High Availability Overview

High Availability seeks to eliminate system downtime and data loss. High Availability provides a set of failure detection and failure recovery mechanisms that can be grouped into various availability-related policies by an IT administrator. The policies are available to application owners to manage the availability-related characteristics of their applications. Additionally, High Availability connects to workload management and workload placement.

Features include KVM-FT, Orchestration Proactive Monitoring, and awareness of infrastructure redundancies; for example, network paths, storage RAIDs, and others. The Stratus Cloud Solution leverages High Availability as protection from compute node failures, and ensures continuous workload operations. High Availability manages the cloud in a holistic method which includes network, storage, and other infrastructure aspects for maintaining your service levels. High Availability in a networking deployment helps mitigate the impact of individual node failures.

Availability Levels

The Stratus Cloud Solution provides three pre-defined High Availability levels:

• Commodity - Failed instances are not automatically restarted
• Business Critical - Failed instances are automatically restarted
• Mission Critical - Failed instances are automatically restarted

The availability levels can be changed at any time.

Support for the IT Administrator

To support the IT administrator, High Availability provides the ability to:

• Attach multiple custom matching tags to each cloud resource.
• Associate prices with the use of behavioral tags.
• Assign custom matching tags and behavioral tags to deployment packages and their individual entities.
• Supply predefined tag names for common infrastructure terms and locations; for example, data center, zone, floor, room, rack, equipment type, brand, model, serial number, country, state, province, city.
• Analyze, both statically and dynamically, the types of infrastructure required to support the set of defined and deployed applications.
• Analyze the reliability and availability of the infrastructure. For example, examine and compare the reliability of multiple brands of servers or switches.
• Examine the utilization of the infrastructure; for example, total resources available, total resources currently in use, total resources in standby, total resources offline for repair, total resources unused, and others.
• Export data used to generate various types of financial reports, such as costs, prices, utilization, uptime, downtime, and others. To assist with billing, the raw data is available in an easily managed form; for example, a comma-separated value spreadsheet.

Support for the Application Owner

To support the application owner, High Availability provides the ability to:

• Specify the desired availability level of a deployment package.
• Determine the cost of using a particular deployment package.
• Request automatic restart of a deployment package or components.
• Alert the application owner in the event an application tagged as restartable does not automatically restart.

High Availability Usage Scenarios

High Availability levels are based on the infrastructure components being monitored; for example, compute, network, and storage. High Availability capabilities apply to all hypervisors supported by OpenStack including KVM, ESX, and HyperV.

Usage scenarios for High Availability include:

• The IT administrator enables selectable availability levels for applications required to set the right levels of protection during specific time periods.
• Application owner deploys the applications with fault-tolerance from the service catalog.
• Application owner changes the level of availability required to meet new business needs.

Key benefits of Stratus Cloud Solution High Availability technology are:

• Selectable availability options through High Availability and fault-tolerant technologies.
• Efficient and dynamic use of resources – resources are not locked to a specific application, availability level, or tenant.
• Eliminates unnecessary infrastructure costs - maintaining multiple copies of data is no longer required.

High Availability images connect to shared storage and prohibit them from using direct-attached storage (DAS). Therefore, there is no waiting period for a failed hypervisor or failed DAS to return to service.