Big Data

116

Solutions

278

Contributors

Big data describes data sets that are so large or complex that traditional data processing applications are inadequate. Challenges include analysis, capture, data curation, search, sharing, storage, transfer, visualization, querying and information privacy. The term often refers simply to the use of predictive analytics or certain other advanced methods to extract value from data, and seldom to a particular size of data set.

Share tech news, updates, or what's on your mind.

Sign up to Post

We use SQL server 2017 as our main data store RDMS

We have a vehicle tracking app with one massive log table (all positions for vehicles)  and a lot of other small tables (like car number plate, users etc), pretty much every query involves a join to this main log table
We are just getting going, and the main table is over a billion rows and over 1TB with a relatively small number of cars in relation to the growth plans for the next year
I know this is small, but we estimate we will grow between 10x to 100x in the next year, which has brought up some architecture questions for long term.

Currently we use SQL server which runs nicely, we will also be adding redis to address reduce load on DB,
Currently the DB grows by 10GB every day
Considering we maybe growing at least by 100GB a day, over time this puts us into the BIG DATA category, and I need to start looking at some options for scaling
Every record that gets inserted to the log table gets an ID, and we use this ID for logging events, alerts etc for fast performance (this works well at the current sizing)

Currently my biggest headache is servers with disk storage, I can find server options upto 10TB quite easily, but after this options are limited and prices sky rocket.
Pricing is a massive issue, we are not Google/Microsoft and cannot afford huge server costs.
Performance is also a huge issue, for example, people expect to run reports and expect very small loading times.
We were also planning to move …
0
Rowby Goren Makes an Impact on Screen and Online
LVL 13
Rowby Goren Makes an Impact on Screen and Online

Learn about longtime user Rowby Goren and his great contributions to the site. We explore his method for posing questions that are likely to yield a solution, and take a look at how his career transformed from a Hollywood writer to a website entrepreneur.

Hi ,

Could please some help to write Recursive Query in Hive (or in Pyspark) .



CREATE   TABLE VSOLD_TESTCALC  AS (
     WITH
        RECURSIVE TEST
        (
          PURCHASENUMBER
          , EffDt
          , NetSOLD
          , NetPurchase --Nate
          , PostSeqNum
          , TransCode
          , Amt
          , TransDescription
          , rnum
          , t_flg
          , SpecPeriodFlg
          , MaxSOLDLimit
          )

         AS
        ( SELECT
           PURCHASENUMBER
         , StartDt                                 AS EffDt
         , CAST(SOLD - sell AS DECIMAL(18,4)) AS NetSOLD
         , CAST(Purchase - sell AS DECIMAL(18,4)) AS NetPurchase
         , renewal_seq       AS PostSeqNum
         , CAST(0 AS                INTEGER)       AS TransCode
         , CAST(0.0 AS              DECIMAL(18,2)) AS Amt
         , CAST('INIT_POSTITION' AS VARCHAR(400))  AS TransDescription
         , rnum
         , t_flg
         , SpecPeriodFlg
         , MaxSOLDLimit
        FROM
           TESTDB.SOLD_CurveInitPoint

        UNION ALL

        SELECT
           b.PURCHASENUMBER
         , b.EffDt
         ,NetSOLD 
         , NetPurchase 
         , b.PostSeqNum
         , b.TransCode
         , b.Amt
         , b.TransDescription END
         , b.rnum
         , b.t_flg
         , Flg
         , b.MaxSOLDLimit
        FROM
           TEST a
         , TESTDB.SOLD_Transactions_TESTCALC b
        WHERE
           b.rnum        = a.rnum+1
      

Open in new window

0
Hi ,
 
      Looking to convert below SQL (INSERT and UPDATE ) statement into Hive Query Language .

INSERT INTO HL_TEST.T_Bck
    SELECT a.PURCHASE_NUMBER
         , a.POST_DATE
         , a.EFF_DATE
         , a.SEQ
         , a.TR_CODE
         , a.TR_SIGN
         , a.IND
         , a.TRANSACTION_AMOUNT
         , a.isValidTransaction
         , CASE WHEN la.FINAL_DRAWDOWN_IND = 'Y' THEN 'Y' ELSE 'N' end AS isBDated 
      FROM backD_trans a   
      JOIN HL_TEST.LOANA la
        ON la.LoanNumber = a.PURCHASE_NUMBER
       AND a.EFF_DATE BETWEEN la.src_strt_dt AND la.src_end_dt
	 GROUP 
		BY a.PURCHASE_NUMBER
         , a.POST_DATE
         , a.EFF_DATE
         , a.SEQ
         , a.TR_CODE
         , a.TR_SIGN
		 , a.IND
         , a.TRANSACTION_AMOUNT
         , a.isValidTransaction
         , CASE WHEN la.FINAL_DRAWDOWN_IND = 'Y' THEN 'Y' ELSE 'N' end;

    UPDATE tgt
      FROM HL_TEST.Transactions tgt, HL_TEST.T_Bck t2
       SET EFF_DATE = t2.POST_DATE
     WHERE tgt.PURCHASE_NUMBER = t2.PURCHASE_NUMBER 
      AND tgt.POST_DATE = t2.POST_DATE
      AND tgt.EFF_DATE = t2.EFF_DATE
      AND tgt.TR_CODE = t2.TR_CODE
      AND tgt.SEQ = t2.SEQ
      AND tgt.TR_SIGN = t2.TR_SIGN 
      AND tgt.TRANSACTION_AMOUNT = t2.TRANSACTION_AMOUNT 
      AND tgt.IND = t2.IND
      AND t2.isBDated = 'Y';

Open in new window

Would appreciate if some one can please help me to run above statement in Hive .

Thanks
0
I would like to know if we can publish to Power BI service (Pro) by using Power BI Desktop ?
0
Would like to understand whether pandas(python) can be used to convert from oracle pl/sql to python(accessing hive) and later ported into spark(Big data) environment or development.

Is to be done using Pyspark for executing in spark(Big data) environment?
0
I have a relatively new company in the UK and it's a technology business that offers custom software services, big data services, and analytics services. I am quite new to marketing. Where should I add the company to make it connected with the UK and US markets?

The answer may include any classifieds or other places where to add the company. It may also be any process (for example LinkedIn prospecting) to reach out to individual companies.
0
Why do companies purchase a product known as Complex Event Processor (TIBCO Streambase CEP, IBM Infosphere CEP) or download Open Source (Siddhi, Esper)?

I understand why companies use real-time analytics in general to make sense of real-time data streams, but I don't understand why CEP.

"Complex event processing (CEP) uses patterns to detect composite events in streams of tuples."
CEP also joins many streams and finds patterns among the whole.

But I don't get it why use CEP and not Spark? Is there any use case you can explain this on?
0
I have a public API endpoint that I am pulling a json file every 30 mins. Right now I am using a python pandas dataframe to pull and upload the file to a cloud storage bucket and then sending to pub sub to process and place into BQ. The problem with this is that the file name stays the same and even though I have  gcs text stream to pub sub if it reads the file once it never reads it again even though the file attributes have changed. My question here is can any one help me with code that will pull from an api web link and stream the data directly to pub sub?

Sample code below:
import json
import pandas as pd
from sodapy import Socrata
from io import StringIO
import datalab.storage as gcs
from google.oauth2 import service_account

client = Socrata("sample.org", None)
results = client.get("xxx")

# Convert to pandas DataFrame
results_df = pd.DataFrame.from_records(results, columns =['segmentid','street','_direction','_fromst','_tost','_length','_strheading','_comments','start_lon','_lif_lat','lit_lon','_lit_lat','_traffic','_last_updt'])
# send results to GCP
gcs.Bucket('test-temp').item('data.json').write_to(results_df.to_json(orient='records', lines=True),'text/json')
0
Hello.. what type of storage architecture would be most useful for kafka, zookeeper running in a kubernetes and containerized environment. Would glusterfs that manages storage persistence, replication, resiliency, and HA be an overkill when the distributed bigdata services do this on their own anyway? We're debating internally why glusterfs would be needed for distributed bigdata systems like Cassandra, Mongo, Kafka, Zookeeper, etc. Thus trying to keep the architecture simpler. Just wondering if anyone has thoughts or experience to share that would help us a bit also.
0
Hi I'm looking for information on Data center growth in Australia. In particular publications on Industry trends on value and growth in Australia.
Anything that talks about isolation of Perth also beneficial. Thanks
0
Introduction to Web Design
LVL 13
Introduction to Web Design

Develop a strong foundation and understanding of web design by learning HTML, CSS, and additional tools to help you develop your own website.

Hi, I'm building a big data streaming pipeline that takes streams from a camera through kinesis to trigger a lambda function. The lambda function will then use AWS machine learning to detect objects and the images are stored in S3 and their metadata is stored in DDB. My problem is that the first frame of the video is being stored on S3 and DynamoDB repeatedly (same image is being stored). Here is the lambda code (the main function):

def process_image(event, context):

    #Initialize clients
    rekog_client = boto3.client('rekognition')
    s3_client = boto3.client('s3')
    dynamodb = boto3.resource('dynamodb')

    s3_bucket = ...
    s3_key_frames_root = ...

    ddb_table = dynamodb.Table(...)
    rekog_max_labels = ...
    rekog_min_conf = float(...)
    label_watch_list = ...
    label_watch_min_conf = ...

    #Iterate on frames fetched from Kinesis
    for record in event['Records']:
        
        frame_package_b64 = record['kinesis']['data']
        frame_package = cPickle.loads(base64.b64decode(frame_package_b64))
        img_bytes.append(frame_package["ImageBytes"])
        frame_count = frame_package["FrameCount"]

        rekog_response = rekog_client.detect_labels(
            Image={
                'Bytes': img_bytes
            },
            MaxLabels=rekog_max_labels,
            MinConfidence=rekog_min_conf
        )

        #Iterate on rekognition labels. Enrich and prep them for storage in DynamoDB
        labels_on_watch_list = []
        

Open in new window

0
I’m trying to crack my head around another key challenge at the moment. That is to Increase
AI’s accuracy to be able to identify demographic gender groups from 70% to 80%.
Currently accuracy of the information is at 70%.

I’m running a Data Management Platform (DMP) which collects demographic tag
information and these information are saved in Treasure Data.

I’m thinking of setting some rules in AI to be able to better
differentiate gender. For example, E-commerce data parameters that tells AI that
visitors who have viewed cosmetics have a higher chance of being female while
gadget viewers would have a higher chance of being male.

Do you think these are good ideas/methods? If yes, why. If no, any better
suggestions please?

The Data I have at the moment is not very large, average database of 100,000
users. I’m not able to increase this number any bigger at the moment.

Would be great if anyone could also advise or share on how best to
approach the challenges above and help shed some light on the challenges faced.

Many thanks in advance!
0
I’m trying to crack my head around a key challenge at the moment. That is to increase
AI’s accuracy to be able to identify demographic age groups to plus minus 5 years
old.

Current AI settings only allow categorization of age groups of 20-24 years old,
25-29 years old…55 to 59 years old, 60 years old and above. Not plus minus.

I’m running a Data Management Platform (DMP) which collects demographic tag
information and these information are saved in Treasure Data.

I’m thinking of filtering groups of 10 years old , example 20-24, 25-29
as one group then feeding it to AI to try to increase accuracy instead of
inserting the full raw data CSV for AI to compute.

Do you think these are good ideas/methods? If yes, why. If no, any better
suggestions please?

The Data I have at the moment is not very large, average database of 100,000
users. I’m not able to increase this number any bigger at the moment.

Would be great if anyone could also advise or share on how best to
approach the challenges above and help shed some light on the challenges faced.

Many thanks in advance!
0
Hello Experts,

I have created the following Hadoop Hive Script.

The script is attempting to store the results of a query into the following location:

LOCATION 'wasb://adfgetstarted@geogstoreacct.blob.core.windows.net/sampleout/';

However, I keep on getting the following error:

FAILED: ParseException line 9:0 Failed to recognize predicate 'ROW'. Failed rule: 'identifier' in table or column identifier
18/01/30 16:08:06 [main]: ERROR ql.Driver: FAILED: ParseException line 9:0 Failed to recognize predicate 'ROW'. Failed rule: 'identifier' in table or column identifier
org.apache.hadoop.hive.ql.parse.ParseException: line 9:0 Failed to recognize predicate 'ROW'. Failed rule: 'identifier' in table or column identifier

Open in new window


The Hive script is as follows:

[code]DROP TABLE IF EXISTS geography;
CREATE EXTERNAL TABLE geography
(
 anonid INT,
 eprofileclass INT,
 fueltypes STRING,
 acorn_category INT,
 acorn_group STRING,
 acorn_type INT,
 nuts4 STRING,
 lacode STRING,
 nuts1 STRING,
 gspgroup STRING,
 ldz STRING,
 gas_elec STRING,
 gas_tout STRING
)
ROW FORMAT DELIMITED
 FIELDS TERMINATED BY ','
STORED AS TEXTFILE
LOCATION 'wasb://adfgetstarted@geogstoreacct.blob.core.windows.net/sampleout/'
TBLPROPERTIES ("skip.header.line.count" = "1");

Create table acorn_category_frequency
 as
select acorn_category,
 count(*) as acorn_categorycount
from geography
group by acorn_category,
ROW FORMAT DELIMITED
 FIELDS TERMINATED BY ','
STORED AS TEXTFILE
LOCATION 'wasb://adfgetstarted@geogstoreacct.blob.core.windows.net/sampleout/';

Open in new window

[/code]

Can someone please help figure out where I'm going wrong in the script?

Thanks
0
Hello Experts,

I have run a hql called samplehive.hql, see attached. However, the script fails with the following error:

FAILED: ParseException line 1:2 cannot recognize input near 'D' 'R' 'O'
18/01/17 20:46:46 [main]: ERROR ql.Driver: FAILED: ParseException line 1:2 cannot recognize input near 'D' 'R' 'O'
org.apache.hadoop.hive.ql.parse.ParseException: line 1:2 cannot recognize input near 'D' 'R' 'O'

I'm very new to Hadoop Hive, can someone take a look at the script and let me know where I'm going wrong

Thanks
samplehive.txt
0
So I have a dataset wherein I have account number and "days past due" with every observation. So for every account number, as soon as the "days past due" column hits a code like "DLQ3" , I want to remove rest of the rows for that account(even if DLQ3 is the first observation for that account).

My dataset looks like :

Observation date Account num   Days past due

2016-09                           200056              DLQ1
2016-09                           200048              DLQ2
2016-09                           389490              NORM
2016-09                           383984              DLQ3.....

So for account 383984, I want to remove all the rows post the date 2016-09 as now its in default.

So in all I want to see when the account hits DLQ3 and when it does I want to remove all the rows post the first DLQ3 observation.
0
My business is exploring the option of recoding item codes as currently its all over the place. Ideally going forward we would like to have only one serial number generated for an item  and that the serial number will be the same as the item number.

Is it possible and what impact it will have to the business.

Thanks
0
Hello,

I am new to Hadoop.  I have a question regarding yarn memory allocation.  If  we have 16GB memory in cluster,  we can have least 3 4GB cluster an keep 4 GB for other uses.  If a job needs 10 GB RAM, would it use 3 containers or  use one container and will start using the ram rest of the RAM ?
0
Hello Guys,

We would like to keep Hadoop prod , dev and QA with standard settings and configurations should sync.   What is the best practise to keep them same?  Since we have 100+ data nodes in PROD and only 8 nodes in Dev and 8 Nodes in QA.

We need to make sure all of them are in sync. What is best practise to keep them same?
0
Fundamentals of JavaScript
LVL 13
Fundamentals of JavaScript

Learn the fundamentals of the popular programming language JavaScript so that you can explore the realm of web development.

Hi,

I am curious if someone knows the best way to set alerts based on certain keywords for financial filings such as 8K, 10K etc. For example, I want to create an alert such that when the following filing appears on the website and has a keyword like "PSU", I get an alert.https://www.sec.gov/Archives/edgar/data/1115128/000156459017019148/quot-8k_20170928.htm

Thanks
0
Hello,

When we create datanodes ,  for the disks do we need to use local disks or SAN disks?  Most of them are recommending the local disks. Why do we need to have local disks?
0
detail data blocks will not query when one of them change
0
I had this question after viewing Advice for vb.net web application structure with code generator - refactoring, rewrite, change ORM?.

Hi Mr. tablaFreak,

Actually i was looking for similar code generator that enables me to create data-intensive asp.net web application in vb.net and after reading this article i think this is the best performance approach for CRUD operations with big data, but i am really not aware of how to bind class records to write literal HTML code in the code behind as you mentioned, so kindly provide your your code generator along with few samples that can help for the same

Your assistance is highly appreciated.
My email is SherifMazar@gmail.com
0
I have large numbers PDF document, from which I need to extract text. The extracted text I use for further processing. I did this for a small subset of documents using Tesseract API in a linear approach and I get the required output. However, this takes a very long time when I have a large number of documents.

I tried to use the Hadoop environment processing capabilities (Map-Reduce) and storage (HDFS) for solving this issue. However, I am facing problem to implement Tesseract API into the Hadoop (Map-Reduce) approach. As Teserract converts the files into intermediate image files, I am confused as to how intermediate result Image files of Tesseract-API-process can be handled inside HDFS.

I have searched and unsuccesfully tried a few options earlier like:

    I have extracted text from PDF by extending FileInputFormat class into my own PdfInputFormat class using Hadoop-Map-Reduce, for this i used Apache PDFBox to extract text from pdf, but when it comes to scanned-pdf's which contains image, this solution does not give me the required results.

    I found few answers on the same topic stating to use -Fuse and that will help or one should generate image files locally and than upload those into hdfs for further processing. Not sure if this is the correct approach.

Would like to know approaches around this.
0
HI,

I am trying to find sample dataset about (cloud) storage server file access logs to conduct my research project. Can anyone please suggest any ideas or places to find this type of sample files? I think maybe something like FTP server's log dataset. because my project focus on file access not web page access.

Thanks in advance.
0

Big Data

116

Solutions

278

Contributors

Big data describes data sets that are so large or complex that traditional data processing applications are inadequate. Challenges include analysis, capture, data curation, search, sharing, storage, transfer, visualization, querying and information privacy. The term often refers simply to the use of predictive analytics or certain other advanced methods to extract value from data, and seldom to a particular size of data set.

Top Experts In
Big Data
<
Monthly
>