Dynamically processing an event using an extensible data model

• US 9,201,910 B2
• Filed: 08/18/2014
• Issued: 12/01/2015
• Est. Priority Date: 03/31/2010
• Status: Active Grant

• Current Assignee: Cloudera Incorporated
• Original Assignee: Cloudera Incorporated
• Inventors: Hsieh, Jonathan Ming-Cyn, Robinson, Henry Noel
• Primary Examiner(s): TO, BAOQUOC N
• Application Number:
  US14/462,395
• Publication Number:
  US 20150019557A1
• Time in Patent Office: 470 Days
• Field of Search: 707/672
• US Class Current: 1/1
• CPC Class Codes: G06F 16/211 Schema design and management
• CPC Class Codes: G06F 16/24568 Data stream processing; Continuous queries
• CPC Class Codes: H04L 67/01 Protocols
• CPC Class Codes: H04L 67/02 based on web technology, e.g. hypertext transfer protocol [HTTP]

Abstract

Systems and methods of dynamically processing an event using an extensible data model are disclosed. One embodiment includes, specifying attributes of the event in a data model; the data model being extensible to add properties to the event as the dataset is streamed from the source to the sink.

Claims

1. A method for dynamically processing an event including a dataset that is streamed from a source to a sink via nodes, the method comprising:

recording, at a respective node, the event including the dataset in a memory of the respective node using a data model;

extracting key-value data from the dataset; and

annotating the event by adding or updating one or more attributes associated with the event in the data model based on analyzing the extracted key-value data, the data model being extensible to add additional attributes to the event by a subsequent node which is configured to further process the dataset as the event is streamed from the source to the sink; and

specifying based on the key-value data, one or more fields of the event in the data model, wherein the one or more fields include one or more of;

a timestamp field, a source machine field, a body field, or a priority field.

2. The method of claim 1, wherein the one or more attributes include (1) a map from string attribute name to an array of bytes, (2) a route to one or more storage locations for the event, or (iii) one or more formats for outputting of the dataset at the sink.

3. The method of claim 1, wherein the key-value data includes a timestamp, and wherein specifying one or more field includes specifying, based on the timestamp, a priority of the event in a priority field of the event in the data model.

4. The method of claim 1, further comprising:

bucketing the event with other events based on the key-value data.

5. The method of claim 4, wherein the key-value data is a timestamp or a data node name.

6. The method of claim 4, wherein the key-value data is a page hit count or a type of web browser.

7. The method of claim 4, wherein the key-value data is a project identifier or a person'"'"'s identifier.

8. The method of claim 1, wherein the data model includes a metadata table extensible for storing the additional attributes.

9. A non-transitory computer readable medium storing a plurality of instructions which, upon execution by a processor, cause the processor to perform a method for dynamically processing an event including a dataset that is streamed from a source to a sink via nodes, the method comprising:

recording, at a respective node, the event including the dataset in a memory of the respective node using a data model;

extracting key-value data from the dataset; and

annotating the event by adding or updating one or more attributes associated with the event in the data model based on analyzing the extracted key-value data, the data model being extensible to add additional attributes to the event by a subsequent node which is configured to further process the dataset as the event is streamed from the source to the sink; and

specifying based on the key-value data, one or more fields of the event in the data model, wherein the one or more fields include one or more of;

a timestamp field, a source machine field, a body field, or a priority field.

10. The medium of claim 9, wherein the one or more attributes include:

(1) a map from a string attribute name to an array of bytes, (2) a route to one or more storage location for the event, or (3) one or more formats for outputting of the dataset at the sink.

11. The medium of claim 9, wherein the key-value data includes a timestamp, and wherein specifying one or more fields including specifying based on the timestamp, a priority of the event in a priority field of the event in the data model.

12. The medium of claim 9, wherein the method performed by the processor further comprises:

Bucketing the event with other events based on the key-valued data.

13. The medium of claim 9, wherein the data model includes a metadata table extensible for storing the additional attributes.

14. A system having a processor and a memory, the memory storing a plurality of instructions which, when executed by the processor, cause the processor to perform a method for dynamically processing an event including a dataset that is streamed from a source to a sink via nodes, the method comprising:

recording, at a respective node, the event including the dataset in a memory of the respective node using a data model;

extracting key-value data from the dataset; and

annotating the event by adding or updating one or more attributes associated with the event in the data model based on analyzing the extracted key-value data, the data model being extensible to add additional attributes to the event by a subsequent node which is configured to further process the dataset as the event is streamed from the source to the sink; and

specifying based on the key-value data, one or more fields of the event in the data model, wherein the one or more fields include one or more of;

a timestamp field, a source machine field, a body field, or a priority field.

15. The system of claim 14, wherein the one or more attributes include:

(1) a map from a string attribute name to an array of bytes, (2) a route to one or more storage locations for the event, or (3) one or more format for outputting of the dataset at the sink.

16. The system of claim 14, wherein the key-value data includes a timestamps, and wherein specifying one or more fields includes specifying, based on the timestamp, a priority event in a priority field of the event in the data model.

17. The system of claim 14, wherein the method performed by the processor further comprises:

bucketing the event with other events based on the key-value data.

20. The method of claim 14, further comprising:

If the batch of messages has not been successfully written based on the acknowledgment message, causing, by the master node, the collector node to rewrite the datasets to the destination location.

18. A method for collecting and aggregating datasets for storage in a file system with fault tolerance, the file system including agent node, a collector node, and a master node, the method comprising:

Collecting the dataset via an agent node operating in a remote machine, wherein the datasets include a batch messages written by the remote machine, and wherein the batch of messages are processed by the agent node when the size or a lapsed time of the batch of messages reaches a selected threshold;

Generating, by the agent node, a batch identifier (ID) for the batch of messages;

Assigning, by the agent node, an event tag to the batch of messages;

Computing, by the agent node, a checksum for the match of messages;

Writing, by the agent node, the batch of messages along with batch ID, the event tag, and the checksum as an entry in a write-ahead-log (WAL) storage maintained by the agent node on the remote machine;

Transmitting, by the agent node, the dataset to a collector machine, wherein the datasets are transmitted in a data model as an event, the data model being extensible to add additional attributes to the event by a subsequent node who is configured to further process the dataset as the event is streamed from a source to a sink;

Verifying the checksum by a collector node operating in the collector machine;

If the checksum is verified, adding, by the collector node, a tag to a map of tags, wherein the map of tags are associated with multiple tags assigned to multiple matches of messages from the datasets;

Writing, by the collector node, the dataset to a destination location; and

If the batch of messages has been successfully written to the destination location, publishing, by the collector node, the tag to the master node.

19. The method of claim 18, further comprising:

Querying, by the master node, for an acknowledgement message confirming that the batch of messages has been successfully written to the destination location; and

If the batch of messages has been successfully written based on the acknowledgment message, causing, by the master node, the agent node to delete the batch of message from the WAL storage.