Abstract: A search server may receive a request to generate a set of indexes based at least in part on content associated with a user. The search server may generate a set of segments to divide the content. For a particular segment the search service then selects an indexing strategy to be used to generate an index for the particular segment. Other indexing strategies may be used to generate indexes for other segments. In addition, along with storing the index, the search server may store information indicating an association with the particular segment and the selected indexing strategy.

Abstract: Client applications that use an open session connection on a unidirectional communication channel, including a plurality of requests, can have such requests aggregated and transformed into individual requests for transmission over a bidirectional communication channel. In at least one embodiment of a proxy disclosed herein, an RPC stream is transformed into multiple individual HTTP requests. By using individual HTTP requests, such requests can be handled easily by existing HTTP infrastructure, such as proxies and HTTP servers. In another embodiment, a plurality of RPC requests can be embedded within an individual HTTP request so as to ensure that the RPC requests are not modified.

Abstract: In one embodiment, entries in a database journal in a non-interactive mode can be partially flushed to maintain database performance. In particular, data records that require decryption prior to flushing a database journal entry can be bypassed in non-interactive mode, while data records that need not be decrypted can be flushed. For example, entries that relate to a creation or deletion of a data record can be performed, while entries relating to modification of an existing record can be bypassed and processed during an interactive mode of operation. Other embodiments also include a transformation of the database journal using a filter so that multiple entries to a same record can be combined.

Abstract: In one embodiment, a system uses a two-layer deterministic function, such as a hashing mechanism, to ensure session stickiness or affinity. A client can generate a session key that can be used for at least two things. First, a load balancer can be selected by the client using the session key by using a consistent or deterministic hashing algorithm so that the same load balancer is selected from a set of weighted load balancers. After selecting the load balancer, the client can send the request to the selected load balancer. The load balancer then uses the same session key, which can be transmitted in an HTTP header, to select a backend host server computer. The result is that session affinity is maintained.