NZPgUG/Past Events
Wellington Meetup July 21st 2016 at Catalyst
Can elephants queue?
Many projects involve something resembling job or message queuing. In this talk we will look at:
- the pros and cons of using a plain old relational database for this type of workload rather than specialized technology
- approaches and designs
- special considerations for PostgreSQL
- performance and tuning
- potential improvements
Using your existing PostgreSQL cluster to queue jobs and messages can be a reasonable choice in some cases. However, it's not without complications. This talk looks at queue-like workloads in detail. The topics covered will include:
- defining and categorizing queue-like structures and access patterns
- reasons to consider building queues inside your relational database
- reasons to avoid building queues inside your relational database
- example use cases
- locking strategies and concurrency
- higher transactions isolation levels
- vacuuming considerations and workarounds
- statistics and indexes
- benchmarking
- comparing PostgreSQL to some other RDBMSs
- some relevant potential future features for PostgreSQL
About the speaker: Thomas Munro joined the Database Server team at EnterpriseDB about a year ago, where he works on Postgres Plus Advanced Server and PostgreSQL. Previously he worked on mission critical software in the finance and software industries in the UK, New Zealand and France over the past two decades.
Wellington Meetup November 19th 2015 at Catalyst
Institutional spatial data management, with Postgis as the exemplar.
The presentation describes a paradigm shift in institutional spatial data management, due in part to Postgis, as well as covering issues pertaining to spatial data, and in particular to New Zealand. It also describes both open source and a mixed commercial/open source approach, which has generated significant uptake of FOSS GIS globally.
Presented by Brent Wood
Brent has been designing, building & using spatial databases since 1981. An open source enthusiast and user for over two decades; working as a data manager, DBA & gis analyst for much of this time, Brent has been using Postgres & Postgis for spatial data management since Postgis arrived on the scene, and through NIWA, where he works, has funded enhancements to open source spatial tools including Postgis, GDAL & QGIS, and has also contributed to GMT (Generic Mapping Tools).
He is a member of the NZOSS Council, and a director of the OSGEO Australasian Chapter.
Wellington Meetup October 15th 2015 at Catalyst
How to handle user-defined relational schema
School Student Management Systems (SMS), and probably many similar systems, contain an interesting theoretical problem, seemingly not handled well in RDBMS.
Individual database users have their own structured data requirements, to be imposed on entities already defined in the system elsewhere. For instance:
Given a "class_group" of "enrolled_students" (M2M, which will also record all the usual data and be linked to pastoral and attendance tables and so on), two separate "teachers" will require different fields, in their own schema. Alice the P.E. teacher might require a "shoe_size" (int) column, a "gym_fee" (float) column, and a "pool_safety_quiz" (text) column; Bob the science teacher wants a "mid_term_test" (integer) column, a "lab_coat_purchased" (Boolean) column, and a "notes" (text) column.
There are several ways to solve this, and most of them seem to be bad.
Hopefully we can come up with the best way!
Presented by Jonathan Harker. In his own words, "Jonathan is a six foot tall ape descendant from somewhere in the vicinity of Kapiti Island who still can't quite get the hang of Thursdays."
Wellington Meetup September 17th 2015 at Catalyst
September talk will be delivered by Finlay Thompson, operations manager at data science agency Dragonfly Science.
About the talk:
"We created a simple extension to the PostgreSQL database system that allowed us to streamline management of Bayesian model output, and create a powerful query engine that is used across a range of reporting.
Bayesian models are a powerful statistical tool, but create large output files that are difficult for analysts to manage. The Monte Carlo Markov chains used to fit the models generate thousands of simulations of parameters. While the simulations allow a great deal of flexibility in reporting, the overhead of managing the large data files usually frustrates this potential. Our PostgreSQL extension addresses this problem.
I will motivate the extension by describing the particular problem we faced at Dragonfly Science, and describe how we currently use PostgreSQL more generally in our work."
Wellington Meetup August 20th 2015 at Catalyst
A Brief Intro To PostgreSQL 9.5
About the talk: A brief introduction to the upcoming PostgreSQL 9.5 with demos of selected new features including ON CONFLICT (AKA "UPSERT"), ROLLUP/CUBE/GROUPING SETS, new JSON functionality, SKIP LOCKED and more depending on time.
About the speaker: Thomas Munro recently joined the Database Server team at EnterpriseDB where he works on Postgres Plus Advanced Server and PostgreSQL. Previously he worked on mission critical software in the finance and software industries in the UK, New Zealand and France over the past two decades.
Wellington Meetup July 16th 2015 at Catalyst
Openstack trove - Database as a service (DbaaS)
It seems that *-as-a-service is the new 'hot thing', and DbaaS is no exception. Catalyst's Openstack team are working on providing it as an add-on to Catalyst's Openstack cloud offering.
Come along and hear Mark Kirkwood (Catalyst's database and cloud storage guy) discuss the basics of DbaaS/what-is-Trove is and then delve a little into how it works.
Where: Level 3, Catalyst House, 150 Willis Street, Wellington
When: 6pm 16th July 2015
For more info please email miroslav at catalyst dot net dot nz
Wellington Meetup June 18th 2015 at Catalyst
Introductory meeting, getting to know what people are doing, and a presentation from Simon Nitz (DMS) about PostgreSQL/PostGIS and LINZ Maps: "6 months ago DMS implemented our IntraMaps web mapping software at LINZ, utilising their national data sets within a PostgreSQL 9.4 environment. IntraMaps is built on MapServer which is also Open source. Access to “LINZ Maps” is available for internal staff and contractors to access with an anticipated 4000 users."
Where: Level 3, Catalyst House, 150 Willis Street, Wellington
When: 6pm 18th June 2015
For more info please email miroslav at catalyst dot net dot nz
Auckland BOF/Meetup January 12th 2015 at LCA
Monday night (Jan. 12), at 7:30 PM, we will have a PostgreSQL 9.4 BOF at the LinuxConf.AU venue. Members of the Auckland PostgreSQL community who can't otherwise make it to LCA are allowed and encouraged to attend this BOF; please contact Josh Berkus so he can get a headcount for pizza.
Wellington Meetup January 20th 2015
We will be having a meetup from 6-9pm in downtown Wellington. Core team member Josh Berkus will present on PostgreSQL 9.4 features.
Please RSVP or there may be no food/drink for you!
Venue details:
LOCATION:
Clyde Quay Wharf, Te Aro, Wellington
That's the wharf between Te Papa and Oriental Parade
Halfway out on the wharf, there's a tunnel that connects the left and right sides. Walk through there and you'll see the condominium office beind glass doors, with a little private cinema. That's where we'll be. Just knock.
PARKING?
If driving, you can drive out the left side of the wharf, almost to the end, then turn around and park along the right side. Coin meter spots.
Or just park for free at New World and walk across the skate park.
LOST? QUESTIONS?
Contact Derek Sivers: 022 351 9100 - derek,at,sivers.org
Wellington Replication Tutorial January 20th 2015
Tuesday, January 20th from 9am to 3pm we will have an extended PostgreSQL replication tutorial, hosted by NIWA New Zealand in downtown Wellington. This will be the same tutorial Josh Berkus will give at LCA, only the extended 4.5 hour version which covers things like replication slots and performance tuning replication. Contact NIWA if you are interested in attending this; there is a fee.
