The Blas from Pas

Wednesday, 4 September 2013

Pivotal SQLFire export/importing CSV files

Pivotal SQLFire offers 2 built in procedures that allow table data to be exported in CSV format as well as imported back from CSV format as shown below. The procedures are as follows

SYSCS_UTIL.EXPORT_TABLE
SYSCS_UTIL.IMPORT_TABLE

Example

In this example we are using a table as follows

  
sqlf> select * from apples_dept;
DEPTNO     |DNAME         |LOC          
----------------------------------------
70         |SUPPORT       |SYDNEY       
60         |DEV           |PERTH        
50         |MARKETING     |ADELAIDE     
40         |OPERATIONS    |BRISBANE     
30         |SALES         |CHICAGO      
20         |RESEARCH      |DALLAS       
10         |ACCOUNTING    |NEW YORK     

7 rows selected

1. Export the table as shown below.

  
sqlf> CALL SYSCS_UTIL.EXPORT_TABLE('APP', 'APPLES_DEPT', '/Users/papicella/sqlfire/vFabric_SQLFire_111_b42624/pasdemos/sqlfireweb/sql/apples_dept.csv', null, null, null);
Statement executed.

2. View export data as shown below.

[Wed Sep 04 20:10:18 papicella@:~/sqlfire/vFabric_SQLFire_111_b42624/pasdemos/sqlfireweb/sql ] $ cat apples_dept.csv
70,"SUPPORT","SYDNEY"
60,"DEV","PERTH"
50,"MARKETING","ADELAIDE"
40,"OPERATIONS","BRISBANE"
30,"SALES","CHICAGO"
20,"RESEARCH","DALLAS"
10,"ACCOUNTING","NEW YORK"

3. Truncate table

  
sqlf> truncate table apples_dept;
0 rows inserted/updated/deleted

4. Import data back into the table

  
sqlf> CALL SYSCS_UTIL.IMPORT_TABLE('APP', 'APPLES_DEPT', '/Users/papicella/sqlfire/vFabric_SQLFire_111_b42624/pasdemos/sqlfireweb/sql/apples_dept.csv', ',', '"', null, 0);
Statement executed.
sqlf> select * from apples_dept;
DEPTNO     |DNAME         |LOC          
----------------------------------------
10         |ACCOUNTING    |NEW YORK     
20         |RESEARCH      |DALLAS       
30         |SALES         |CHICAGO      
40         |OPERATIONS    |BRISBANE     
50         |MARKETING     |ADELAIDE     
60         |DEV           |PERTH        
70         |SUPPORT       |SYDNEY       

7 rows selected

More Information

http://pubs.vmware.com/vfabric53/index.jsp?topic=/com.vmware.vfabric.sqlfire.1.1/reference/system_procedures/derby/rrefimportproc.html

http://pubs.vmware.com/vfabric53/index.jsp?topic=/com.vmware.vfabric.sqlfire.1.1/reference/system_procedures/derby/rrefexportproc.html

Tuesday, 6 August 2013

CloudFoundry - Managing your MYSQL database service

After deploying a simple java based application using MYSQL in CloudFoundry I found it easy enough to connect to the database it's using for my application as follows. Handy when / if I need to reset the database data or perhaps add additional data which I didn't add at deployment time.

1. Login into the public CloudFoundry instance

2. Click on your application

3. Click the manage button for the ClearDB service which is for MYSQL

4. This takes you to the ClearDB console as shown below.

5. Click on the database instance name and it takes you to a page where you can some graphs and most importantly the connect details into your MYSQL database. For that click on the TAB named "Endpoint Information". Here you find the hostname and the access credentials for the MYSQL database. In my example it is using the default MYSQL port of 3306. The screen shot is not showing the password for obvious reasons.

6. Finally use the connect details below to connect to MYSQL from a third party JDBC tool such as DBVisualizer.I am querying my CUSTOMER table which is par of my Books Web Based Application

Thursday, 1 August 2013

Taking Pivotal Cloud Foundry for a Test Drive

If you haven't hear about CloudFoundry and Pivotal PaaS a good place to start is here. https://www.cloudfoundry.com/ In the example below we deploy a simple WAR file which contains a JSP and servlet using the public PaaS available for trail use.

1. Make sure you create an account which you can do for free at https://www.cloudfoundry.com/

2. You will need a version of ruby such as 1.9.x like I have below.

[Thu Aug 01 12:56:14 papicella@:~/vmware/vFabric/cloud-foundry/apps/java-demo ] $ ruby -v
ruby 1.9.3p286 (2012-10-12 revision 37165) [x86_64-darwin11.4.2]

3. Next you need to install the CouldFoundry command line tool known as "cf" as shown below.

> sudo gem install cf

4. At this point we can target the Pubic CF instance as shown below.

[Wed Jul 31 13:02:53 papicella@:~ ] $ cf target api.run.pivotal.io
Setting target to https://api.run.pivotal.io... OK

5. Now lets login using our login we created at #1 above as well as selecting a deployment stage as follows

[Wed Jul 31 13:10:28 papicella@:~/vmware/vFabric/cloud-foundry ] $ cf login

target: https://api.run.pivotal.io

Email> papicella@vmware.com

Password> ********

Authenticating... OK

1: development

2: production

3: staging

Space> 1

Switching to space development... OK

6. In this example we have our WAR file sitting ina directory where we will push this to CloudFoundry

[Thu Aug 01 12:57:35 papicella@:~/vmware/vFabric/cloud-foundry/apps/java-demo ] $ d

total 32

-rw-r--r-- 1 papicella staff 12502 7 Oct 2011 hello.war

drwxr-xr-x 3 papicella staff 102 31 Jul 20:00 ./

drwxr-xr-x 11 papicella staff 374 1 Aug 11:49 ../

7. Now lets deploy our simple WAR file as shown below.

[Thu Aug 01 13:03:27 papicella@:~/vmware/vFabric/cloud-foundry/apps/java-demo ] $ cf push

Name> hellojava_pas

Instances> 1

1: 128M

2: 256M

3: 512M

4: 1G

Memory Limit> 512M

Creating hellojava_pas... OK

1: hellojava_pas

2: none

Subdomain> hellojava_pas

1: cfapps.io

2: none

Domain> cfapps.io

Binding hellojava_pas.cfapps.io to hellojava_pas... OK

Create services for application?> n

Bind other services to application?> n

Save configuration?> n

Uploading hellojava_pas... OK

Preparing to start hellojava_pas... OK

Checking status of app 'hellojava_pas'...

0 of 1 instances running (1 starting)

1 of 1 instances running (1 running)

Push successful! App 'hellojava_pas' available at http://hellojava_pas.cfapps.io

8. Finally access application from your browser

Finally if we connect to our CloudFoundry dashboard we can monitor / view all our deployment applications as per the screen shot below.

Friday, 26 July 2013

New Open Source Download: Browser-based Schema Management for Pivotal SQLFire

A while ago I started working on a web based utility for Pivotal SQLFire known as SQLFire*Web. Today it was finally released as an open source project. See the link below for more details.

http://blog.gopivotal.com/products/new-open-source-download-browser-based-schema-management-for-pivotal-sqlfire

Friday, 19 July 2013

Pivotal SQLFire - Shutting down members when authentication is on

In a previous blog entry I detailed how to use the DEV BULIT IN authentication mode for SQLFire.

http://theblasfrompas.blogspot.com.au/2013/07/securing-schemas-in-pivotal-sqlfire.html

When you need to shutdown members with authentication on you would do the following.

1. Create a directory called "shutdown".

2. Copy sqlfire.properties into that directory

sqlfire.auth-provider=BUILTIN
sqlfire.sql-authorization=true
sqlfire.user.admin=adminpassword
sqlfire.user.locatoradmin=adminpassword
sqlfire.user.serveradmin=adminpassword

3. Change into the shutdown directory and run a script as follows.

sqlf shut-down-all -locators=localhost[41111] -user=admin -password=adminpassword

This will only shutdown the members you would still need to stop the locator if you wanted to do that.

Wednesday, 17 July 2013

Securing Schemas in Pivotal SQLFire

Authentication/authorization is off by default in SQLFire if servers have not specified it. That is why SQLFire connections don't need a user/password if not configured explicitly on server. Once authorization has been configured on server, users will have access to only their database objects. This is illustrated below. In this example we use PLAIN text for the passwords just to illustrate how we secure a schema

1. Create a locator script as follows

sqlf locator start -user=admin -password=adminpassword -sqlfire.sql-authorization=true -peer-discovery-address=Pas-Apicellas-MacBook-Pro.local -peer-discovery-port=41111 -client-bind-address=Pas-Apicellas-MacBook-Pro.local -client-port=1527 -dir=locator

2. Create a server start script as follows

sqlf server start $JVM_OPTS -user=admin -password=adminpassword -sqlfire.sql-authorization=true -max-heap=$MAXHEAP -initial-heap=$MINHEAP -critical-heap-percentage=85 -eviction-heap-percentage=75 -server-groups=MYGROUP -conserve-sockets=false -locators=Pas-Apicellas-MacBook-Pro.local[41111] -client-bind-address=Pas-Apicellas-MacBook-Pro.local -client-port=1528 -dir=server1

Note: Ensure that a sqlfire.properties file exists for the locator/server as shown below.

sqlfire.properties

sqlfire.auth-provider=BUILTIN
sqlfire.sql-authorization=true
sqlfire.user.admin=adminpassword
sqlfire.user.locatoradmin=adminpassword
sqlfire.user.serveradmin=adminpassword

3. As the admin using connect with sqlf and create a database user account which will be the schema owner.

  
sqlf> connect client '127.0.0.1:1527;user=admin;password=adminpassword';
sqlf> CALL SYS.CREATE_USER('pas', 'pas');
Statement executed.

4. Connect with a JDBC client as shown below.

  
private Connection getConnection() throws SQLException
 {
  Connection conn = null;
  conn = DriverManager.getConnection
      ("jdbc:sqlfire://127.0.0.1:1527/", "pas", "pas");
  return conn; 
 }

When the JDBC connection connects successfully it will then be in the context of the schema "pas" and will have full rights / access to that schema to be able to create database objects etc.

If we provide an invalid username/password we get an error as follows, hence securing access to the schema with a built in username we created at step #3 above.

Exception in thread "main" java.sql.SQLNonTransientConnectionException: Connection authentication failure occurred. Reason: userid 'pas' or password invalid.

Pivotal HD - Talking true SQL on HDFS

Pivotal HD is an Apache Hadoop distribution that natively integrates the industry-leading Pivotal Greenplum massively parallel processing (MPP) database technology with the Apache Hadoop framework. See this link for it's release information. http://www.gopivotal.com/pivotal-products/pivotal-data-fabric/pivotal-hd

In this blog entry I want to detail, how we bring a full compliant ansi standard SQL into hadoop. The two components of Pivotal HD critical for this are as follows.

HAWQ - SQL query processor based on GPDB running on HDFS

HAWQ is a parallel SQL query engine that combines the merits of the Greenplum Database Massively Parallel Processing (MPP) relational database engine and the Hadoop parallel processing framework. HAWQ supports SQL and native querying capability against various data sources in different popular formats. It provides linear scalable storage solution for managing terabytes or petabytes of data at low cost

GFXF - Extension Framework component of HAWQ to create external tables

GPXF is the external table interface to HAWQ that enables querying data stored in HDFS, HBase, and HIVE. In addition, it exposes an API that enables a parallel connection to additional data sources.

Example

So here is a very simple example of how this works. We already have an existing Pivotal HD install with the HAWQ and GPXF components.

1. First lets generate some data using a simple script as follows

for i in {1..1000000}
do
echo "$i|person$i"
done

2. Create the file person.txt as shown below and verify we have 1 million person entries

  
[gpadmin@pivhdsne pas]$ ./data2.sh > person.txt
[gpadmin@pivhdsne pas]$ cat person.txt | wc -l
1000000
[gpadmin@pivhdsne pas]$ tail person.txt 
999991|person999991
999992|person999992
999993|person999993
999994|person999994
999995|person999995
999996|person999996
999997|person999997
999998|person999998
999999|person999999
1000000|person1000000

3. Now lets copy the file person.txt onto the HDFS as shown below

hadoop fs -put person.txt /

4. At this point we have the data file on HDFS so we can create an external table using HAWQ so we can query the data using SQL. To create an external table we would use SQL as follows

  
demo=# CREATE EXTERNAL TABLE person ( id int, name text)
demo-# LOCATION ('gpxf://pivhdsne:50070/person.txt?Fragmenter=HdfsDataFragmenter') FORMAT 'TEXT' (DELIMITER = '|');
CREATE EXTERNAL TABLE
Time: 3.843 ms

GPXF enables External table interface inside HAWQ to read data stored in Hadoop ecosystem which enables loading and querying data stored in

HDFS
HBase
Hive

The supported data formats for GPXF include the following.

Text
Avro
Hive - Text, Sequence and RCFile formats
HBase

Note: GPXF has an extensible framework API to enable custom connector development for other data sources and custom formats

5. Now at this point we can query data sitting on our HDFS using SQL , some examples below.

- Getting a count(*)

  
demo=# select count(*) from person;
  count  
---------
 1000000
(1 row)

Time: 1396.451 ms

- Querying the first 10 records only

  
demo=# select * from person limit 10;
 id |   name   
----+----------
  1 | person1
  2 | person2
  3 | person3
  4 | person4
  5 | person5
  6 | person6
  7 | person7
  8 | person8
  9 | person9
 10 | person10
(10 rows)

Time: 263.435 ms

More Information

Fore more information on Pivotal HD see the links below.

http://www.gopivotal.com/pivotal-products/pivotal-data-fabric/pivotal-hd

http://www.greenplum.com/products/pivotal-hd

http://pivotalhd.cloudfoundry.com/index.html

Tuesday, 25 June 2013

Greenplum GPLOAD with updates/merge ability

“gpload” is a data loading utility that acts as an interface to Greenplum Database’s external table parallel loading feature. The Greenplum EXTERNAL TABLE feature allows us to define network data sources as tables that we can query to speed up the data loading process. Using a load specification defined in a YAML formatted control file, “gpload” executes a load by invoking the Greenplum parallel file server (gpdist) – Greenplum’s parallel file distribution program, creating an external table definition based on the source data defined, and executing an INSERT, UPDATE or MERGE operation to load the source data into the target table in the database.

In this example we show how we can load 500,000 records and then update those 500,000 records with GPLOAD.

1. Create a table to load the data into as shown below.

drop table if exists apples.people cascade;

create table apples.people (id int, name text)
DISTRIBUTED BY (id);

2. The data we are loading is defined as follows

[Tue Jun 25 22:39:24 gpadmin@:~/demos/gpload/inserts ] $ head person.txt
1,person1
2,person2
3,person3
4,person4
5,person5
6,person6
7,person7
8,person8
9,person9
10,person10

3. Create a YAML formatted control file to load the data into the table PEOPLE as shown below.

test.yml

  
VERSION: 1.0.0.1
DATABASE: gpadmin
USER: pas
HOST: 127.0.0.1
PORT: 5432
GPLOAD:
    INPUT:
     - SOURCE:
          LOCAL_HOSTNAME:
            - 127.0.0.1 
          PORT: 8100
          FILE: [ /Users/gpadmin/demos/gpload/inserts/person.txt ]
     - COLUMNS:
         - "id":
         - "name": 
     - FORMAT: text
     - DELIMITER: ','
     - ENCODING: 'UTF8'
     - NULL_AS: ''
     - ERROR_LIMIT: 100000
     - ERROR_TABLE: apples.err_people
    OUTPUT:
     - TABLE: apples.people 
     - MODE: INSERT

Now at this point lets update the same data set with new updated data , once again using gpload BUT this time the YAML control file is a little different here.

5. The data we are UPDATING is defined as follows

[Tue Jun 25 22:54:39 gpadmin@:~/demos/gpload/upserts ] $ head people.txt
1,person1-updated
2,person2-updated
3,person3-updated
4,person4-updated
5,person5-updated
6,person6-updated
7,person7-updated
8,person8-updated
9,person9-updated
10,person10-updated

6. Create a YAML formatted control file to merge/update the data into the table PEOPLE as shown below.

test.yml

  
VERSION: 1.0.0.1
DATABASE: gpadmin
USER: pas
HOST: 127.0.0.1
PORT: 5432
GPLOAD:
    INPUT:
     - SOURCE:
          LOCAL_HOSTNAME:
            - 127.0.0.1 
          PORT: 8100
          FILE: [ /Users/gpadmin/demos/gpload/upserts/people.txt ]
     - COLUMNS:
         - id: int 
         - name: text 
     - FORMAT: text
     - DELIMITER: ','
     - ENCODING: 'UTF8'
     - NULL_AS: ''
     - ERROR_LIMIT: 100000
     - ERROR_TABLE: apples.err_people
    OUTPUT:
     - TABLE: apples.people 
     - MODE: MERGE
     - MATCH_COLUMNS:
           - id
     - UPDATE_COLUMNS:
           - name 
     - MAPPING:
         id: id 
         name: name

8. Finally verified we indeed have updated the data in the PEOPLE table.

  
gpadmin=# select * from apples.people limit 10;
 id |       name       
----+------------------
  1 | person1-updated
  2 | person2-updated
  3 | person3-updated
  4 | person4-updated
  5 | person5-updated
  6 | person6-updated
  7 | person7-updated
  8 | person8-updated
  9 | person9-updated
 10 | person10-updated
(10 rows)

Thursday, 13 June 2013

Creating a multi threaded insert client for SQLFire 1.1

In this example below we show how to create a multi threaded insert client to insert 1,000,000 records into SQLFire table. In this example below the table is partitioned with synchronous persistence turned on.The distributed system includes one locator and 5 data members.

1. Create Table as shown below

  
drop diskstore store1;

CREATE DISKSTORE STORE1;

drop table person;

create table person 
(id int primary key,
 name varchar(40))
PARTITION BY COLUMN (id)
REDUNDANCY 1
PERSISTENT 'STORE1' SYNCHRONOUS;

2. Multi Threaded Insert client Code.

  
package com.pivotal.fe.test;

import java.io.IOException;
import java.net.URL;
import java.sql.Connection;
import java.sql.DriverManager;
import java.sql.PreparedStatement;
import java.sql.SQLException;
import java.util.ArrayList;
import java.util.List;
import java.util.Properties;
import java.util.concurrent.Callable;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.Future;
import java.util.concurrent.TimeUnit;

public class MultiThreadInsert 
{
 private String url;
 private String driverClassName;
 private int RECORDS;
 private int COMMIT_POINT;
 private int nThreads;
 
 public MultiThreadInsert() throws IOException 
 {
  loadProperties();
 }
 
 public void loadProperties() throws IOException
 {
     Properties props = new Properties();
     URL propertiesUrl = ClassLoader.getSystemResource("sqlfiretest.properties");
     props.load(propertiesUrl.openStream());
     
     url = (String) props.getProperty("url");
     driverClassName = (String) props.getProperty("driverclassname");
     RECORDS = Integer.parseInt((String) props.getProperty("records"));
     COMMIT_POINT = Integer.parseInt((String) props.getProperty("commit_point"));
     nThreads = Integer.parseInt((String) props.getProperty("nThreads"));
     
 }
 
 @SuppressWarnings("unchecked")
 public void start() throws InterruptedException, SQLException 
 {
  System.out.printf("Starting %d threads for %d records connecting to %s\n", nThreads, RECORDS, url);
  
        final ExecutorService executorService = Executors.newFixedThreadPool(nThreads);

        ArrayList list = new ArrayList();
        for (int i = 0; i < nThreads; i++) {
            list.add(new RunData(url, driverClassName, i+1));
        }
        long start = System.currentTimeMillis();
        
        List<Future<?>> tasks = executorService.invokeAll(list, 5, TimeUnit.MINUTES);
        
        for(Future<?> f : tasks){
         try {
    f.get();
   } catch (ExecutionException e) {
    // TODO Auto-generated catch block
    e.printStackTrace();
   }
        }
        
     long end = System.currentTimeMillis() - start;
     
     float elapsedTimeSec = end/1000F;

        System.out.println(String.format("Elapsed time in seconds %f", elapsedTimeSec));
        
     executorService.shutdown();
        System.exit(0);
 }
 
 private class RunData implements Callable 
 {
     int counter = 0;
        int increment;
        Connection conn;
        String url;
        String driverClassName;
        
        private RunData(String url, String driverClassName, int increment) 
        {
            this.increment = increment;
            this.url = url;
            this.driverClassName = driverClassName;
        }

     public Connection getConnection() throws SQLException, ClassNotFoundException
     {
      Class.forName(driverClassName);
      Connection conn = null;
      conn = DriverManager.getConnection(url);
      //System.out.println("auto commit = " + conn.getAutoCommit());
      return conn; 
     }
     
        public void run() 
        {
      PreparedStatement stmt = null;
      String sql = "insert into person values (?, ?)";
      int counter = 0, size = 0;
      long startTime, endTime;
      
            int dataSize = RECORDS / nThreads;
            try 
            {
    conn = getConnection();
   } 
            catch (Exception e1) 
   {
    // TODO Auto-generated catch block
    e1.printStackTrace();
   }
            
            System.out.printf("Start: %d  End: %d \n",(dataSize * (increment - 1)), (dataSize * increment));
      try 
      {
       stmt = conn.prepareStatement(sql);
       
       startTime = System.currentTimeMillis();
       for (int i = (dataSize * (increment - 1)); i < (dataSize * increment); i++)
       {
        counter = counter + 1;
        size = size + 1;
        stmt.setInt(1, i);
        stmt.setString(2, "Person" + i);
        stmt.addBatch();
        
        if (counter % COMMIT_POINT == 0)
        {
         stmt.executeBatch();
         endTime = System.currentTimeMillis();
         System.out.printf("Insert batch of %d records took | %d | milliseconds\n", size, (endTime - startTime));
         startTime = System.currentTimeMillis();
         size = 0;
        }
       }
       
       /* there might be more records so call stmt.executeBatch() prior to commit here */
       stmt.executeBatch();
       
       //System.out.printf("Number of records submitted %d.\n", counter);
       
                 
      } 
   catch (SQLException e) 
   {
    // TODO Auto-generated catch block
    e.printStackTrace();
   }
   finally
   {
    if (stmt != null)
    {
     try 
     {
      stmt.close();
     } 
     catch (SQLException e) 
     {
      // TODO Auto-generated catch block
      e.printStackTrace();
     }
    }
   }      
        }
        
        public Object call() throws Exception 
        {
            run();
            return counter;
        }
  
 }

 /**
  * @param args
  * @throws InterruptedException 
  * @throws SQLException 
  * @throws IOException 
  */
 public static void main(String[] args) throws InterruptedException, SQLException, IOException 
 {
  // TODO Auto-generated method stub
  MultiThreadInsert test = new MultiThreadInsert();
  test.start();
 }
}

3. Output when run as follows.

Note: This was run on my MAC laptop which had 5 cache servers running on it. This would perform much better if I had 5 physical machines for each of the SQLFire cache server members.

[Fri Jul 05 22:33:47 papicella@:~/vmware/ant-demos/sqlfire/performance-test ] $ ant
Buildfile: /Users/papicella/vmware/ant-demos/sqlfire/performance-test/build.xml

init:

compile:

package:

run-insert-client:
     [echo] Starting insert client with jvm args : -server -showversion -Xms512m -Xmx512m
     [java] java version "1.6.0_37"
     [java] Java(TM) SE Runtime Environment (build 1.6.0_37-b06-434-11M3909)
     [java] Java HotSpot(TM) 64-Bit Server VM (build 20.12-b01-434, mixed mode)
     [java]
     [java] Starting 8 threads for 1000000 records connecting to jdbc:sqlfire://127.0.0.1:1527/;single-hop-enabled=true
     [java] Start: 875000 End: 1000000
     [java] Start: 125000 End: 250000
     [java] Start: 625000 End: 750000
     [java] Start: 500000 End: 625000
     [java] Start: 750000 End: 875000
     [java] Start: 375000 End: 500000
     [java] Start: 250000 End: 375000
     [java] Start: 0 End: 125000
     [java] Insert batch of 10000 records took | 1541 | milliseconds
     [java] Insert batch of 10000 records took | 1544 | milliseconds
     [java] Insert batch of 10000 records took | 1548 | milliseconds
     [java] Insert batch of 10000 records took | 1551 | milliseconds
     [java] Insert batch of 10000 records took | 1552 | milliseconds
     [java] Insert batch of 10000 records took | 1554 | milliseconds
     [java] Insert batch of 10000 records took | 1560 | milliseconds
     [java] Insert batch of 10000 records took | 1568 | milliseconds
     [java] Insert batch of 10000 records took | 552 | milliseconds
     [java] Insert batch of 10000 records took | 559 | milliseconds
     [java] Insert batch of 10000 records took | 563 | milliseconds
     [java] Insert batch of 10000 records took | 576 | milliseconds
     [java] Insert batch of 10000 records took | 581 | milliseconds
     [java] Insert batch of 10000 records took | 618 | milliseconds
     [java] Insert batch of 10000 records took | 600 | milliseconds
     [java] Insert batch of 10000 records took | 600 | milliseconds
     [java] Insert batch of 10000 records took | 630 | milliseconds
     [java] Insert batch of 10000 records took | 649 | milliseconds
     [java] Insert batch of 10000 records took | 643 | milliseconds
     [java] Insert batch of 10000 records took | 656 | milliseconds
     [java] Insert batch of 10000 records took | 687 | milliseconds
     [java] Insert batch of 10000 records took | 705 | milliseconds
     [java] Insert batch of 10000 records took | 705 | milliseconds
     [java] Insert batch of 10000 records took | 665 | milliseconds
     [java] Insert batch of 10000 records took | 682 | milliseconds
     [java] Insert batch of 10000 records took | 612 | milliseconds
     [java] Insert batch of 10000 records took | 712 | milliseconds
     [java] Insert batch of 10000 records took | 630 | milliseconds
     [java] Insert batch of 10000 records took | 646 | milliseconds
     [java] Insert batch of 10000 records took | 656 | milliseconds
     [java] Insert batch of 10000 records took | 661 | milliseconds
     [java] Insert batch of 10000 records took | 686 | milliseconds
     [java] Insert batch of 10000 records took | 475 | milliseconds
     [java] Insert batch of 10000 records took | 497 | milliseconds
     [java] Insert batch of 10000 records took | 499 | milliseconds
     [java] Insert batch of 10000 records took | 501 | milliseconds
     [java] Insert batch of 10000 records took | 534 | milliseconds
     [java] Insert batch of 10000 records took | 530 | milliseconds
     [java] Insert batch of 10000 records took | 534 | milliseconds
     [java] Insert batch of 10000 records took | 505 | milliseconds
     [java] Insert batch of 10000 records took | 534 | milliseconds
     [java] Insert batch of 10000 records took | 513 | milliseconds
     [java] Insert batch of 10000 records took | 535 | milliseconds
     [java] Insert batch of 10000 records took | 550 | milliseconds
     [java] Insert batch of 10000 records took | 504 | milliseconds
     [java] Insert batch of 10000 records took | 517 | milliseconds
     [java] Insert batch of 10000 records took | 528 | milliseconds
     [java] Insert batch of 10000 records took | 534 | milliseconds
     [java] Insert batch of 10000 records took | 523 | milliseconds
     [java] Insert batch of 10000 records took | 531 | milliseconds
     [java] Insert batch of 10000 records took | 517 | milliseconds
     [java] Insert batch of 10000 records took | 571 | milliseconds
     [java] Insert batch of 10000 records took | 598 | milliseconds
     [java] Insert batch of 10000 records took | 641 | milliseconds
     [java] Insert batch of 10000 records took | 652 | milliseconds
     [java] Insert batch of 10000 records took | 650 | milliseconds
     [java] Insert batch of 10000 records took | 498 | milliseconds
     [java] Insert batch of 10000 records took | 518 | milliseconds
     [java] Insert batch of 10000 records took | 542 | milliseconds
     [java] Insert batch of 10000 records took | 577 | milliseconds
     [java] Insert batch of 10000 records took | 557 | milliseconds
     [java] Insert batch of 10000 records took | 529 | milliseconds
     [java] Insert batch of 10000 records took | 550 | milliseconds
     [java] Insert batch of 10000 records took | 552 | milliseconds
     [java] Insert batch of 10000 records took | 525 | milliseconds
     [java] Insert batch of 10000 records took | 541 | milliseconds
     [java] Insert batch of 10000 records took | 544 | milliseconds
     [java] Insert batch of 10000 records took | 522 | milliseconds
     [java] Insert batch of 10000 records took | 537 | milliseconds
     [java] Insert batch of 10000 records took | 511 | milliseconds
     [java] Insert batch of 10000 records took | 554 | milliseconds
     [java] Insert batch of 10000 records took | 556 | milliseconds
     [java] Insert batch of 10000 records took | 513 | milliseconds
     [java] Insert batch of 10000 records took | 503 | milliseconds
     [java] Insert batch of 10000 records took | 468 | milliseconds
     [java] Insert batch of 10000 records took | 539 | milliseconds
     [java] Insert batch of 10000 records took | 581 | milliseconds
     [java] Insert batch of 10000 records took | 569 | milliseconds
     [java] Insert batch of 10000 records took | 541 | milliseconds
     [java] Insert batch of 10000 records took | 544 | milliseconds
     [java] Insert batch of 10000 records took | 626 | milliseconds
     [java] Insert batch of 10000 records took | 613 | milliseconds
     [java] Insert batch of 10000 records took | 662 | milliseconds
     [java] Insert batch of 10000 records took | 594 | milliseconds
     [java] Insert batch of 10000 records took | 644 | milliseconds
     [java] Insert batch of 10000 records took | 637 | milliseconds
     [java] Insert batch of 10000 records took | 675 | milliseconds
     [java] Insert batch of 10000 records took | 658 | milliseconds
     [java] Insert batch of 10000 records took | 608 | milliseconds
     [java] Insert batch of 10000 records took | 624 | milliseconds
     [java] Insert batch of 10000 records took | 538 | milliseconds
     [java] Insert batch of 10000 records took | 511 | milliseconds
     [java] Insert batch of 10000 records took | 484 | milliseconds
     [java] Insert batch of 10000 records took | 440 | milliseconds
     [java] Insert batch of 10000 records took | 462 | milliseconds
     [java] Insert batch of 10000 records took | 476 | milliseconds
     [java] Elapsed time in seconds 8.502000

BUILD SUCCESSFUL
Total time: 9 seconds

Tuesday, 14 May 2013

Naming Members in vFabric SQLFire

I find it useful to give a member a meaningful name. In SQLFire you could simply give each member a name by adding a property "name" as follows to the sqlfire.properties file for the member.

sqlfire.properties

# sqlfire.properties for data store or accessor member
license-serial-number=XXXXXXXXXXX
name=server1

Note: The same can be done with GemFire as well.

Then when the system is up the ID for each system member includes the given name as shown below.

  
sqlf> select substr(id, 1 , 35) as "Member" from sys.members;
Member                             
-----------------------------------
172.16.62.1(server2:38971)<v2>:4265
172.16.62.1(server1:38970)<v1>:1660
127.0.0.1(38744):29535             

3 rows selected

Thursday, 2 May 2013

JMX access to vFabric SQLFire

With the release of vFabric SQLFire 11 we can now start a JMX manager with the locator itself. To do that we add the following to the sqlfire.properties file of the locator itself.

jmx-manager=true
jmx-manager-start=true
jmx-manager-ssl=false
jmx-manager-http-port=8083

Then with the locator started we can verify we have it running on the default port of 1099 as shown below.

[Thu May 02 09:45:49 papicella@:~/sqlfire/vFabric_SQLFire_11_b40332/pasdemos/agent-test/locator ] $ netstat -an | grep 1099
tcp4       0      0 127.0.0.1.1099         127.0.0.1.64803        ESTABLISHED
tcp4       0      0 127.0.0.1.1099         127.0.0.1.64801        ESTABLISHED
tcp4       0      0 127.0.0.1.64801        127.0.0.1.1099         ESTABLISHED
tcp4       0      0 127.0.0.1.64803        127.0.0.1.1099         ESTABLISHED
tcp4       0      0 127.0.0.1.1099         127.0.0.1.64799        ESTABLISHED
tcp4       0      0 127.0.0.1.64799        127.0.0.1.1099         ESTABLISHED
tcp46      0      0 *.1099                 *.*                    LISTEN

Finally start jconsole and connect using a service URL as follows

Format:

service:jmx:rmi://{hotname}/jndi/rmi://{hostname}:1099/jmxrmi

Example:

service:jmx:rmi://Pas-Apicellas-MacBook-Pro.local/jndi/rmi://Pas-Apicellas-MacBook-Pro.local:1099/jmxrmi

Once connected you can browse the MBean as shown in the image below.

More Information

http://pubs.vmware.com/vfabric53/index.jsp?topic=/com.vmware.vfabric.sqlfire.1.1/manage_guide/jmx/jmx_intro.html

Wednesday, 1 May 2013

Explain Plan in vFabric SQLFire improved

With the recently released vFabric SQLFire 11 version the query execution plan is much easier to read then previously. An example below.

  
[Wed May 01 11:32:11 papicella@:~/sqlfire/vFabric_SQLFire_11_b40332/pasdemos/sqlfire ] $ sqlf
sqlf version 10.4
sqlf> connect peer 'bind-address=localhost;mcast-port=12333;host-data=false' as peerClient;
sqlf> explain select * from emp where deptno = 20;
MEMBER_PLAN                                                                                                                     
--------------------------------------------------------------------------------------------------------------------------------
ORIGINATOR 192.168.14.167(73118)<v6>:61492 BEGIN TIME 2013-05-01 11:32:39.735 END TIME 2013-05-01 11:32:39.777
DISTRIBUTION to &
Slowest Member Plan:
member 192.168.14.167(72048)<v1>:42223 begin_execution 2013-05-01 11:32:39.74 end_execution 2013-05-01 11:&
Fastest Member Plan:
member 192.168.14.167(72048)<v1>:42223 begin_execution 2013-05-01 11:32:39.74 end_execution 2013-05-01 11:&

3 rows selected
sqlf> select STMT_ID, STMT_TEXT from SYS.STATEMENTPLANS;
STMT_ID                             |STMT_TEXT                                                                                                                       
---------------------------------------------------------------------------------------------------------------------------------------------------------------------
00000001-ffff-ffff-ffff-000400000016| select * from emp where deptno = <?>                                                                                           

1 row selected
sqlf> explain '00000001-ffff-ffff-ffff-000400000016';
stmt_id 00000001-ffff-ffff-ffff-000400000016 SQL_stmt select * from emp where deptno = <?> begin_execution 2013-05-01 11:32:39.735 end_execution 2013-05-01 11:32:39.777
QUERY-SCATTER  execute_time 0.0 ms
  QUERY-SEND 
    RESULT-RECEIVE 
      SEQUENTIAL-ITERATION (0.38%) execute_time 0.136 ms returned_rows 5 no_opens 1
        RESULT-HOLDER  returned_rows 5 no_opens 1
          DISTRIBUTION-END (99.61%) execute_time 35.073 ms returned_rows 5
member 192.168.14.167(72048)<v1>:42223 begin_execution 2013-05-01 11:32:39.74 end_execution 2013-05-01 11:32:39.774
QUERY-RECEIVE 
  RESULT-SEND 
    RESULT-HOLDER  returned_rows 5 no_opens 1
      ROWIDSCAN (1.71%) execute_time 0.148 ms returned_rows 5 no_opens 1 node_details EMP : 
        CONSTRAINTSCAN (98.28%) execute_time 8.482 ms returned_rows 5 no_opens 1 scan_qualifiers None scanned_object APP.6__EMP__DEPTNO:base-table:APP.EMP scan_type  node_details WHERE : ((DEPTNO = CONSTANT:20) and true)

Monday, 15 April 2013

Handling DML Events Synchronously with vFabric SQLFire

SQLFire provides synchronous cache plug-in mechanisms to handle cache events. This example is a synchronous listener. A listener enables you to receive after-event notifications of changes to a table (insert, update and delete). Any number of listeners can be defined for the same table. Listener callbacks are called synchronously, so they will cause the DML operation to block if the callback blocks.

CommandTableEventCallBackListenerImpl.java

  
package pivotal.au.demo.poc.listener;

import java.sql.ResultSet;
import java.sql.SQLException;
import pivotal.au.demo.poc.domain.Command;
import pivotal.au.demo.poc.executor.ExecutorCommand;
import pivotal.au.demo.poc.executor.ExecutorFactory;

import com.vmware.sqlfire.callbacks.Event;
import com.vmware.sqlfire.callbacks.Event.Type;
import com.vmware.sqlfire.callbacks.EventCallback;

public class CommandTableEventCallBackListenerImpl implements EventCallback
{ 
 public void close() throws SQLException 
 {
 }

 public void init(String configuration) throws SQLException 
 {  
  System.out.println("configuration = " + configuration);
 
  System.out.println("CommandTableEventCallBackListenerImpl.init");
  
 }

 public void onEvent(Event event) throws SQLException 
 {
  if (event.getType() == Type.AFTER_INSERT)
  {
   ResultSet rset = event.getNewRowsAsResultSet();
   Command cmd = 
     new Command(rset.getInt(1), 
        rset.getString(2),
        rset.getString(3),
        rset.getString(4),
        rset.getString(5));
   
   System.out.println("Table[" + event.getTableName() + "] Command = " + cmd.toString());
   handleEvent(cmd); 
  }
  else
  {
   System.out.println("Not processing event " + event.getType().toString());
  }
  
 }
 
 private void handleEvent (Command cmd)
 {
  System.out.println("Handling event for Command with id = " + cmd.getId());
  
  ExecutorCommand execCommand = null;
  
  if (cmd.getType().equalsIgnoreCase("OS"))
  {
   execCommand = ExecutorFactory.getOSExecutorImpl();
   execCommand.runCommand(cmd.getCommand(), null);
  }
  else
  {
   // expecting to execute SQL so check if firing on sqlfire or greenplum at this stage
   execCommand = ExecutorFactory.getSQLExecutorImpl();
   if (cmd.getExecuteOnGreenplum().equalsIgnoreCase("Y"))
   {
    execCommand.runCommand(cmd.getCommand(), "GP");
   }
   
   if (cmd.getExecuteOnSqlfire().equalsIgnoreCase("Y"))
   {
    execCommand.runCommand(cmd.getCommand(), "SQLFIRE");
   }
  }

 }

}

Attach Listener to a table.

  
CREATE TABLE command_table 
(ID INT generated always as identity NOT NULL, 
 EXECUTE_ON_SQLFIRE VARCHAR(1) default 'N',
 EXECUTE_ON_GREENPLUM VARCHAR(1) default 'Y',
 command_type varchar(10),
 COMMAND VARCHAR(200) not null
 )
SERVER GROUPS (MYGROUP);

call sys.ADD_LISTENER('CommandTableEventCallBackListenerImpl', 'apples', 'command_table', 'pivotal.au.demo.poc.listener.CommandTableEventCallBackListenerImpl', '', 'MYGROUP');

More Information

http://pubs.vmware.com/vfabricNoSuite/index.jsp?topic=/com.vmware.vfabric.sqlfire.1.1/caching_database/cache-plug-ins.html

Thursday, 11 April 2013

ADO .NET c# Connection Pooling with vFabric SQLFire

Whether it's accessing a database using JAVA or in this case c# I always want to use a connection pool and in this example I show and simple way to do this with a c# ADO .NET client accessing SQLFire.

1. Add a reference to your Visual Studio project in the VMware.Data.SQLFire.dll. This DLL is installed in the

vFabric_SQLFire_11_bNNNNN\adonet\lib directory.

2. Reference the driver namespace in each source file where you want to use SQLFire components. For example, include this directive with all other references required in your application:

using VMware.Data.SQLFire;

3. Create a c# console application as follows

  
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using VMware.Data.SQLFire;
using System.Data;

namespace SQLFireDemo
{
    class QueryDemo
    {
        private string sqlfHost = "192.168.1.4";
        private int sqlfPort = 1527;

        public QueryDemo()
        {
        }

        private string GetConnectionString()
        {
            return string.Format(@"server={0}:{1}", sqlfHost, sqlfPort);
        }

        public void run()
        {
            using (SQLFClientConnection conn = new SQLFClientConnection(GetConnectionString()))
            {
                conn.Open();
                SQLFCommand command = new SQLFCommand
                      (string.Format("SELECT * FROM dept"), conn);
                SQLFDataReader reader = command.ExecuteReader();

                try
                {
                    StringBuilder row = new StringBuilder();
                    while (reader.Read())
                    {
                        Console.WriteLine(string.Format("Dept[deptno={0}, dname={1}]",
                                          reader.GetString(0),
                                          reader.GetString(1)));
                    }
                }
                catch (Exception e)
                {
                    Console.WriteLine(e);
                }
                finally
                {
                    reader.Close();
                }
            }

        }

        static void Main(string[] args)
        {
            QueryDemo test = new QueryDemo();
            test.run();
        }
    }
}

Output as follows

Dept[deptno=10, dname=ACCOUNTING]
Dept[deptno=20, dname=RESEARCH]
Dept[deptno=30, dname=SALES]
Dept[deptno=40, dname=OPERATIONS]
Dept[deptno=50, dname=MARKETING]
Dept[deptno=60, dname=DEV]
Dept[deptno=70, dname=SUPPORT]
Press any key to continue . . .

Monday, 8 April 2013

vFabric GemFire and the Native Client World using c#

Recently I had to step out of my comfort zone and learn how to create a c# client to access a GemFire 7 distributed system for a demo to a customer. OIf course there was more to it then just that but this outlines what you need to do to connect as a c# client to GemFire. I was using the following here.

Visual Studio 2012
GemFire 32 bit Naive Client

1. Install GemFire Native client 32 bit or 64 bit depending on your OS. It can be downloaded from the following location.

https://my.vmware.com/web/vmware/info/slug/application_platform/vmware_vfabric_gemfire/7_0

2. Once installed setup an ENV variable as shown below pointing to the location of the native client install.

C:\Windows\system32>echo %GFCPP%
C:\vFabric_NativeClient_32bit_7010

3. In your Visual Studio 2012 Project / Solution add a reference to GemFire DLL as shown below.

4. In Visual Studio 2012 create a cache.xml as shown below. This client cache is going to use a locator to connect to a cache server instance for the client itself.

xml/cache.xml

  
<?xml version="1.0"?>
<!DOCTYPE client-cache PUBLIC
    "-//GemStone Systems, Inc.//GemFire Declarative Caching 7.0//EN"
    "http://www.gemstone.com/dtd/cache7_0.dtd">

<client-cache>
  <pool name="client" subscription-enabled="true">
    <locator host="172.16.62.1" port="10334" />
  </pool>

  <region name="CommandRegion">
    <region-attributes refid="PROXY" pool-name="client">
    </region-attributes>
  </region>
  
  <region name="changeTrackingRegion">
    <region-attributes data-policy="normal" pool-name="client">
    </region-attributes>
  </region>
</client-cache>

5. Create 2 c# classes as shown below.

GemFireClient.cs

  
using GemStone.GemFire.Cache.Generic;
using System;
using System.Collections.Generic;
using System.IO;
using System.Linq;
using System.Text;
using System.Threading.Tasks;

namespace pivotal.au.company.poc
{
    class GemFireClient
    {
        private static bool isStarted = false;
        private static GemFireClient instance = new GemFireClient();
        private string configFileLocation = "xml/cache.xml";
        private Properties<string, string> properties = Properties<string, string>.Create<string, string>();
        private CacheFactory cacheFactory;
        private Cache cache;
        IRegion<string, string> ctrRegion;

        private GemFireClient()
        {
            Console.WriteLine("Reading properties file xml/cache.xml...");
            string clientCacheXml = getCacheConfigLocation(configFileLocation);
            properties.Insert("cache-xml-file", clientCacheXml);
            Serializable.RegisterPdxSerializer(new ReflectionBasedAutoSerializer());

            cacheFactory = CacheFactory.CreateCacheFactory(properties);
            cache = cacheFactory.Create();
  
            ctrRegion = cache.GetRegion<string, string>("changeTrackingRegion");
            ctrRegion.GetSubscriptionService().RegisterRegex("."); 
            
            Console.WriteLine("ctrRegion size = " + ctrRegion.Count);
        }

        public static GemFireClient getInstance()
        {
            return instance;
        }

        public void closeClientCache()
        {
            cache.Close();
            Console.WriteLine("Client Cache closed...");
        }

        public Cache getCache()
        {
            return cache;
        }

        private static string getCacheConfigLocation(string cacheXml)
        {
            var directoryName = Path.GetDirectoryName(System.Reflection.Assembly.GetExecutingAssembly().Location);
            if (File.Exists(System.Environment.GetEnvironmentVariable("COMPANY_CONFIG") + "/" + cacheXml) == true)
            {
                return System.Environment.GetEnvironmentVariable("COMPANY_CONFIG") + "/" + cacheXml;
            }
            else if (File.Exists(Path.Combine(directoryName, cacheXml)) == true)
            {
                return Path.Combine(directoryName, cacheXml);
            }
            else
            {
                throw new SystemException("Unable to find /" + cacheXml);
            }
        }
    }
}

GemFireTest.cs

  
using pivotal.au.company.poc.domain;
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using GemStone.GemFire.Cache.Generic;

namespace pivotal.au.company.poc
{
    class GemFireTest
    {
        GemFireClient gfClient;

        public void doInsert()
        {
            gfClient = GemFireClient.getInstance();

            // get command region
            IRegion<string, Command> commandRegion = gfClient.getCache().GetRegion<string, Command>("CommandRegion");

            // insert a Command Object into the region
            Command command = new Command();
            command.eventType = "INSERT";
            command.tableName = "Holiday";
            command.tableKey = "1";
            command.sequence = 31;
            command.payload = new Dictionary<object, object>()
                { 
                  {"Id", "1"},
               {"name", "apples"},
                  {"createdate", "10-10-2009"}
                };

            Console.WriteLine(command.ToString());

            commandRegion[command.tableKey] = command;

        }

        public void queryCommandRegion()
        {
            gfClient = GemFireClient.getInstance();

            Console.WriteLine("about to query commandRegion");

            QueryService<string, Command> queryService = gfClient.getCache().GetQueryService<string, Command>();
            Query<Command> qry = queryService.NewQuery("SELECT * FROM /CommandRegion");
            ISelectResults<Command> results = qry.Execute();
            SelectResultsIterator<Command> iter = results.GetIterator();
            while (iter.MoveNext())
            {
                Console.WriteLine(iter.Current.ToString());
            }

        }

        public void closeCache()
        {
            gfClient.closeClientCache();
        }

        public void run()
        {
            GemFireTest test = new GemFireTest();
            test.doInsert();
            test.queryCommandRegion();
            test.closeCache();
        }

    }
}

Output omitted but this should give you the general idea.

Search This Blog

Wednesday, 4 September 2013

More Information

Tuesday, 6 August 2013

Thursday, 1 August 2013

Friday, 26 July 2013

Friday, 19 July 2013

Wednesday, 17 July 2013

More Information

Tuesday, 25 June 2013

Thursday, 13 June 2013

Tuesday, 14 May 2013

Thursday, 2 May 2013

Wednesday, 1 May 2013

Monday, 15 April 2013

Thursday, 11 April 2013

Monday, 8 April 2013