US 2026/0187157 A1Published application

Workload compression using large language models

AssigneeTeradata US, Inc.

InventorLouis M. Burger

Life of this patent

Estimating the term…

Loading patent…

Abstract

A system includes a storage device and a plurality of processing nodes in communication with the storage device. At least one processing node of the plurality of processing nodes receives query history. The query history contains information on a plurality of executed queries. The at least one processing node generates, using a large language model (“LLM”), a document for each query describing the information associated with a respective query. The at least one processing node divides generated documents into a plurality of groups. The at least one processing node compresses the documents to reduce a number of documents in each group until a predetermined threshold is reached. The at least one processing node stores a final compressed set of workload documents. The at least one processing node performs workload analysis using workloads described in the final compressed set of workload documents. A method and computer-readable medium are also disclosed.

Claims15

Dependent claims

A system comprising:

a storage device comprising one or more data storage facilities comprising computer-readable storage media;
a plurality of processing nodes in communication with the storage device, wherein each processing node includes one or more processors and memories comprising instructions executable by the one or more physical processors, and wherein at least one processing node from the plurality of processing nodes is configured to:
receive query history indicative of queries executed by a database management system, wherein the query history contains query telemetry collected in query logs and contains information on a plurality of executed queries;
generate a workload document for each query describing the information associated with a respective query based on the query telemetry wherein the workload document comprises a text-based representation of the query telemetry;
divide generated workload documents into a plurality of groups;
compress, with a large language model (“LLM”), the workload documents to reduce a number of documents in each group until a predetermined threshold is reached, wherein the predetermined threshold is a maximum amount of information that can be handled by the LLM, and wherein compression may include eliminating an amount of documents based on a set of parameters employed by the LLM;
store a final compressed set of workload documents;
and
perform workload analysis using workloads described in the final compressed set of workload documents.

6 more elements

A method comprising:

receiving, with a processor, query history indicative of queries executed by a database management system, wherein the query history contains query telemetry collected in query logs and contains information on a plurality of executed queries;
generating, with the processor, a workload document for each query describing the information associated with a respective query based on the query telemetry, wherein the workload document comprises a text-based representation of the query telemetry;
dividing, with the processor, generated workload documents into a plurality of groups;
compressing, with a large language model (“LLM”) executed by the processor, the workload documents to reduce a number of documents in each group until a predetermined threshold is reached, wherein the predetermined threshold is a maximum amount of information that can be handled by the LLM, and wherein compression may include eliminating an amount of documents based on a set of parameters employed by the LLM;
storing, with the processor, a final compressed set of workload documents;
and
performing, with the processor, workload analysis using workloads described in the final compressed set of workload documents.

4 more elements

A non-transitory computer-readable medium encoded with a plurality of instructions executable by a processor, the plurality of instructions comprising:

instructions to receive query history indicative of queries executed by a database management system, wherein the query history contains query telemetry collected in query logs and contains information on a plurality of executed queries;
instructions to generate a workload document for each query describing the information associated with a respective query based on the query telemetry, wherein the workload document comprises a text-based representation of the query telemetry;
instructions to divide generated workload documents into a plurality of groups;
instructions to compress, with a large language model (“LLM”), the workload documents to reduce a number of documents in each group until a predetermined threshold is reached, wherein the predetermined threshold is a maximum amount of information that can be handled by the LLM, and wherein compression may include eliminating an amount of documents based on a set of parameters employed by the LLM;
instructions to store a final compressed set of workload documents;
and
instructions to perform workload analysis using workloads described in the final compressed set of workload documents.

4 more elements

Dependencies are inferred from the claim text.