PGConf.Russia 2018

PostgreSQL upgrade is not as painful as it sounds

For the majority of System Administrators and DBAs performing an upgrade for RDBMS, let alone a major one, is a pain. That’s because one of the key factors that plays a role in a decision if and when to perform an upgrade is the downtime that it might come to during the process. This is true for any databases but especially important for those that are in production or under a high load.

Often, a major upgrade get’s cancelled and a DBA needs to go back to an older version due to the lack of experience or some basic errors that could have been easily avoided at the planning stage.

In our consultancy, we perform upgrades for our clients regularly and it allowed us to streamline the process and take some preventative measures that help us to perform it quickly, efficiently and with minimal or no downtime.

In this talk, I will share some key steps and tools that will help any DBA to become better at major upgrade performance. I will answer the following questions:

How to prepare for an upgrade of PostgreSQL? What one needs to do at the planning stage? How to plan your actions during the actual upgrade process? How to perform an upgrade successfully without going back to the older version? What actions one must perform following an upgrade?

I will also go through the two most popular processes of an upgrade: pg_upgrade и pg_dump/pg_restore, will compare some of the benefits and downfalls using each of these. I will also discuss some of the main issues one might face throughout the process and ways to avoid them.

This talk would be of interest to those who are new to PostgreSQL, as well as experienced DBAs who would like to learn more about upgrades or those who, in general, would like to understand why major upgrades should NOT be avoided like the plague.

pgpro_scheduler and cryptocurrency transactions

Apart from its main purpose of scheduling tasks, pgpro_scheduler can also deal with chained transactions. It can be used in various scenarios of asynchronous data processing.

This tutorial demonstrates pgpro_scheduler features that ensure secure processing of chained transactions. We'll be using cryptocurrency transactions as an example.

pgpro_scheduler is included into Postgres Pro Enterprise as an extension.

PostGIS and realtime systems

PostGIS is a spatial extension to PostgreSQL that enables spatial datatypes, access methods and a set of functions to perform geometric operations on them.

Typically PostGIS is used to select a small subset of a big static dataset. In this talk I'll cover issues that arise when working with big dynamic data flows, and ways to resolve them, on examples that we've met developing Juno ride sharing service backend.

VOPS: vectorized executor for Postgres

PostgreSQL looks very competitive with other mainstream databases on OLTP workload (execution of large number of simple queries). But on OLAP queries, requiring processing of larger volumes of data, DBMS-es oriented on analytic queries processing can provide an order of magnitude better speed. The following factors limit Postgres OLAP performance:

• Unpacking tuple overhead (tuple_deform)
• Interpretation overhead (Postgres executor has to interpret query execution plan)
• Abstraction penalty (support of abstract data types)
• Pull model overhead (operators are pulling tuples from heap page one-by-one, resulting numerous repeated accesses to the page)
• MVCC overhead (extra per-tuple storage + visibility check cost)

All this issues can be solved using vectorized executor, which proceed bulk of values at once. In this presentation I will show how vector operations can be implemented in Postgres as standard Postgres extension, not affecting Postgres core. The approach is based on introducing special types: tile types, which can be used instead of normal (scalar) types and implement vector operations. Postgres extension mechanism, such as UDT (user-defined type), FDW (foreign data wrappers), executor hooks are used to let users work with vectorized tables almost in the same way as with normal tables. But more than 10 times faster because of vector operations.