Abstract: A Domain Name System (DNS) server that is operable to apply DNS-based load balancing policies based on geolocation and time of day is described herein. In an embodiment, the DNS server determines a geolocation and a time of day associated with a plurality of DNS queries and then obtains a domain name resolution policy that is associated with one or more of the geolocation and the time of day. The policy specifies a plurality of geographically disparate data centers and a ratio by which domain names are to be resolved thereto. The ratio at least specifies that more domain names are to be resolved to one of the plurality of geographically disparate data centers than another. The DNS server applies the domain name resolution policy to resolve the domain names in the DNS queries.

Abstract: Embodiments are directed to establishing caches that provide authoritative domain name system (DNS) answers to DNS requests. In one scenario, a computer system establishes a cache that stores authoritative DNS answers to DNS queries. The cache corresponds to a specified DNS zone that includes authoritative DNS answers for a subset of DNS queries. The cache is configured to store the authoritative DNS answers for at least a specified period of time during which the authoritative DNS answers are updatable. The cache then receives an update indicating that at least one cached DNS answer is out-of-date and the computer system purges the out-of-date DNS answer from the cache, ensuring that the cache continually provides authoritative DNS answers for DNS queries assigned to the specified DNS zone.

Abstract: Techniques for parallel processing of directory searches are provided. A directory search is received on a multicore or multiprocessor architecture machine. At least two threads processing on different cores or processors of the machine handle different aspects of the directory search. A first thread identifies matches in the directory for the search and serially populates the matches to search results. The second thread serially and in parallel grabs the matches in the search results and filters some out and modifies other ones of them to complete processing on the search. The search results are then returned to an initial requester making the initial search.

Abstract: Techniques for parallel processing of directory searches are provided. A directory search is received on a multicore or multiprocessor architecture machine. At least two threads processing on different cores or processors of the machine handle different aspects of the directory search. A first thread identifies matches in the directory for the search and serially populates the matches to search results. The second thread serially and in parallel grabs the matches in the search results and filters some out and modifies other ones of them to complete processing on the search. The search results are then returned to an initial requester making the initial search.