Weekly tweet digest for 2011-05-22

  • Contemplating odd note left by a neighbor reminding me I am loved paired with flowers taken from my yard. http://yfrog.com/gzxmljcpj #
  • The other side has foil stickers and <3s from the universe on it http://yfrog.com/h7oh7cjj #
  • #pgcon day 0 http://chesnok.com/u/29 #postgres #
  • Discussing whether global deadlocks are inevitable in multi-master replication #pgcon #clusteringsummit #
  • OH: "Order of commit is not the same thing as apparent order of execution. And we don't want to get into that right now." #pgcon #
  • having to remind myself that i don't need to correct people that are wrong on the internet. #
  • lovely meeting a few new faces, hearing ppl's impressions of portlandia and now turning in early the night before developer meeting #pgcon #
  • We still have a few more openings for lightning talks! Submit yours!! http://chesnok.com/u/1Z #pgcon #
  • OH: "If you think of it as a local data wrapper, it makes more sense." #pgcon #
  • OH: "We've now fixed our problem of being ahead of schedule." #
  • Just updated the commitfest application for the 9.2 development season. Guess that makes it officiall! https://commitfest.postgresql.org/ #
  • Looking at http://reorg.projects.postgresql.org/index.html #pgcon #
  • OH: "User-defined demons." #advancedfeatures #
  • mmm. scotch. #
  • Next year, @magnushagander is in charge of weather. That means rain again #pgcon #
  • Watching @pramsey keynote at #pgcon http://flic.kr/p/9JX9Ds #
  • "another thing for Canadians to be smug about. We invented GIS!" (re the Canada Land Inventory) -@pramsey #pgcon #
  • Announcing Postgres Open! September 14-16 in Chicago http://chesnok.com/u/2c #
  • "PostGIS is our biggest, highest profile third-party extension" that pushes the limits of Postgres #pgcon #
  • oops, we've all been talking about @pwramsey #pgcon #
  • Now @pwramsey sharing why PostGIS didn't need to be part of Postgres, but could operate well as an external extension. #pgcon #
  • "another thing for Canadians to be smug about. We invented GIS!" (re the Canada Land Inventory) -@pwramsey #pgcon #
  • +1 RT @robertmhaas: Scalability patterns: "New nodes are never free." -@robtreat2 #pgcon #
  • Lightning talks are looking awesome (5:30pm today!) #pgcon #
  • Updated PL Summit meeting agenda http://wiki.postgresql.org/wiki/PgCon_2011_PL_Summit#Tentative_Agenda #pgcon #
  • Getting ready for my talk http://www.pgcon.org/2011/schedule/events/366.en.html #pgcon #
  • Woo! Got a patch committed http://chesnok.com/u/2d #pgcon #postgres #
  • Almost time for lightning talks. First talk includes a live demo. In 5 minutes. Will it work? #danger #pgcon #
  • We will post all the lightning talk slides on the Postgres wiki in the morning. Now, to the party! #pgcon #
  • Lightning talks mostly uploaded (except the CBC talk that was slightly too large… working on it :) http://chesnok.com/u/2e #
  • Just heard from someone planning a 1 million table postgres instance. #pgcon #
  • SSI (serializable snapshot isolation) has never been implemented in production in another database. #postgres #pgcon #
  • "And therefore transaction one is its own grandfather.. and that just doesn't work." #pgcon #ssi #
  • "The people who wrote the SQL standard got this wrong. They got it wrong more than once." #ssi #pgcon #
  • pro-tip: a read-only transaction can create a cycle of dependencies that make a transaction sequence not serializable. #pgcon #
  • Basic idea: find a cycle in a graph, and break the transaction that caused it to preserve serializability. #ssi #pgcon #hopeigotthatright #
  • SSI rule is: don't let a transaction both have a rw-conflict *in* AND a rw-conflict *out* #ssi #pgcon #
  • SIREAD locks don't cause blocking (predicate locks are just flags). And they can persist beyond transaction commit. #pgcon #ssi #
  • Agenda for tomorrow's Procedural Language summit http://wiki.postgresql.org/wiki/PgCon_2011_PL_Summit #pgcon #
  • Almost the end of #pgcon :( #
  • "If it was hard to write it should be hard to use." -Jan Wieck #slony #pgcon #
  • "I don't want to have anything that says "release it" on the back." #janwieck #pgcon #
  • Enjoying @3dmashup's slides on PL/OpenCL, causing lots of conversation #pgcon #plsummit #
  • Great opening slides giving state-of PL/Python, status of PL/Perl, PL/TCL and PL/Lua. Next up PL/PHP! #pgcon #plsummit #
  • We're at Patty Bolands Pub #pgcon #
  • We're doing CREATE TABLE timetrials #pgcon #pubtrack #
  • Pubtrack #pgcon http://flic.kr/p/9KwiBF #
  • More details on the pubtrack at #pgcon http://chesnok.com/u/2i #

PgCon Pub Track: Learning more about Synchronous Replication

So, we’re at the Pub and doing “create a billion tables” time trials with Jan Urbanski using Python and Josh Berkus using Perl.

We’re also hacking on a test framework the Slony developers have, specifically hacking with Steve Singer. What we discovered is that sync rep doesn’t wait for a WAL segment to be *replayed* before it returns. In the pg_stat_replication table, we see sent_location, write_location and flush_location synchronized, but not replay_location.

This makes sense from a database perspective, but may be surprising behavior for application developers. There are patches out there (according to what I just heard from Bernd) to make synchronous replication wait for replay on the slave, but it’s not certain when that will be committed. It definitely won’t be part of version 9.1.

I just wrote up configuration details from a database administrator’s perspective, and am planning on doing some additional work to make a highly condensed configuration tutorial for our main docs. We definitely need to explain this more clearly for users, who might be thinking of it more from an application perspective.

twittering on 2011-05-21

  • Lightning talks mostly uploaded (except the CBC talk that was slightly too large… working on it :) http://chesnok.com/u/2e #
  • Just heard from someone planning a 1 million table postgres instance. #pgcon #
  • SSI (serializable snapshot isolation) has never been implemented in production in another database. #postgres #pgcon #
  • "And therefore transaction one is its own grandfather.. and that just doesn't work." #pgcon #ssi #
  • "The people who wrote the SQL standard got this wrong. They got it wrong more than once." #ssi #pgcon #
  • pro-tip: a read-only transaction can create a cycle of dependencies that make a transaction sequence not serializable. #pgcon #
  • Basic idea: find a cycle in a graph, and break the transaction that caused it to preserve serializability. #ssi #pgcon #hopeigotthatright #
  • SSI rule is: don't let a transaction both have a rw-conflict *in* AND a rw-conflict *out* #ssi #pgcon #
  • SIREAD locks don't cause blocking (predicate locks are just flags). And they can persist beyond transaction commit. #pgcon #ssi #
  • Agenda for tomorrow's Procedural Language summit http://wiki.postgresql.org/wiki/PgCon_2011_PL_Summit #pgcon #
  • Almost the end of #pgcon :( #
  • "If it was hard to write it should be hard to use." -Jan Wieck #slony #pgcon #
  • "I don't want to have anything that says "release it" on the back." #janwieck #pgcon #

twittering on 2011-05-20

Announcing Postgres Open

On behalf of the Postgres Open organizing committee, I’m pleased to share this announcement:

Postgres Open 2011, a conference for data innovators focused on disruption of the database industry through PostgreSQL, will take place September 14-16, 2011 at Chicago’s Westin Michigan Avenue hotel.

“PostgreSQL’s consistent addition of new features and enhancements, while remaining focused on reliability and performance, has provided myYearbook a solid foundation to create new and innovative applications,” said Gavin Roy, CTO at myYearbook. “We are looking forward to the Postgres Open Conference as a venue to share, network, and learn innovative ways to leverage Postgres in our environment.”

Postgres Open, a community-organized, non-profit conference, addresses the breadth of PostgreSQL usage, from core database system design to enterprise database use. Inviting entrepreneurs and technologists on the leading edge of data management, the conference will focus on open source database innovation and changes in the database market. Postgres Open includes regular talks, keynotes and hands-on tutorials.

We’re pleased to announce that VMWare and EnterpriseDB are joining the conference as founding sponsors.

The theme of the inaugural conference is “disruption of the database industry”. Topics will include new features in the latest version of PostgreSQL, use cases, product offerings and important announcements. Invited talks and presentations will cover many of the innovations in version 9.1, such as nearest-neighbor indexing, serializable snapshot isolation, and transaction-controlled synchronous replication. Vendors will also be announcing and demonstrating new products and services to enhance and extend PostgreSQL.

Postgres Open 2011′s main program (September 15-16) will be preceded by a day of intensive, half-day tutorials.

The Call For Papers for Postgres Open will open in late May.

Our program committee includes:
Robert Haas, Major Contributor, PostgreSQL committer,
Josh Berkus, Core Team member,
Greg Smith, Major Contributor to PostgreSQL and author of High
Performance PostgreSQL 9.0,
Gavin Roy, CTO of MyYearbook.com and
Selena Deckelmann, Major Contributor to PostgreSQL.

If you’d like to receive announcements as the conference progresses, please visit the website and add your email address to our list.

For information concerning sponsorship, please send email to sponsorship@postgresopen.org for a copy of our prospectus.

twittering on 2011-05-18

  • Discussing whether global deadlocks are inevitable in multi-master replication #pgcon #clusteringsummit #
  • OH: "Order of commit is not the same thing as apparent order of execution. And we don't want to get into that right now." #pgcon #
  • having to remind myself that i don't need to correct people that are wrong on the internet. #
  • lovely meeting a few new faces, hearing ppl's impressions of portlandia and now turning in early the night before developer meeting #pgcon #
  • We still have a few more openings for lightning talks! Submit yours!! http://chesnok.com/u/1Z #pgcon #

PgCon Day 1 – Cluster summit and catching up with folks

Yesterday, I spent my morning at the Clustering summit, catching up on what the cluster hackers have been up to for the last year. I was lucky enough to sit next to Jan Wieck and Kevin Grittner. You may remember Kevin from his work on serializable snapshot isolation.

There were some pretty awesome side conversations about where folks think work needs to be done next, and conflict resolution for multi- (or many-) master setups.

I gave a quick update on Bucardo 5, which had an alpha release last week, supports many-master and has has experimental support for non-Postgres targets. The first two targets are text and MongoDB.

The Postgres project has given the generic name “binary replication” to all the features like WAL shipping, streaming replication and synchronous replication. Simon Riggs also gave his update on these features at the Clustering Summit today. He observed that the 9.1 release is the culmination of 7 years of work on replication subsystems. Simon pointed out that synchronous replication is the best, and most obvious, use case for the binary replication at the core of Postgres. And also pointed out that he was quite pleased with the ultimate design.

For the afternoon, I spent some time with folks on the infrastructure team, giving Magnus well-deserved congratulations for his induction into -core, and meeting up with folks from all over at the Royal Oak and Keg, a reasonable steakhouse in town.

Looking forward to the developers meeting today!