pgDay Asia 2016

Abstract:
TBD

Bio:
TBD

Abstract:
This paper will mainly talk about the scalability and performance improvements done in PostgreSQL 9.6

This paper will mostly focus on various scalability bottleneck and performance problem with 9.5 like ProcArrayLock for taking snapshot, ClogControlLock Contention and some other issues.

Paper will also highlights how these problems are solved in 9.6 and various other performance improvements done in 9.6

In Detail we will discuss about below topic:
1. ProcArrayLock Contention
2. ClogControlLock
3. Parallel Seq Scan
4. Freezing Large Table
5. Index Only scan with Partial indexes
6. Selectivity estimation for intarray.
7. Replace buffer manager spinlock with atomic operations
8. About Sorting Improvement

Bio:
Dilip is working for EnterpriseDB and has been with company since Dec 2015, He his working on Performance scalability work on Postgres. Previously he was working at Senior technical lead position in Huawei. He has worked in various databases (PostgreSql, In-house in memory database) for more than 9 years. Dilip has worked in various database development, maintenance projects. He has worked on optimizer, statistics collecton and executor module for in-house db. He has presented the paper on CSN based snapshot in PgCon 2015.

Abstract:
My talk includes our OLTP ( 300-500GB ) and OLAP (14 TB ) implementation of Postgresql @Inmobi.

Bio:
We are using Postgresql for our budget management system ( OLTP ) which is replicated to four data centers. It also includes all the monitoring and alerting systems ( TailnMail, Nagios, Pgbadger, Graphite, Diamond Logstash , Grafana etc ) that we are currently using to monitor all our production servers across four Data Centers. I will ensure to keep my talk very precise and wrap it up quick.

Abstract:
The traditional database executors are based on the fact that “I/O cost dominates execution”. These executor models are inefficient in terms of CPU instructions. Now most of the workloads fits into main memory, which is consequence of two broad trends :

A. Growth in the amount of memory (RAM) per node/machine.
B. Prevalence of high speed SSD.

So biggest bottleneck is CPU usage efficiency not I/O. So in order improve performance significantly, we are required to drastically reduce number of CPU instruction to be executed without impacting whole functionality. This is only possible with code specialization i.e. execute only code specific to given object/query/procedure instead of executing the generalized code with various branches. In this talk, I would like to explain how can we specialize the code for a given relation, some implementation details and also performance gain observed. Also I would highlight how code can be specialized for the procedure and also for query.

Bio:
I Kumar Rajeev Rastogi working with Huawei Technology since 2008. Prior to that I used to work with Aricent Technology for 3+ years. I have total experience of 10+ years.

I first encountered PostgreSQL in 2009, when tasked with integrating our in house "In Memory Storage" with PostgreSQL. From there on worked on to develop many features for our in house DB based on PostgreSQL. I am active PostgreSQL community members for many year and have contributed more than 8 patches to community, recent one being the "Index Scan optimization", which improved performance by up to 35%. Currently working on to apply Code specialization technology on PostgreSQL to gain huge performance improvement for our in house DB. I have more than 10 patents in DB domain. I have already presented papers in many earlier conference like PGCon, PGDay India

Abstract:
TBA

Bio:
TBA

Abstract:
How the VACUUM works actually, and why we need FREEZING to avoid XID wrap-around? This session will cover PostgeSQL MVCC, VACUUM, FREEZING, visiblity map. In addition, I will talk about some latest features related to VACUUM for next version PostgreSQL.

Bio:
Masahiko Sawada is a Database engineer in Japan, and providing technical support and database migration from other DBMS to PostgreSQL. And his primary interests are maintenance, replication and patch review.

Abstract:
Greenplum is a PostgreSQL fork, optimized for Analytics and Data Warehouse use cases. Pivotal announced in early 2015 that a number of products will go Open Source, one of them is Greenplum Database. This talk provides an overview over the history of Greenplum, the entire process of bringing the product into Open Source, all the stumbling blocks we ran into, and explains how contributors can participate.

Bio:
Andreas Scherbaum is working with PostgreSQL since 1997. He is involved in several PostgreSQL related community projects, member of the Board of Directors of the European PostgreSQL User Group and also wrote a PostgreSQL book (in German). Since 2011 he is working for EMC/Greenplum/Pivotal and tackles very big databases.

Abstract:
In this modern business today, we pace faster to serve customers and clients everyday. Banks, Telcos, and Retail Industries has the most Industries spending more budget on IT Infrastructure to help their businesses grow faster while keeping to stay competitive. Changing their transactional Database server into PostgreSQL has been the big option lately. It rocks because it improves performance, with better support and lower TCO. Using PostgreSQL, is a must in today's modern world. Why should we spend hundreds of thousands of dollars in CAPEX, while we can use OPEX instead with only a couple of thousands? It rocks because it improves the company's ability to spare more budget to marketing and becoming more competitive than its competitor. Using PostgreSQL for Transactional System is the Inevitable thing, it is very disruptive. It Rocks!

Bio:
TBA

Abstract:
The talk will look at Postgres;s native JSON Support. It will cover the postgresql JSON data type and functions and look at how Postgres can be used as a "no-sql" data store.

Bio:
Mark Clarke is a developer, security professional and trainer who works at Jumping Bean, an open source integration company, in Johannesburg South Africa. He has over 15 year experience of leveraging open source and free software to build robust solutions for customers. He is passionate about technology and creating a vibrant and innovative start-up industry and culture in the developing world using open source and free software.

Abstract:
One of the toughest challenges for data scientists is the speed at which their queries are processed when they are exploring for patterns in very large datasets. PostgreSQL has not traditionally been designed for such analytics, but the newly added TABLESAMPLE feature can be exploited to analyze big data very quickly. This talk is about using this awesome new feature to extract data statistics very quickly from datasets that would otherwise take a very long time to return with results.

Bio:
I am a PostgreSQL Consultant currently working with 2ndQuadrant, focusing on productization of PostgreSQL projects. My experience with PostgreSQL started with EnterpriseDB when the company was founded in 2004. I was primarily responsible for release/sustenance management to ensure timely and top quality delivery of the flagship product, Postgres Plus Advanced Server. After spending more than 7 years with the company, I moved to OpenSCG in 2011 for an onsite consulting role in USA. This role was focused on business intelligence using PostgreSQL and also migration of a few databases to PostgreSQL. I started working with 2ndQuadrant in 2015.

Abstract:
The talk will include the features and tuning that a DBA can do for performance enhancements of different kind of queries. On the basis of my experience with PostgreSQL, there are some right ways to write a query which can enhance the performance with a big difference in time. I will also discuss things which you should not do while writing a query.

* System tuning for better performance
* Right ways to write different kind of queries
* How to make query performance better
* What should be avoided while writing a query
* Role of a DBA for database performance

Bio:
https://in.linkedin.com/in/himanchali-7bb8282a

Abstract:
How PostgreSQL features and capabilities are effectively used by the business applications in the following areas -

- Replication slots impacting High Availability

How Replication slots (introduced in 9.4) have greatly helped our production high availability environment reduce the manual intervention of our technical resources to ensure business continuity

- PostgreSQL database maintenance features impacting database performance and scalability

How VACUUMING the databases has greatly improved database server performance post bulk loads

- Migrating databases and applications to PostgreSQL

Challenges faced migrating databases and applications to PostgreSQL from commercial databases like Oracle

Mitigating Development effort needed for database code migration and challenges encountered migrating Applications

Bio:
I am Venkata B, Nagothi with 10+ years of comprehensive experience as database consultant and currently working as PostgreSQL Database Consultant at Fujitsu, Australia.

Abstract:
Typical PostgreSQL shops will have a Pacemaker+Corosync/Heartbeat for a HA solution. It is a common and proven setup. However, is it easy to manage?

Bio:
Wei Shan is a database engineer with more than 4 years of experience. He has worked on majority of the relational databases before. He was working heavily on Oracle/MSSQL until 9 months ago where he got a new opportunity to work on PostgreSQL/MySQL full-time.

Abstract:
Less than 10 years ago, it could be a serious dispute between non-database computer engineers about which open-source RDBMS is better and faster: PostgreSQL or MySQL. Today it sounds ridiculously: now we compare PostgreSQL to industry-standard commercial databases, mostly to Oracle. Despite the fact that we can still learn something even from MySQL-derived databases, this development is clearly a result of massive learning from commercial databases and adopting enterprise features.v
On many PostgreSQL events last year, there were some talks and lots of offline discussions about Postgres current architecture limitations and lack some obvious features. As a consequence, more than one team was inspired and started to think about or even to implement proofs of concept.

So I figured that it is a good time first to give an overview of how PostgreSQL successfully learned new good things from commercial databases, from both technical and marketing perspective, and how that shaped the spread of PostgreSQL on enterprise market. Then I will discuss a set of concepts and features, in which the commercial databases are far ahead. These include not necessarily the trendy multi-master replication, but also more down-to-earth and tedious areas, such as undo management, I/O related issues, performance profiling, and even documentation. I will advocate why PostgreSQL needs these features, and how to approach naturally integrating them.

Bio:
Ilya is a CEO and consultant at PostgreSQL-Consulting LLC. His previous experience encompasses such commercial databases as Oracle and DB2. Currently, Ilya works primarily with PostgreSQL with a focus on its performance and effective use in production. He sees the mission of PostgreSQL in substituting the commercial databases in high-performance mission-critical applications.

Abstract:
An insight into PostgreSQL on IBM newly announced LinuxONE which explores the ultimate transactional database on IBM System z and learn how Power 8 can help to achieve the acceleration to business value for clients, ISV and developers through waitless computing by enhancing the performance of data intensive workloads. Power 8 is designed for superior performance for Big Data and Analytics from the processor, chip and memory, up to the system software. IBM is really focus on developing support for PostgreSQL because it encourages clients to move away from expensive database licensing. As IBM is focusing more on open source ISVs.

Bio:
Li Wen, WONG – ASEAN System z Brand Leader

Li Wen joined IBM ASEAN as System z Technical Specialist in 2006. She provided technical consultation to clients and Business Partners across ASEAN and India/South Asia with primary focus on System z performance analysis, capacity planning, server consolidation, and sizing for ISV products.v
In 2007, she was appointed to be the Technical Advocate and Subject Matter Expert for SAP on System z and was actively engaging the ASEAN customers to drive SAP on System z, providing technical consultation, solution design, configuration, and presentation support.

Since 2011, she has been driving new workloads focusing on Business Applications solution and new footprints on System z. She also leads the product management to support growth in the System z usage.

In 2015, she became the Server Solutions Country Manager for IBM Indonesia, overlooking IBM Mainframe, POWER and Linux Server.

In 2016, she returned to ASEAN region as the IBM Mainframe Brand Leader.

Abstract:
In this talk I will talk about doing KNN (nearest neighbor) search in Postgres with PostGIS and ordering the result based on more than just physical distance. This ordering is the core of the dating app Tantan, which I will use to illustrate the talk. At Tantan we have more than one million users every day that look through 200 million potential matches each day, and we need to optimize what we show to maximize the potential chance for a match. We do that by ordering the potential matches based on a number of criteria, such as distance from you, recent activity and popularity.

Bio:
Victor plays a major role in the backend team at the Chinese dating company Tantan, where he handles all parts from development in Go to database management of PostrgreSQL. He was part of the original development team and has first-hand experience facing the tremendous growth and challenges Tantan has encountered along the way.

Abstract:
InMobi is worlds largest independent mobile advertising network serving 150 billion ads per month. PostgreSQL is one of the data stores behind this scale and this talk is about various issues which we encountered in production and also how did we solve the same.

Bio:
The issues are categorised under various umbrella and we will deep dive into the same during the course of the talk.

1. Database User Control
2. Change Control
3. Replication
4. Partitions
5. Reliability Issues
6. Performance
7. Miscellaneous

Abstract:
TBD

Bio:
TBD

Abstract:
In recent years, the applicable enterprise area of PostgreSQL has been significantly enlarged by performance and functionality enhancement. As a result, there is a emerging demand to migrate database systems to PostgreSQL from Oracle.

In this presentation, we will report on the following two tools which greatly help our migration projects:
* SQL Imcompatibility detection tool (db_syntax_diff)
* Oracle emulating function libraries "orafce" and our enhancement

We are working at the open source software center (OSS center) at NTT which is the largest telecommunications company group in Japan. For the purpose of software cost reduction, we have been carried out a lot of database migration projects so far.

Most of our customers are preferring so called "DB migration project", in which only the middle-ware DBMS will be replaced and AP itself will be reused. This is because such projects are usually considered more effective in terms of development time and cost than the "rebuild" projects, in which the system will be build again from scratch. Nevertheless, the differences between old and new DBMSs often block the migration projects technically, and considerable number of projects may be giving up eventually.

We found the major factors that make the migration difficult is following two:
* Cost of the estimation work
* Technical difficulty of the estimation work itself

As a solution of these problems, we have developed a non-compatible SQL auto-detection tool named db_syntax_diff. This tool can extract PostgreSQL incompatible SQLs in AP for Oracle and append each of them a difficulty index for migration. We've performed more than 30 database migration projects using this tool.

As another effort to solve migration problems, we are also working on function enhancement of orafce (Oracle emulation libraries). Based on the knowledge of frequently seen incompatibilities in the above estimations, we have added a large number of new entries to the orafce.

Bio:
I'm dealing with PostgreSQL from 2014 at that company.
The business in one's charge :
* PostgreSQL support (inquiry correspondence, etc.)
* Update verification of PostgreSQL major / minor for support
Currently, we have conducted verification towards 9.5 support.

In addition, we have also tested products related to PostgreSQL. I've investigated pgpool-II and pacemaker as a representative of the HA cluster.

Abstract:
Upwork is the world’s largest freelance talent marketplace, with 10+MM freelancers and 4+MM clients, and pays out more than $1+B annually in salaries. The data processing pipeline at Upwork consists of PG production DB, GPDB data warehouse, and multiple Vitesse DB data marts. I will detail the distinctive role of each system, and provide details on the setup of the pipeline, specifically how data flow from production to DW to data mart. If time permits, I will also give some examples of how the platform works for the data scientists at Upwork.

Bio:
CK Tan is Founder at Vitesse Data, a database startup in Silicon Valley that provide data processing accelerators for Postgres, Greenplum DB and Spark. CK was Director of Data Services at Elance-oDesk (now Upwork) where he ran the data group. Prior to that, he was Sr Architect at Greenplum. Early in his career, CK worked on MPP database at Informix XPS. Before Informix, he worked as a research staff at UW-Madison Database Research on Shore Storage Manager and Exodus Storage Manager.

Abstract:
The talk will focus on PostgreSQL and Hadoop working together and how they can be used together to function better and perform advanced stuff.

Bio:
A PostgreSQL hacker

Abstract:
Using PostgreSQL for your data warehouse gives great benefits in performance, scale and cost. Simply switching your data warehouse or data mart from SQL Server or Oracle could save your organisation hundreds of thousands of dollars.

In this session we will explore features that make PostgreSQL a great DW solution, with a mix of presentation and demonstration. By the time the session ends, you will understand the technical and business benefits of PostgreSQL, as well as some cool integration tips for your first project.

Bio:
Ron Dunn is a data warehouse specialist with over 30 years of experience in the field. As well as working with PostgreSQL data warehouses through his Evangelist role at Ajilius, he runs the PostgreSQL Data Warehousing & Business Intelligence Meetup group in Australia. He has worked with companies across the Asia Pacific, ranging from small businesses to the largest conglomerates, and is firmly committed to the growth of PostgreSQL in enterprise computing.

Un-conference / Ligntning Talks

Lightning Talks!