a monitoring server is not available? In today’s blog, I answer that question by exploring approaches for high availability.
Let’s first look at what high availability means for a management and monitoring tool. There are several aspects here:
- Tolerance or survivability when a monitoring server fails because of an issue such as power supply failure. Running Unified Manager in a VM delegates that kind of issue to the hypervisor layer.
- Availability across multiple data centers—being able to continue to work in data center A even when connectivity to data center B is down.
- Data continuity—being able to collect metrics on a continuous basis with no data loss or holes.
When you deploy Unified Manager on physical hardware, on either Windows or Red Hat Enterprise Linux, you can depend on OS-level high-availability features such as Microsoft Cluster Server. MSCS allows you to run an active-passive configuration in which a standby server is available to take over with any failover. This configuration fits certain scenarios, such as when dedicated monitoring servers are required. This capability has been available for Unified Manager for some time. Red Hat Enterprise Linux customers can use Veritas Cluster Server.
You can deploy Unified Manager in a virtualized environment either by using our prepackaged VMWare OVA or by self-installing into a VM. Customers often consider other configurations as well. In a typical scenario, two or more data centers are involved, each operated independently. With Unified Manager 7.2, customers can choose to install two copies, one in each data center, and run in an active-active configuration. There is no explicit synchronization between the two instances, but both can be configured to monitor NetApp ONTAP® clusters throughout the network, providing resiliency to both hardware failures and network outages. It’s necessary to manually maintain the configuration in sync.
Some customers choose to deploy a dedicated virtualized environment for their monitoring tools. The whole virtualized environment may get replicated to an offsite backup location. Because Unified Manager can be configured to support multiple IP addresses, the DR site replica copy is accessible on an alternate IP address. This allows an active-passive configuration with manual failover, driven by whole monitoring environment failover. This configuration can support longer-term operations at the DR site, assuming that the primary Unified Manager is configured to monitor both the primary and DR sites.
Several high-availability deployment alternatives are available in Unified Manager 7.2 to provide options that support the DR policy. If you have questions about which deployment option is right for you, feel free to comment below or connect with a Unified Manager subject matter expert on the NetApp OnCommand community.
My next blog will look at what the team has done recently to improve usability and make Unified Manager easy and intuitive to use.