Scalable analysis platform for semi-structured data

Patent metadata

Field	Value
Publication number	US-9613068-B2
Application number	US-201414213941-A
Country	US
Kind code	B2
Filing date	Mar 14, 2014
Priority date	Mar 15, 2013
Publication date	Apr 4, 2017
Grant date	Apr 4, 2017

Abstract

A data transformation system includes a schema inference module and an export module. The schema inference module is configured to dynamically create a cumulative schema for objects retrieved from a first data source. Each of the retrieved objects includes (i) data and (ii) metadata describing the data. Dynamically creating the cumulative schema includes, for each object of the retrieved objects, (i) inferring a schema from the object and (ii) selectively updating the cumulative schema to describe the object according to the inferred schema. The export module is configured to output the data of the retrieved objects to a data destination system according to the cumulative schema.

First claim

The invention claimed is:

1. A data transformation system comprising:
one or more computing devices comprising one or more hardware processors and memory and configured to implement:
a schema inference module configured to
dynamically create a cumulative schema for objects retrieved from a first data source, wherein:
each of the retrieved objects includes (i) data and (ii) metadata describing the data; and
dynamically creating the cumulative schema includes, for each object of the retrieved objects, (i) inferring a schema from the object and (ii) selectively updating the cumulative schema to describe the object according to the inferred schema;

collect statistics on the data types of the retrieved objects; and
based on the statistics on the data types, determine whether the data of the retrieved objects is typed correctly; and

an export module configured to output the data of the retrieved objects to a data destination system according to the cumulative schema.

2. The data transformation system of claim 1 , wherein the data destination system includes a data warehouse.

3. The data transformation system of claim 2 , wherein the data warehouse stores relational data.

4. The data transformation system of claim 3 , wherein the export module is configured to convert the cumulative schema into a relational schema and output the data of the retrieved objects to the data warehouse according to the relational schema.

5. The data transformation system of claim 4 , wherein the export module is configured to generate commands for the data warehouse that update a schema of the data warehouse to reflect any changes made to the relational schema.

6. The data transformation system of claim 4 , wherein the export module is configured to create at least one intermediate file from the data of the retrieved objects according to the relational schema, wherein the at least one intermediate file has a predefined data warehouse format.

7. The data transformation system of claim 6 , wherein export module is configured to bulk load the at least one intermediate file into the data warehouse.

8. The data transformation system of claim 1 , further comprising an index store configured to store the data from the retrieved objects in columnar form.

9. The data transformation system of claim 8 , wherein the export module is configured to generate row-based data from the stored data in the index store.

10. The data transformation system of claim 8 , wherein the schema inference module is configured to create a time index in the index store that maps time values to identifiers of the retrieved objects.

11. The data transformation system of claim 10 , wherein, for each retrieved object of the retrieved objects, the time value denotes at least one of (i) a transaction time corresponding to creation of the retrieved object or (ii) a valid time corresponding to the retrieved object.

12. The data transformation system of claim 8 , further comprising a write-optimized store configured to (i) cache additional objects for later storage in the index store and (ii) in response to a size of the cache reaching a threshold, package the additional objects together for bulk loading into the index store.

13. The data transformation system of claim 1 , wherein the schema inference module is configured to collect statistics on the metadata of the retrieved objects.

14. The data transformation system of claim 1 , wherein dynamically creating the cumulative schema further includes:
determining that a particular field has a different data type for at least two of the retrieved objects; and
selectively updating the cumulative schema to indicate that the particular field is type polymorphic.

15. The data transformation system of claim 1 , wherein the schema inference module is configured to, in response to the statistics on data types, recast the data of some of the retrieved objects.

16. The data transformation system of claim 1 , wherein the schema inference module is configured to, in response to in response to a determination that the data of at least one retrieved object is typed incorrectly, report the data of some of the retrieved objects to a user as potentially being typed incorrectly.

17. The data transformation system of claim 1 , wherein the schema inference module is configured to collect statistics on the data of the retrieved objects.

18. The data transformation system of claim 17 , wherein the statistics includes at least one of minimum, maximum, average, and standard deviation.

19. The data transformation system of claim 1 , further comprising a data collector module configured to receive relational data from the first data source and generate the objects for use by the schema inference module.

20. The data transformation system of claim 19 , wherein the data collector module is configured to eventize the relational data by creating (i) a first column indicating a table from which each item of the relational data is retrieved and (ii) a second column indicating a timestamp associated with each item of the relational data.

21. The data transformation system of claim 1 , further comprising a scheduling module configured to assign processing jobs to the schema inference module and the export module according to predetermined dependency information.

22. The data transformation system of claim 1 , wherein the export module is configured to partition the cumulative schema into multiple tables, wherein each of the multiple tables includes columns that appear together in the retrieved objects.

23. The data transformation system of claim 22 , wherein the export module is configured to partition the cumulative schema according to columns found in corresponding groups of the retrieved objects that each have a different value for an identifier element.

24. The data transformation system of claim 1 , wherein the schema inference module records a source identifier for each of the retrieved objects.

25. The data transformation system of claim 24 , wherein, for each object of the retrieved objects, the source identifier includes a unique identifier of the first data source and a position of the object within the first data source.

Assignees

• Amiato Inc
• Amazon Tech Inc

Inventors

• Tsirogiannis Dimitris
• Binkert Nathan A
• Harizopoulos Stavros
• Shah Mehul A
• Sowell Benjamin
• Kaplan Bryan D
• Meyer Kevin R

Classifications

• G06F16/211 Primary

  Schema design and management · CPC title

• G06F16/86 Primary

  Mapping to a database · CPC title

• G06F16/254

  Extract, transform and load [ETL] procedures, e.g. ETL data flows in data warehouses · CPC title

• G06F16/235

  Update request formulation · CPC title

• G06F16/22

  Indexing; Data structures therefor; Storage structures · CPC title