Netezza, as an appliance, is equipped with essential failover components to ensure smooth operation even in the face of hardware issues. This architecture guarantees an availability rate exceeding 99.99%. Two hosts constitute a cluster in every Netezza appliance, enabling seamless takeover if one host encounters a failure.
Netezza leverages Linux-HA (High Availability) and Distributed Replicated Block Device for managing the host cluster and data mirroring between these hosts respectively.
Regarding data storage, one-third of each disk in the array houses the primary copy of user data. Another third mirrors the primary data from a different disk, while the remaining third serves as temporary storage.
In the event of a disk failure, the mirror copy will be utilized, and the SPU (Service Processing Unit) attached to the faulty disk will be updated with the disk containing the mirror copy. If a disk track fails, the track will be marked as invalid, and valid data will be copied from the mirror copy onto a new track.
Should any issues arise with an S-Blade, the workload will be reassigned to other available S-Blades.
All failures will be notified based on event monitors defined and enabled. Similar to the dual host setup for high availability, the appliance also features dual power systems. All connections between components like host to SPA (Service Processing Array) and SPA to disk array also have a secondary.
Any hardware-related issues can be monitored using the NZAdmin GUI tool.
In this article, we will delve into the concepts of failover and high availability in the context of Netezza, a high-performance data warehouse appliance. We will explore how Netezza ensures uninterrupted service by implementing a robust failover architecture.
The Netezza architecture consists of one or more nodes, each with its own storage and processing capabilities. The system is designed to be highly parallel, leveraging massively parallel processing (MPP) technology.
Netezza employs an active-active/active-passive high availability configuration. In the active-active model, all nodes are actively serving queries. In case of failure of a node, the workload is automatically redistributed among the remaining active nodes.
In an active-passive configuration, one node acts as the active node while the others are passive. The passive nodes can take over the role of the active node in case of failure. This is achieved through a process called "standby switchover."
Standby switchover involves transferring the database connection from the failed active node to a standby (passive) node. The process ensures minimal downtime and guarantees service continuity.
In addition to failover, Netezza also provides features that enhance high availability, such as:
By understanding Netezza's failover and high availability architecture, we can appreciate how this powerful data warehouse appliance ensures uninterrupted service and maximizes uptime. With its robust design and intelligent failover mechanisms, Netezza stands as a reliable solution for mission-critical data warehousing needs.