Arista Networks, Inc.

Oracle International Corporation

Stephen Schleimer

Michael Greenwald

Derek Taylor

Ryder B Rishel

Daniel Greene

A network device includes a database that includes database entries, an external data that includes external data entries, an interpreter, and a standing query instance (SQI). The interpreter is programmed to convert an external data entry into a temporary database entry and provide the temporary database entry to the database. The SQI is programmed to obtain a database entry and the temporary database entry from the database and generate an output based on the database entry and the temporary database entry.

External data source linking to queries in memory

A network device includes a database that includes a number of tables, an observer, and a query service (QS) manager. The observer monitors an element of a first table of the number of tables and provide a value of the element to a query instance of a number of query instances. The QS manager generate the number of query instances. The query instance of the number of query instances executes a compiled query in response to receiving the value.

Standing queries in memory

A network device includes a standing query instance (SQI) and a database. The SQI is programmed to generate an output stored in a low resource storage table of a database. The database is programmed to monitor a storage capacity of the low resource storage table based on at least, in part, the output stored in the low resource storage table; determine, based on the monitoring, when the storage capacity reaches a predetermined amount; and disable an upstream element in response to the determination.

Query result flow control in a network switch

A network device includes a query service instance (QSI), a database, and an observer. The QSI is programmed to generate an output based on a, at least in part, a first table of a database; send the output to the database for storage; and send a query completion indicator to the database. The database is programmed to store the output in a second table of the database. The observer is programmed to obtain the query completion indicator from the database and notify an entity that the QSI is complete in response to obtaining the query completion indicator.

Use of null rows to indicate the end of a one-shot query in network switch

A network device includes a database that includes a materialized table and a non-materialized table. The network device also includes a standing query instance that generates an output that is stored in the non-materialized table. The database obtains a request that specifies the output, and provides the output in response to the request.

Non-materialized tables with standing queries

A network device includes an external subscription table, an interpreter, and a state machine. The external subscription table includes a subscription associated with a client. The interpreter is programmed to obtain the subscription; generate an entity, based on the subscription, that produces an output; and notify the client of the output. The state machine is programmed to send the output to the client based on the subscription.

Interface for a client of a network device

A method for operating a network element includes obtaining a write request that specifies a variable length data. The method includes identifying a row of a table based on the write request. The method includes processing the row to identify an empty portion of a variable length data storage portion of the row. The method includes determining an offset that that specifies the location of the empty portion. The method includes storing the offset and a length of the variable length data in a fixed length storage element of the fixed length data storage portion. The method includes storing the variable length data in the empty portion of the variable length data storage portion.

Packed row representation for efficient network serialization with direct column indexing in a network switch

Methods and systems are described for applying the use of shards within a single memory address space. A database request is processed by providing the request from a client to a processor, the processor then distributing the request to multiple threads within a single process but executing in a shared memory address environment, wherein each thread performs the request on a distinct shard, and aggregating the results of the multiple threads being aggregated and returning a final result to the client. By parallelizing operations in this way, the request response time can be reduced and the total amount of communication overhead can be reduced.

Growing community of inventors

Campbell, CA, United States of America

Stephen Schleimer