Parallel Querying

The K2 infrastructure supports parallel querying, in which a single search query can be distributed simultaneously to multiple collections.

In the case of a single K2 Server attached to multiple collections, the K2 Server simultaneously processes the request against each collection. The K2 Server then merges the results for display to the user.

In the case of a larger installation, separate K2 Servers may be involved. For example, consider a company that must provide fast and accurate search results across many collections in its large corporate intranet. Its documents reside in six separate collections: C1 through C6 (Figure 2-2). Assume that an application needs to search collections C3, C4, and C5. In that case, the K2 Broker distributes the search to the two appropriate K2 Servers. One server distributes the search to collections C3 and C4; the other server searches collection C5. All three searches are conducted simultaneously. Final merging of the results is performed by the K2 Broker.

Figure 2-2    Parallel querying, mirroring, and failover in a large deployment