Here's a short recap about my speaking engagements in June

AMOOCON

AMOOCON takes place from 4th to 6th June in Rostock, Germany. The conference has it's roots in the FOSS VoIP communities but has a more broad focus these days. I will hold a "full length" PostgreSQL advocacy talk as well as a 20 minute PostgreSQL 9.0 primer.

Netways OSDC

It's my second time at the OSDC, again in Nuremberg, Germany on 23rd & 24th of June. I'll hold a talk and give a short presentation on the native replication mechanisms PostgreSQL is going to provide with the upcoming 9.0 release.

As every year, there might be talks & presentations on these topics at the Metalab shortly before or after the two events; follow me on twitter or watch the event list for updates if you're interested in these.

So it's April by now, more than half a year since my last post. Maybe this blogging business was just a temporary thing after all. I guess my reluctance to post anything new was caused by the lack of definitives in my life as of lately, but this is hopefully changing in the near future.

As promised, here are the slides of my presentation I held at the Metalab, titled "PostgreSQL Performance: Eine Landvermessung".

And I just stumbled over this blog posting by Brendan Gregg who works for Sun's Fishworks team and was amazed by the level of detail that Solaris' instrumentation data provides. Good stuff!

FrOSCon

FrOSCon 2009 was a nice break from the stress at work, replacing it by stress in the weekend. The atmosphere was nice as usual and the planning good as every year. And with Andreas Scherbaum playing airport- and venue taxi the transportation didn't leave any room for improvement ;).

A few of the things that stuck with me were

Virtualization & Cloud Management

There's a lot of stuff going on in the Virtualization world, since by now everybody noticed that just hypervising things doesn't cut the slack and that you need to manage the stuff you deployed somehow. Which is a good thing, by the way.

The last two months were very interesting and positively demanding.

A catchy headline, as written by The Register. To quote more from the story (Full report with tech details):

The malware logs the magnetic-stripe data and personal identification number of cards used at an infected machine and provides an intuitive interface for retrieving the information using the ATM's receipt printer, [..] Since late 2007 or so, there have been at least 16 updates to the software, an indication that the authors are working hard to perfect their tool.

This is a nice example of what happens, when you ignore the things that are necessary to run an important area of your core business. The business area being the operation of the ATM machines (guess how bank teller utilization would look like if you throw out all ATMs). And a few of the things to run such a part competently would be: security (of the systems, the network), service lifecycle management and configuration management.

I just stumbled over an old architecture diagram from one of the projects I used to work on. The type of services and project in question are left as an exercise to the curious reader, since this is not the point of this posting.

What I wanted to show is, how complex multi-tiered applications can be these days, especially when you phase in new services or try to replace old ones by setting up the new services to run in parallel to the existing ones.

To quote Jeff Atwood in his blog:

Server Fault is a sister site to Stack Overflow, which we launched back in September 2008. It uses the same engine, but it's not just for programmers any more:

Server Fault is for system administrators and IT professionals, people who manage or maintain computers in a professional capacity. If you are in charge of ...
* servers
* networks
* many desktop PCs (other than your own)
... then you're in the right place to ask your question! Well, as long as the question is about your servers, your networks, or desktops you support, anyway.

Slony-I is a trigger-based replication solution which allows you to replicate database tables and sequences asynchronously from one master to several read-only slaves (which can also be cascaded).

Trigger-based means, that each table and sequence which gets replicated has triggers assigned, which will fire whenever the content of the given database object changes. The stored procedures, which are associated in the triggers, will then record the changes and store these in a replication log table. Separate daemons monitor the log table for changes and distribute the changes according to their defined rules.

This approach allows for extremely flexible setups, having different master servers for different tables, but this comes at a price.

As we've seen in our previous example, doing log shipping with pg_standby can be quite a hassle if you take your slave servers regularly online to use them for queries and then want to resume replication again.

The guys from Skype were probably faced by exactly the same problems when they decided to write walmgr.

If you're not familiar with log shipping I strongly suggest to read the previous post first.