Filesystem I/O at the Linux Plumbers Conference

http://osdldbt.sourceforge.net/dl380/4disk/sraid10/ext3/read-write/
graph from software raid, RAID10, no partition table, ext3, read-write load

If you haven’t heard, the Linux Plumbers Conference is happening September 17-19, 2008 in Portland, OR. It’s a gathering designed to attract Linux developers – kernel hackers, tool developers and problem solvers.

A few of us that met through the Portland PostgreSQL User Group (PDXPUG) pitched an idea for a talk on filesystem performance. We wanted to examine performance conventional wisdom and put it to the test on some sweet new hardware, recently donated for performance testing Postgres. We’re asking questions like: Is RAID5 really the worst performing configuration for a database? How much does partition alignment really matter? Is there one Linux filesystem that a DBA should always choose for best performance under any load? Is adaptive readahead all that?

Our talk was accepted, so we’ve been furiously gathering data, and drawing interesting conclusions, ever since. Gabrielle Roth and I are presenting, using the results of extensive testing conducted by Mark Wong, a database benchmarking expert and author of pg_top. We’ll be sharing 6 different assumptions about filesystem performance, tested on five different filesystems, under five types of loads generated by fio, a benchmarking tool designed by kernel hacker Jens Axboe to test I/O.

PDXPUG Day on July 20 – Register now!

pgday 2007

photo courtesy of Dan Browning

Registration for PDXPUG Day on July 20, 2008 is open! Please sign up and let us know what size t-shirt you’d like. We’re requesting a $20 donation (by cash or check) at the door. All proceeds to to Software in the Public Interest, a 501(c)3 organization that is used to fund PostgreSQL advocacy.

Registration for OSCON is not required to attend.

Registering also gets you in the door at the Gotham Tavern, our after-party location close to the convention center!

Our line-up of talks includes:

PostgreSQL Unit Testing with pgTAP – David Wheeler
Inside the PostgreSQL Shared Buffer Cache – Greg Smith
Muldis D – Portable Databases At Full Power – Darren Duncan
A Streaming Database Talk – Rafael J. Fernández-Moctezuma
Using GLORP to connect Squeak Smalltalk to PostgreSQL – RandalSchwartz
Fighting Disease with PostgreSQL Full Text Search and JRuby on Rails – Mike Herrick
All Your GIS Are Belong to You – Abe Gillespie
What’s PgUS – Joshua Drake

Sign up today!

Call for proposals for PDXPUG PgDay, due June 20, 2008

pgday 2007
(Photo from PgDay 2007)

Please submit a talk! The call will be open for 2 weeks and proposals are due June 20th. Follow the link for details on submitting.

http://pugs.postgresql.org/node/400

PDXPUG PgDay will be on July 20, 2008. This is a one-day conference happening the day before OSCON at the Oregon Convention Center.

We are inviting anyone who has something interesting to share about PostgreSQL to send us a proposal!

We’d like to have at least one 1.5 hour tutorial and up to five 45-minute talks.

We welcome talks in any of the following areas:

* Case studies involving interesting and innovative uses of PostgreSQL from an application developer, PostgreSQL developer or administrative user perspective
* Converting from other databases to PostgreSQL
* Howtos for database administration tasks (partitioning, backups, replication, writing stored procedures)
* Practical advice on configuration, monitoring and database management

PgCon Lightning talk: User Groupalooza

I gave a Lightning talk today about PostgreSQL User Groups. I wasn’t able to get through ALL my slides – but I only had to rush through the last three. (click on the cat below to download – 5MB)

splash for user group talk

Lightning talks are some of my favorite sessions. I got to announce the incorporation of PostgreSQL-EU! We had a talk about DBIx::Cache (which you should all check out!), a cool open source lab in Japan that Hiroshi Saito works for (only for Japanese, but very cool), Gavin Roy talked about Staplr and a new benchmarking tool called Playr that was just released on Thursday, and six more talks! We hope to publish the rest of the slides shortly.

At PgCon in Ottawa, Day 1

Oops, this got published early. I’ll be updating this shortly.. probably over lunch.

Magnus posted some photos already!

Summary of last night:

We had a great time last night at the pub for registration, and then over to the Yahoo! party. We had fun talking databases, beer, and the difficulty in trying to understand Australian accents. I learned from Tom (not Tom Lane) that Adelaide, Australia has a fantastic wine-growing region. Alvarro once again recommended that Jeff commit his time interval data type. I think Jeff was convinced to do this for 8.4. Now I have to update the documentation!

ptop – the talk

ptop slide1

Check out the slides from my ptop talk. I was very happy with the end result of the presentation. I had a few people talk to me about the project afterward, including the author of innotop, a perl-based monitoring tool for MySQL. I’m sure we can steal some ideas from that project!

I haven’t gotten much done with ptop since I’ve been back. I’ll have some time after LUG Radio Live!

PostgreSQL Conference – Keynote commentary

Theo Schlossnagle wrote a great blog post about Joshua Drake’s keynote. I wanted to respond to some of his comments here.

I totally agree with Theo that we need to be disruptive! One criticism I have had in the past of the PostgreSQL community is that it has been too centralized, and not willing to experiment with social networking, wikis and other non-hierarchical tools that will get more *end* *users* actively contributing.

I think all of that is changing with the creation of the PostgreSQL-EU and USPgA groups, and the fast uptake in PostgreSQL User Groups in the last few months.

Also, I also respectfully disagree with some comments that were made in other sessions about “us not wanting” the type of users that choose MySQL. *I* want those users to come to my user group meetings. There’s tons we can learn from each other.

We will make PostgreSQL better if we hear and respond directly to criticism from users of the most popular open source database. I’m not saying that Tom Lane and Bruce Momjian need to do that. PUGs should be doing this, filtering out the good/important stuff and communicating information back to -hackers and -core. That’s a great service the user groups can provide to postgresql.org.

Finally, I am so excited about a Baltimore/Washington PUG!! I had several people approach me about a group. I think even the MySQL guy will be interested. 🙂

PostgreSQL Conference East Flickr group

DSC_0051.JPG

I created a Flickr group for the conference and added a few pictures today.

The weather in Maryland is just like Portland this week! Kind of drizzly and overcast. The sun peeked out for a few minutes at sunset, and I had a nice walk around the University of Maryland at College Park campus today.

I met several people in the UMIACS department, including my new hero, Mike, who created over 100 individual wireless network accounts for conference attendees. The rooms look great, and the foyer where we’ll be having breakfast is very pretty. UMD’s campus is just amazing and huge. If you’ve got an extra hour to spare, take a walk around!