Implement slowly altering dimensions in an information lake utilizing AWS Glue and Delta

In an information warehouse, a dimension is a construction that categorizes information and measures as a way to allow customers to reply enterprise questions. As an example an instance, in a typical gross sales area, buyer, time or product are dimensions and gross sales transactions is a truth. Attributes inside the dimension can change over time—a buyer can change their handle, an worker can transfer from a contractor place to a full-time place, or a product can have a number of revisions to it. A slowly altering dimension (SCD) is an information warehousing idea that comprises comparatively static information that may change slowly over a time frame. There are three main kinds of SCDs maintained in information warehousing: Sort 1 (no historical past), Sort 2 (full historical past), and Sort 3 (restricted historical past). Change information seize (CDC) is a attribute of a database that gives a capability to determine the information that modified between two database hundreds, in order that an motion will be carried out on the modified information.

As organizations throughout the globe are modernizing their information platforms with information lakes on Amazon Easy Storage Service (Amazon S3), dealing with SCDs in information lakes will be difficult. It turns into much more difficult when supply methods don’t present a mechanism to determine the modified information for processing inside the information lake and makes the information processing extremely advanced if the information supply occurs to be semi-structured as an alternative of a database. The important thing goal whereas dealing with Sort 2 SCDs is to outline the beginning and finish dates to the dataset precisely to trace the modifications inside the information lake, as a result of this offers the point-in-time reporting functionality for the consuming functions.

On this put up, we give attention to demonstrating determine the modified information for a semi-structured supply (JSON) and seize the complete historic information modifications (SCD Sort 2) and retailer them in an S3 information lake, utilizing AWS Glue and open information lake format Delta.io. This implementation helps the next use instances:

• Monitor Sort 2 SCDs with begin and finish dates to determine the present and full historic information and a flag to determine the deleted information within the information lake (logical deletes)
• Use consumption instruments resembling Amazon Athena to question historic information seamlessly

Answer overview

This put up demonstrates the answer with an end-to-end use case utilizing a pattern worker dataset. The dataset represents worker particulars resembling ID, identify, handle, cellphone quantity, contractor or not, and extra. To reveal the SCD implementation, take into account the next assumptions:

• The information engineering workforce receives each day recordsdata which can be full snapshots of information and don’t include any mechanism to determine supply report modifications
• The workforce is tasked with implementing SCD Sort 2 performance for figuring out new, up to date, and deleted information from the supply, and to protect the historic modifications within the information lake
• As a result of the supply methods don’t present the CDC functionality, a mechanism must be developed to determine the brand new, up to date, and deleted information and persist them within the information lake layer

The structure is carried out as follows:

• Supply methods ingest recordsdata within the S3 touchdown bucket (this step is mimicked by producing the pattern information utilizing the supplied AWS Lambda perform into the touchdown bucket)
• An AWS Glue job (Delta job) picks the supply information file and processes the modified information from the earlier file load (new inserts, updates to the prevailing information, and deleted information from the supply) into the S3 information lake (processed layer bucket)
• The structure makes use of the open information lake format (Delta), and builds the S3 information lake as a Delta Lake, which is mutable, as a result of the brand new modifications will be up to date, new inserts will be appended, and supply deletions will be recognized precisely and marked with a delete_flag worth
• An AWS Glue crawler catalogs the information, which will be queried by Athena

The next diagram illustrates our structure.

Stipulations

Earlier than you get began, be sure you have the next conditions:

Deploy the answer

For this answer, we offer a CloudFormation template that units up the providers included within the structure, to allow repeatable deployments. This template creates the next assets:

• Two S3 buckets: a touchdown bucket for storing pattern worker information and a processed layer bucket for the mutable information lake (Delta Lake)
• A Lambda perform to generate pattern information
• An AWS Glue extract, remodel, and cargo (ETL) job to course of the supply information from the touchdown bucket to the processed bucket

To deploy the answer, full the next steps:

1. Select Launch Stack to launch the CloudFormation stack:

2. Enter a stack identify.
3. Choose I acknowledge that AWS CloudFormation would possibly create IAM assets with customized names.
4. Select Create stack.

After the CloudFormation stack deployment is full, navigate to AWS CloudFormation console to notice the next assets on the Outputs tab:

• Information lake assets – The S3 buckets scd-blog-landing-xxxx and scd-blog-processed-xxxx (known as scd-blog-landing and scd-blog-processed within the subsequent sections on this put up)
• Pattern information generator Lambda perform – SampleDataGenaratorLambda-<CloudFormation Stack Identify> (known as SampleDataGeneratorLambda)
• AWS Glue Information Catalog database – deltalake_xxxxxx (known as deltalake)
• AWS Glue Delta job – <CloudFormation-Stack-Identify>-src-to-processed (known as src-to-processed)

Word that deploying the CloudFormation stack in your account incurs AWS utilization prices.

Check SCD Sort 2 implementation

With the infrastructure in place, you’re prepared to check out the general answer design and question historic information from the worker dataset. This put up is designed to be carried out for an actual buyer use case, the place you get full snapshot information each day. We take a look at the next facets of SCD implementation:

• Run an AWS Glue job for the preliminary load
• Simulate a state of affairs the place there aren’t any modifications to the supply
• Simulate insert, replace, and delete eventualities by including new information, and modifying and deleting present information
• Simulate a state of affairs the place the deleted report comes again as a brand new insert

Generate a pattern worker dataset

To check the answer, and earlier than you can begin your preliminary information ingestion, the information supply must be recognized. To simplify that step, a Lambda perform has been deployed within the CloudFormation stack you simply deployed.

Open the perform and configure a take a look at occasion, with the default hello-world template occasion JSON as seen within the following screenshot. Present an occasion identify with none modifications to the template and save the take a look at occasion.

Select Check to invoke a take a look at occasion, which invokes the Lambda perform to generate the pattern information.

When the Lambda perform completes its invocation, it is possible for you to to see the next pattern worker dataset within the touchdown bucket.

Run the AWS Glue job

Affirm should you see the worker dataset within the path s3://scd-blog-landing/dataset/worker/. You’ll be able to obtain the dataset and open it in a code editor resembling VS Code. The next is an instance of the dataset:

{"emp_id":1,"first_name":"Melissa","last_name":"Parks","Handle":"19892 Williamson Causeway Suite 737nKarenborough, IN 11372","phone_number":"001-372-612-0684","isContractor":false}
{"emp_id":2,"first_name":"Laura","last_name":"Delgado","Handle":"93922 Rachel Parkways Suite 717nKaylaville, GA 87563","phone_number":"001-759-461-3454x80784","isContractor":false}
{"emp_id":3,"first_name":"Luis","last_name":"Barnes","Handle":"32386 Rojas SpringsnDicksonchester, DE 05474","phone_number":"127-420-4928","isContractor":false}
{"emp_id":4,"first_name":"Jonathan","last_name":"Wilson","Handle":"682 Tempo Springs Apt. 011nNew Wendy, GA 34212","phone_number":"761.925.0827","isContractor":true}
{"emp_id":5,"first_name":"Kelly","last_name":"Gomez","Handle":"4780 Johnson TunnelnMichaelland, WI 22423","phone_number":"+1-303-418-4571","isContractor":false}
{"emp_id":6,"first_name":"Robert","last_name":"Smith","Handle":"04171 Mitchell Springs Suite 748nNorth Juliaview, CT 87333","phone_number":"261-155-3071x3915","isContractor":true}
{"emp_id":7,"first_name":"Glenn","last_name":"Martinez","Handle":"4913 Robert ViewsnWest Lisa, ND 75950","phone_number":"001-638-239-7320x4801","isContractor":false}
{"emp_id":8,"first_name":"Teresa","last_name":"Estrada","Handle":"339 Scott ValleynGonzalesfort, PA 18212","phone_number":"435-600-3162","isContractor":false}
{"emp_id":9,"first_name":"Karen","last_name":"Spencer","Handle":"7284 Coleman Membership Apt. 813nAndersonville, AS 86504","phone_number":"484-909-3127","isContractor":true}
{"emp_id":10,"first_name":"Daniel","last_name":"Foley","Handle":"621 Sarah Lock Apt. 537nJessicaton, NH 95446","phone_number":"457-716-2354x4945","isContractor":true}
{"emp_id":11,"first_name":"Amy","last_name":"Stevens","Handle":"94661 Younger Lodge Suite 189nCynthiamouth, PR 01996","phone_number":"241.375.7901x6915","isContractor":true}
{"emp_id":12,"first_name":"Nicholas","last_name":"Aguirre","Handle":"7474 Joyce MeadowsnLake Billy, WA 40750","phone_number":"495.259.9738","isContractor":true}
{"emp_id":13,"first_name":"John","last_name":"Valdez","Handle":"686 Brian Forges Suite 229nSullivanbury, MN 25872","phone_number":"+1-488-011-0464x95255","isContractor":false}
{"emp_id":14,"first_name":"Michael","last_name":"West","Handle":"293 Jones Squares Apt. 997nNorth Amandabury, TN 03955","phone_number":"146.133.9890","isContractor":true}
{"emp_id":15,"first_name":"Perry","last_name":"Mcguire","Handle":"2126 Joshua Forks Apt. 050nPort Angela, MD 25551","phone_number":"001-862-800-3814","isContractor":true}
{"emp_id":16,"first_name":"James","last_name":"Munoz","Handle":"74019 Banks EstatesnEast Nicolefort, GU 45886","phone_number":"6532485982","isContractor":false}
{"emp_id":17,"first_name":"Todd","last_name":"Barton","Handle":"2795 Kelly Shoal Apt. 500nWest Lindsaytown, TN 55404","phone_number":"079-583-6386","isContractor":true}
{"emp_id":18,"first_name":"Christopher","last_name":"Noble","Handle":"Unit 7816 Field 9004nDPO AE 29282","phone_number":"215-060-7721","isContractor":true}
{"emp_id":19,"first_name":"Sandy","last_name":"Hunter","Handle":"7251 Sarah CreeknWest Jasmine, CO 54252","phone_number":"8759007374","isContractor":false}
{"emp_id":20,"first_name":"Jennifer","last_name":"Ballard","Handle":"77628 Owens Key Apt. 659nPort Victorstad, IN 02469","phone_number":"+1-137-420-7831x43286","isContractor":true}
{"emp_id":21,"first_name":"David","last_name":"Morris","Handle":"192 Leslie Groves Apt. 930nWest Dylan, NY 04000","phone_number":"990.804.0382x305","isContractor":false}
{"emp_id":22,"first_name":"Paula","last_name":"Jones","Handle":"045 Johnson Viaduct Apt. 732nNorrisstad, AL 12416","phone_number":"+1-193-919-7527x2207","isContractor":true}
{"emp_id":23,"first_name":"Lisa","last_name":"Thompson","Handle":"1295 Judy Ports Suite 049nHowardstad, PA 11905","phone_number":"(623)577-5982x33215","isContractor":true}
{"emp_id":24,"first_name":"Vickie","last_name":"Johnson","Handle":"5247 Jennifer Run Suite 297nGlenberg, NC 88615","phone_number":"708-367-4447x9366","isContractor":false}
{"emp_id":25,"first_name":"John","last_name":"Hamilton","Handle":"5899 Barnes PlainnHarrisville, NC 43970","phone_number":"341-467-5286x20961","isContractor":false}

Obtain the dataset and hold it prepared, as a result of you’ll modify the dataset for future use instances to simulate the inserts, updates, and deletes. The pattern dataset generated for you can be totally completely different than what you see within the previous instance.

To run the job, full the next steps:

1. On the AWS Glue console, select Jobs within the navigation pane.
2. Select the job src-to-processed.
3. On the Runs tab, select Run.

When the AWS Glue job is run for the primary time, the job reads the worker dataset from the touchdown bucket path and ingests the information to the processed bucket as a Delta desk.

When the job is full, you possibly can create a crawler to see the preliminary information load. The next screenshot reveals the database out there on the Databases web page.

4. Select Crawlers within the navigation pane.
5. Select Create crawler.

6. Identify your crawler delta-lake-crawler, then select Subsequent.

7. Choose Not but for information already mapped to AWS Glue tables.
8. Select Add an information supply.

9. On the Information supply drop-down menu, select Delta Lake.
10. Enter the trail to the Delta desk.
11. Choose Create Native tables.
12. Select Add a Delta Lake information supply.

13. Select Subsequent.

14. Select the function that was created by the CloudFormation template, then select Subsequent.

15. Select the database that was created by the CloudFormation template, then select Subsequent.

16. Select Create crawler.

17. Choose your crawler and select Run.

Question the information

After the crawler is full, you possibly can see the desk it created.

To question the information, full the next steps:

1. Select the worker desk and on the Actions menu, select View information.

You’re redirected to the Athena console. For those who don’t have the most recent Athena engine, create a brand new Athena workgroup with the most recent Athena engine.

2. Below Administration within the navigation pane, select Workgroups.

3. Select Create workgroup.

4. Present a reputation for the workgroup, resembling DeltaWorkgroup.
5. Choose Athena SQL because the engine, and select Athena engine model 3 for Question engine model.

6. Select Create workgroup.

7. After you create the workgroup, choose the workgroup (DeltaWorkgroup) on the drop-down menu within the Athena question editor.

8. Run the next question on the worker desk:

SELECT * FROM "deltalake_2438fbd0"."worker";

Word: Replace the right database identify from the CloudFormation outputs earlier than working the above question.

You’ll be able to observe that the worker desk has 25 information. The next screenshot reveals the whole worker information with some pattern information.

The Delta desk is saved with an emp_key, which is exclusive to each change and is used to trace the modifications. The emp_key is created for each insert, replace, and delete, and can be utilized to seek out all of the modifications pertaining to a single emp_id.

The emp_key is created utilizing the SHA256 hashing algorithm, as proven within the following code:

df.withColumn("emp_key", sha2(concat_ws("||", col("emp_id"), col("first_name"), col("last_name"), col("Handle"),
            col("phone_number"), col("isContractor")), 256))

Carry out inserts, updates, and deletes

Earlier than making modifications to the dataset, let’s run the identical job yet another time. Assuming that the present load from the supply is identical because the preliminary load with no modifications, the AWS Glue job shouldn’t make any modifications to the dataset. After the job is full, run the earlier Choose question within the Athena question editor and ensure that there are nonetheless 25 energetic information with the next values:

• All 25 information with the column isCurrent=true
• All 25 information with the column end_date=Null
• All 25 information with the column delete_flag=false

After you verify the earlier job run with these values, let’s modify our preliminary dataset with the next modifications:

1. Change the isContractor flag to false (change it to true in case your dataset already reveals false) for emp_id=12.
2. Delete all the row the place emp_id=8 (be certain to avoid wasting the report in a textual content editor, as a result of we use this report in one other use case).
3. Copy the row for emp_id=25 and insert a brand new row. Change the emp_id to be 26, and ensure to alter the values for different columns as properly.

After we make these modifications, the worker supply dataset appears to be like like the next code (for readability, now we have solely included the modified information as described within the previous three steps):

{"emp_id":12,"first_name":"Nicholas","last_name":"Aguirre","Handle":"7474 Joyce MeadowsnLake Billy, WA 40750","phone_number":"495.259.9738","isContractor":false}
{"emp_id":26,"first_name":"John-copied","last_name":"Hamilton-copied","Handle":"6000 Barnes PlainnHarrisville-city, NC 5000","phone_number":"444-467-5286x20961","isContractor":true}

4. Now, add the modified fake_emp_data.json file to the identical supply prefix.

5. After you add the modified worker dataset to Amazon S3, navigate to the AWS Glue console and run the job.
6. When the job is full, run the next question within the Athena question editor and ensure that there are 27 information in whole with the next values:

SELECT * FROM "deltalake_2438fbd0"."worker";

Word: Replace the right database identify from the CloudFormation output earlier than working the above question.

7. Run one other question within the Athena question editor and ensure that there are 4 information returned with the next values:

SELECT * FROM "AwsDataCatalog"."deltalake_2438fbd0"."worker" the place emp_id in (8, 12, 26)
order by emp_id;

Word: Replace the right database identify from the CloudFormation output earlier than working the above question.

You will note two information for emp_id=12:

• One emp_id=12 report with the next values (for the report that was ingested as a part of the preliminary load):
  • emp_key=44cebb094ef289670e2c9325d5f3e4ca18fdd53850b7ccd98d18c7a57cb6d4b4
  • isCurrent=false
  • delete_flag=false
  • end_date=’2023-03-02’
• A second emp_id=12 report with the next values (for the report that was ingested as a part of the change to the supply):
  • emp_key=b60547d769e8757c3ebf9f5a1002d472dbebebc366bfbc119227220fb3a3b108
  • isCurrent=true
  • delete_flag=false
  • end_date=Null (or empty string)

The report for emp_id=8 that was deleted within the supply as a part of this run will nonetheless exist however with the next modifications to the values:

• isCurrent=false
• end_date=’2023-03-02’
• delete_flag=true

The brand new worker report will likely be inserted with the next values:

• emp_id=26
• isCurrent=true
• end_date=NULL (or empty string)
• delete_flag=false

Word that the emp_key values in your precise desk could also be completely different than what’s supplied right here for instance.

8. For the deletes, we verify for the emp_id from the bottom desk together with the brand new supply file and interior be a part of the emp_key.
9. If the situation evaluates to true, we then verify if the worker base desk emp_key equals the brand new updates emp_key, and get the present, undeleted report (isCurrent=true and delete_flag=false).
10. We merge the delete modifications from the brand new file with the bottom desk for all of the matching delete situation rows and replace the next:
    1. isCurrent=false
    2. delete_flag=true
    3. end_date=current_date

See the next code:

delete_join_cond = "worker.emp_id=employeeUpdates.emp_id and worker.emp_key = employeeUpdates.emp_key"
delete_cond = "worker.emp_key == employeeUpdates.emp_key and worker.isCurrent = true and employeeUpdates.delete_flag = true"

base_tbl.alias("worker")
        .merge(union_updates_dels.alias("employeeUpdates"), delete_join_cond)
        .whenMatchedUpdate(situation=delete_cond, set={"isCurrent": "false",
                                                        "end_date": current_date(),
                                                        "delete_flag": "true"}).execute()

11. For each the updates and the inserts, we verify for the situation if the bottom desk worker.emp_id is the same as the new modifications.emp_id and the worker.emp_key is the same as new modifications.emp_key, whereas solely retrieving the present information.
12. If this situation evaluates to true, we then get the present report (isCurrent=true and delete_flag=false).
13. We merge the modifications by updating the next:
    1. If the second situation evaluates to true:
       1. isCurrent=false
       2. end_date=current_date
    2. Or we insert all the row as follows if the second situation evaluates to false:
       1. emp_id=new report’s emp_key
       2. emp_key=new report’s emp_key
       3. first_name=new report’s first_name
       4. last_name=new report’s last_name
       5. handle=new report’s handle
       6. phone_number=new report’s phone_number
       7. isContractor=new report’s isContractor
       8. start_date=current_date
       9. end_date=NULL (or empty string)
       10. isCurrent=true
       11. delete_flag=false

See the next code:

upsert_cond = "worker.emp_id=employeeUpdates.emp_id and worker.emp_key = employeeUpdates.emp_key and worker.isCurrent = true"
upsert_update_cond = "worker.isCurrent = true and employeeUpdates.delete_flag = false"

base_tbl.alias("worker").merge(union_updates_dels.alias("employeeUpdates"), upsert_cond)
    .whenMatchedUpdate(situation=upsert_update_cond, set={"isCurrent": "false",
                                                            "end_date": current_date()
                                                            }) 
    .whenNotMatchedInsert(
    values={
        "isCurrent": "true",
        "emp_id": "employeeUpdates.emp_id",
        "first_name": "employeeUpdates.first_name",
        "last_name": "employeeUpdates.last_name",
        "Handle": "employeeUpdates.Handle",
        "phone_number": "employeeUpdates.phone_number",
        "isContractor": "employeeUpdates.isContractor",
        "emp_key": "employeeUpdates.emp_key",
        "start_date": current_date(),
        "delete_flag":  "employeeUpdates.delete_flag",
        "end_date": "null"
    })
    .execute()

As a final step, let’s convey again the deleted report from the earlier change to the supply dataset and see how it’s reinserted into the worker desk within the information lake and observe how the entire historical past is maintained.

Let’s modify our modified dataset from the earlier step and make the next modifications.

14. Add the deleted emp_id=8 again to the dataset.

After making these modifications, my worker supply dataset appears to be like like the next code (for readability, now we have solely included the added report as described within the previous step):

{"emp_id":8,"first_name":"Teresa","last_name":"Estrada","Handle":"339 Scott ValleynGonzalesfort, PA 18212","phone_number":"435-600-3162","isContractor":false}

15. Add the modified worker dataset file to the identical supply prefix.
    
16. After you add the modified fake_emp_data.json dataset to Amazon S3, navigate to the AWS Glue console and run the job once more.
17. When the job is full, run the next question within the Athena question editor and ensure that there are 28 information in whole with the next values:

SELECT * FROM "deltalake_2438fbd0"."worker";

Word: Replace the right database identify from the CloudFormation output earlier than working the above question.

18. Run the next question and ensure there are 5 information:

SELECT * FROM "AwsDataCatalog"."deltalake_2438fbd0"."worker" the place emp_id in (8, 12, 26)
order by emp_id;

You will note two information for emp_id=8:

• One emp_id=8 report with the next values (the previous report that was deleted):
  • emp_key=536ba1ba5961da07863c6d19b7481310e64b58b4c02a89c30c0137a535dbf94d
  • isCurrent=false
  • deleted_flag=true
  • end_date=’2023-03-02’
• One other emp_id=8 report with the next values (the brand new report that was inserted within the final run):
  • emp_key=536ba1ba5961da07863c6d19b7481310e64b58b4c02a89c30c0137a535dbf94d
  • isCurrent=true
  • deleted_flag=false
  • end_date=NULL (or empty string)

The emp_key values in your precise desk could also be completely different than what’s supplied right here for instance. Additionally word that as a result of it is a identical deleted report that was reinserted within the subsequent load with none modifications, there will likely be no change to the emp_key.

Finish-user pattern queries

The next are some pattern end-user queries to reveal how the worker change information historical past will be traversed for reporting:

• Question 1 – Retrieve an inventory of all the workers who left the group within the present month (for instance, March 2023).

SELECT * FROM "deltalake_2438fbd0"."worker" the place delete_flag=true and date_format(CAST(end_date AS date),'%Y/%m') ='2023/03'

Word: Replace the right database identify from the CloudFormation output earlier than working the above question.

The previous question would return two worker information who left the group.

• Question 2 – Retrieve an inventory of recent workers who joined the group within the present month (for instance, March 2023).

SELECT * FROM "deltalake_2438fbd0"."worker" the place date_format(start_date,'%Y/%m') ='2023/03' and iscurrent=true

Word: Replace the right database identify from the CloudFormation output earlier than working the above question.

The previous question would return 23 energetic worker information who joined the group.

• Question 3 – Discover the historical past of any given worker within the group (on this case worker 18).

SELECT * FROM "deltalake_2438fbd0"."worker" the place emp_id=18

Word: Replace the right database identify from the CloudFormation output earlier than working the above question.

Within the previous question, we will observe that worker 18 had two modifications to their worker information earlier than they left the group.

Word that the information outcomes supplied on this instance are completely different than what you will notice in your particular information primarily based on the pattern information generated by the Lambda perform.

Clear up

When you will have completed experimenting with this answer, clear up your assets, to stop AWS prices from being incurred:

1. Empty the S3 buckets.
2. Delete the stack from the AWS CloudFormation console.

Conclusion

On this put up, we demonstrated determine the modified information for a semi-structured information supply and protect the historic modifications (SCD Sort 2) on an S3 Delta Lake, when supply methods are unable to supply the change information seize functionality, with AWS Glue. You’ll be able to additional prolong this answer to allow downstream functions to construct further customizations from CDC information captured within the information lake.

Moreover, you possibly can prolong this answer as a part of an orchestration utilizing AWS Step Features or different generally used orchestrators your group is accustomed to. You may also prolong this answer by including partitions the place acceptable. You may also preserve the delta desk by compacting the small recordsdata.

Concerning the authors

Nith Govindasivan, is a Information Lake Architect with AWS Skilled Companies, the place he helps onboarding prospects on their trendy information structure journey by means of implementing Massive Information & Analytics options. Outdoors of labor, Nith is an avid Cricket fan, watching nearly any cricket throughout his spare time and enjoys lengthy drives, and touring internationally.

Vijay Velpula is a Information Architect with AWS Skilled Companies. He helps prospects implement Massive Information and Analytics Options. Outdoors of labor, he enjoys spending time with household, touring, mountaineering and biking.

Sriharsh Adari is a Senior Options Architect at Amazon Internet Companies (AWS), the place he helps prospects work backwards from enterprise outcomes to develop revolutionary options on AWS. Through the years, he has helped a number of prospects on information platform transformations throughout trade verticals. His core space of experience embrace Know-how Technique, Information Analytics, and Information Science. In his spare time, he enjoys taking part in sports activities, binge-watching TV reveals, and taking part in Tabla.