#DB2-9 The DB2Night Show #131: DB2’s GOT TALENT Grand Finale!

This post is reference to my presentation in the Grand Finale of DB2’s Got Talent 2014 at DB2Night Show season. This was my last presentation in this season and the show theme was DB2 Success Story. I have selected one of my favourite topic i.e. Federated Database Performance. I have been working on the Federated setup since last 1.5 years. I faced many performance issues on production for which resolutions were difficult that is the reason I spent most of my time in searching and learning different ways for performance improvement in such distributed systems. I had given the presentation on Federated Query performance before on The DB2Night Show #128. This one is the extension of my previous presentation. You can check out the Federation basics and recap of the Episode #128 presentation at this link.

I discussed how distributed data across multiple databases can be accessed using the Federated database systems. Also specified the components of the federation like wrappers, servers, nicknames etc. Here are the few steps to setup the federated database.

CREATE WRAPPER NET8
LIBRARY ‘libdb2net8.a’ ;

CREATE SERVER ora_server TYPE oracle
VERSION 11.2.0 WRAPPER net8
OPTIONS (NODE ‘ORCL’);

CREATE USER MAPPING FOR userName
SERVER ora_server
OPTIONS (REMOTE_AUTHID ‘username’, REMOTE_PASSWORD ‘pwd) ;

CREATE NICKNAME ora_emp FOR ora_server.scott.emp;

 

#DB2-8 The DB2Night Show #130: DB2’s GOT TALENT Top 7 Finalists Compete!

This post is based on the Multi-Temperature Storage Management Feature Introduced in DB2 10.1. I presented this topic at DB2Night Show on March 21, 2014. The theme for this episode was Storage. Since I am obsessed with the new fabulous features introduced in this version, without any second thought I chose this topic for this round of Finale.

This episode decided our progression for the Grand Finale and fortunately based on votes I qualified for the Grand Finale.
Here is the quick overview of the Multi-Temperature Storage Management Feature.

Storage Groups:

Storage groups are the logical groupings of the automatic storage paths. The storage paths grouped by a particular are identified by the same storage path characteristics like latency, overhead, transfer-rate etc. Once the storage group is created the, we can assign the Automatic Storage tablespaces to the respective storage group. This association of the tablespaces with the storage group is completely dynamic and we can change it whenever required. Here is the visual illustration of how storage groups are formed.

 

 

 

Multi-Temperature Storage Management:

 

#DB2-7 The DB2Night Show #129: DB2’s GOT TALENT Top 9 Finalists Compete

This post is based on one of the ‘Index Jump Scan‘ – one of the terrific feature introduced in DB2 10.1. I presented the topic at DB2NightShow Finals-2. The theme for this episode was ‘DB2 Performance’ and I found this topic at the right time.
This was actually a courageous decision of presenting this topic in front of the Scott – one of the pioneer of this topic.
Still I tried my best to introduce the community with this topic and present a comparative study of the scenario without and with the Index Jump Scan. Here is the quick summary of my presentation.

Index Gap:
Being a DBA I always worked on the long ad-hoc queries which has been written for the reporting tools like actuate reports and Birt reports. It is always painful to tune these queries because of the number of predicate conditions and the composite indexes already designed on the tables involved in such queries. Frankly speaking my face used to be turned down when developers asked me to tune such a long queries for which I need to scroll my editor. 😉
Ideally query predicates should be consistent with that of the composite index on the column. For e.g

Q1: select FIRSTNME,MIDINIT,LASTNAME
           from EMPLOYEE where FIRSTNME=’CHRISTINE’
          and MIDINIT=‘I’ and LASTNAME=‘HAAS’;

I1:  create index EMPLOYEE_ID1 on EMPLOYEE
        ( FIRSTNME ASC,MIDINIT ASC, LASTNAME ASC)  ALLOW REVERSE SCAN;

I have query Q1 in which WHERE condition is specified on the columns FIRSTNME, MIDINIT and LASTNAME. For optimal performance I created a composite index I1 on combination of these columns. Such queries where the query predicates are consistent with that of the columns in the composite indexes such queries are said to be consistent. Such queries will provide optimal performance.
Unfortunately, this is not the case in most of the application scenarios. Queries are in-consistent with that of the composite indexes. Following is the sample example of the in-consistent queries.

Q2: select FIRSTNME, MIDINIT, LASTNAME
from EMPLOYEE where FIRSTNME=’CHRISTINE‘
and LASTNAME=’HAAS‘;

Predicates of the query Q2 are not consistent with index I1 since column MIDINIT is not there in the WHERE clause.
In such case the query is said to be in-consistent and it is said to have Index Gap. In our example, we are having the index gap for query Q2 on MIDINT column.

 

#DB2-6 The DB2Night Show #128: DB2’s GOT TALENT Top 10 Finalists Compete

This post is based on ‘Performance Tuning in Federated systems‘ presentation in DB2’s Got Talent 2014 at db2nightshow. This was my first presentation on federated systems. I worked closely on the federated systems so I wanted to share my experience with all the db2 community. Since there were too many things to tell I ran very fast while talking but overall I took it very good at the end. Here is an overview of Performance Tuning in Federated Systems.

What is Federation ?

1. Federation allows the users and applications to access the data from more than one RDBMS in the single           requests.

2. Its a special type of distributed database management system that allows the distributed request across            the multiple databases at the same time.

3. Federation consists of a DB2 instance, a database or federated database and one or more remote                     databases of same or different type. I work on the federated system with ORACLE as a remote data                  source in the federation.

How Federation Works:

Application can communicate with Federated server with any supported interface like JDBC or ODBC etc.