Software virtual machine for data ingestion
View Patent ↗In general, this disclosure is directed to a software virtual machine that provides high-performance transactional data acceleration optimized for multi-core computing platforms. The virtual machine utilizes an underlying parallelization engine that seeks to maximize the efficiencies of multi-core computing platforms to provide a highly scalable, high performance (lowest latency), virtual machine. In some embodiments, the virtual machine may be viewed as an in-memory virtual machine with an ability in its operational state to self organize and self seek, in real time, available memory work boundaries to automatically optimize maximum available throughput for data processing acceleration and content delivery of massive amounts of data.
1. A device comprising:
a multi-core hardware processor having a plurality of execution cores;
an in-memory database comprising data stored within a plurality of memory partitions, wherein each of the memory partitions is associated with a different one of the execution cores;
a data ingest engine that establishes a plurality of concurrent network connections to external devices to receive incoming transactions; and
a parallelization engine that decomposes incoming transactions for the in-memory database into sub-transactions, deploys the sub-transactions to the execution cores associated with the respective memory partitions that store data for the sub-transactions, and concurrently executes the sub-transactions of the incoming transactions on the respective execution cores to which the sub-transactions are deployed, wherein the incoming transactions include a first transaction and a second transaction, wherein the first transaction and the second transaction include first data and second data, respectively, for insertion to the in-memory database,
wherein the parallelization engine comprises a first compiler task and a second compiler task that execute concurrently to decompose, respectively, the first transaction into a first set of sub-transactions and the second transaction into a second set of sub-transactions;
a first statement execution task that applies an assignment algorithm to the first data to identify a first one of the execution cores to execute one of the first set of sub-transactions; and
a second statement execution task that applies the assignment algorithm to the second data to identify a second one of the execution cores to execute one of the second set of sub-transactions,
wherein the one of the first set of sub-transactions and the one of the second set of sub-transactions execute concurrently on the first one of the execution cores and the second one of the execution cores, respectively.
2. The device of claim 1 , further comprising: a plurality of virtual machines that each operate in accordance with a virtual instruction set, wherein each of the virtual machines executes on a different one of the execution cores,
wherein the first statement execution task executes on one of the virtual machines to spawn and deploy, for each of the first set of sub-transactions, a sub-execution task to execute the sub-transaction on the identified one of the execution cores for the sub-transaction.
3. The device of claim 2 ,
wherein each of the virtual machines comprises a scheduler for the corresponding execution core, and
wherein each of the schedulers maintains a separate run list data structure that stores a reference to tasks scheduled for execution by the corresponding one of the plurality of execution cores.
4. The device of claim 2 ,
wherein each of the virtual machines includes a cross-queue data structure that stores references to tasks migrated to the virtual machine by a scheduler of a different one of the virtual machines.
5. The device of claim 1 ,
wherein the first statement execution task spawns a sub-execution task for each of the first set of sub-transactions; and
wherein the second statement execution task spawns a sub-execution task for each of the second set of sub-transactions,
wherein the sub-execution task for each of the first set of sub-transactions and the sub-execution task for each of the second set of sub-transactions execute concurrently on different ones of the execution cores.
6. The device of claim 1 , further comprising:
a content delivery engine that directs the parallelization engine to execute an aggregate transaction to aggregate respective data received in the incoming transactions,
wherein the content delivery engine outputs the data aggregated by the parallelization engine.
7. The device of claim 6 , wherein the parallelization engine decomposes the aggregate transaction into aggregate sub-transactions, the device further comprising:
a plurality of virtual machines that each operate in accordance with a virtual instruction set, wherein each of the virtual machines executes on a different one of the execution cores,
wherein the first statement execution task executes on one of the virtual machines to apply the assignment algorithm to the aggregate sub-transactions to identify the execution cores associated with the respective memory partitions that store respective data for the aggregate sub-transactions,
wherein the first statement execution task spawns and deploys, for each of the aggregate sub-transactions, a sub-execution task to execute the aggregate sub-transaction on the identified one of the execution cores for the aggregate sub-transaction,
wherein the sub-execution task for each of the aggregate sub-transactions executes concurrently to retrieve the respective data for the aggregate sub-transactions, and
wherein the first statement execution task receives and aggregates the respective data and sends the aggregated data to the content delivery engine.
8. A method comprising:
establishing a plurality of concurrent network connections to external devices with a data ingest engine;
receiving incoming transactions with the data ingest engine by the concurrent network connections;
decomposing, with a parallelization engine executing on a multi-core hardware processor having a plurality of execution cores, incoming transactions for an in-memory database into sub-transactions, wherein the in-memory database comprises data stored within a plurality of memory partitions, wherein each of the memory partitions is associated with a different one of the execution cores;
deploying, with the parallelization engine, the sub-transactions to the execution cores associated with respective memory partitions that store data for the sub-transactions; and
concurrently executing the sub-transactions of the incoming transactions on the respective execution cores to which the sub-transactions are deployed, wherein the incoming transactions include a first transaction and a second transaction, wherein the first transaction and the second transaction include first data and second data, respectively, for insertion to the in-memory database;
concurrently executing a first compiler task and a second compiler task of the parallelization engine in parallel to decompose, respectively, the first transaction into a first set of sub-transactions and the second transaction into a second set of sub-transactions;
executing a first statement execution task that applies an assignment algorithm to the first data to identify a first one of the execution cores to execute one of the first set of sub-transactions;
executing a second statement execution task that applies the assignment algorithm to the second data to identify a second one of the execution cores to execute one of the second set of sub-transactions; and
executing the one of the first set of sub-transactions and the one of the second set of sub-transactions concurrently on the first one of the execution cores and the second one of the execution cores, respectively.
9. The method of claim 8 , further comprising:
executing a plurality of virtual machines that each operate in accordance with a virtual instruction set, wherein each of the virtual machines executes on a different one of a plurality of execution cores of the multi-core hardware processor,
wherein the first statement execution task executes on one of the virtual machines to spawn and deploy, for each of the first set of sub-transactions, a sub-execution task to execute the sub-transaction on the identified one of the execution cores for the sub-transaction.
10. The method of claim 9 ,
wherein each of the virtual machines comprises a scheduler for the corresponding execution core, and
wherein each of the schedulers maintains a separate run list data structure that stores a reference to tasks scheduled for execution by the corresponding one of the plurality of execution cores.
11. The method of claim 10 ,
wherein each of the virtual machines includes a cross-queue data structure that stores references to tasks migrated to the virtual machine by a scheduler of a different one of the virtual machines.
12. The method of claim 8 , further comprising:
spawning, with the first statement execution task, a sub-execution task for each of the first set of sub-transactions; and
spawning, with the second statement execution task, a sub-execution task for each of the second set of sub-transactions,
wherein the sub-execution task for each of the first set of sub-transactions and the sub-execution task for each of the second set of sub-transactions execute concurrently on different ones of the execution cores.
13. The method of claim 8 , further comprising:
directing, with a content delivery engine, the parallelization engine to execute an aggregate transaction to aggregate respective data received in the incoming transactions;
outputting, with the content delivery engine, the data aggregated by the parallelization engine.
14. The method of claim 13 , wherein the parallelization engine decomposes the aggregate transaction into aggregate sub-transactions, the method further comprising:
executing a plurality of virtual machines that each operate in accordance with a virtual instruction set, wherein each of the virtual machines executes on a different one of a plurality of execution cores of the multi-core hardware processor;
executing the statement execution task on one of the virtual machines to apply an assignment algorithm to the aggregate sub-transactions to identify the execution cores associated with the respective memory partitions that store respective data for the aggregate sub-transactions,
wherein the first statement execution task spawns and deploys, for each of the aggregate sub-transactions, a sub-execution task to execute the aggregate sub-transaction on the identified one of the execution cores for the aggregate sub-transaction,
wherein the sub-execution task for each of the aggregate sub-transactions executes concurrently to retrieve the respective data for the aggregate sub-transactions, and
wherein the first statement execution task receives and aggregates the respective data and sends the aggregated data to the content delivery engine.
15. A computer-readable storage media comprising instructions that, when executed, cause a multi-core hardware processor having a plurality of execution cores to:
establish a plurality of concurrent network connections to external devices with a data ingest engine;
receive incoming transactions with the data ingest engine by the concurrent network connections;
decompose, with a parallelization engine, incoming transactions for an in-memory database into sub-transactions, wherein the in-memory database comprises data stored within a plurality of memory partitions, wherein each of the memory partitions is associated with a different one of the execution cores;
deploy, with the parallelization engine, the sub-transactions to the execution cores associated with respective memory partitions that store data for the sub-transactions; and
concurrently execute the sub-transactions of the incoming transactions on the respective execution cores to which the sub-transactions are deployed, wherein the incoming transactions include a first transaction and a second transaction, wherein the first transaction and the second transaction include first data and second data, respectively, for insertion to the in-memory database;
concurrently execute a first compiler task and a second compiler task of the parallelization engine to decompose, respectively, the first transaction into a first set of sub-transactions and the second transaction into a second set of sub-transactions;
execute a first statement execution task that applies an assignment algorithm to the first data to identify a first one of the execution cores to execute one of the first set of sub-transactions;
execute a second statement execution task that applies the assignment algorithm to the second data to identify a second one of the execution cores to execute one of the second set of sub-transactions; and
execute the one of the first set of sub-transactions and the one of the second set of sub-transactions concurrently on the first one of the execution cores and the second one of the execution cores, respectively.
16. The computer-readable storage media of claim 15 , wherein the instructions, when executed, further cause the multi-core hardware processor to: execute a plurality of virtual machines that each operate in accordance with a virtual instruction set, wherein each of the virtual machines executes on a different one of a plurality of execution cores of the multi-core hardware processor,
wherein the first statement execution task executes on one of the virtual machines to spawn and deploy, for each of the first set of sub-transactions, a sub-execution task to execute the sub-transaction on the identified one of the execution cores for the sub-transaction.