Always-on is a key design principle in the Optimaize SearchCluster architecture. Every part is redundant, no single point of failure.
The data store
SearchCluster stores the data within OpenSearch, an open-source, enterprise-grade search software, a continuation of Elasticsearch, with Apache Lucene at its core. This is a distributed network database where each record is stored multiple times on separate hardware.
Multiple nodes (hardware) are allowed to go offline without impact on the overall availability of the service. The database syncs and scales out.
Optimaize has been working with this data store for over 10 years, without downtime. Thousands of businesses use and stress-test every new feature and every new release of this underlying technology. A plethora of tools is available to work with.
Besides the redundancy within the life cluster, the snapshot and restore feature allows for a recovery from a complete failure, such as sabotage or human error.
The app nodes
The stateless matcher nodes, where record matching happens scale with demand and work load. Nodes can go online and offline as needed without affecting the service availability.
The hardware
The redundancy is in the network, not on a single machine. SearchCluster runs best on dedicated hardware with selected hardware components, namely locally attached SSDs. It also runs on common cloud, AWS, Azure.
The host centers
SearchCluster indices are available across multiple data centers and therefore they are still available in case of a complete data center outage.
Such host center outages have historically been caused by: network failures, router failures, router maintenance, DDoS attacks and circumvention, human error, electricity outage and failing generator startup, and burning down of a whole data center.
The API access points
The REST endpoints are accessible on separate, fully redundant URLs with different top level and second level domain, IP address, and SSL certificate.
<- Back to SearchCluster page.