Version 8.1 of the free and open source database PostgreSQL has just been launched, with a performance boost, and 120 new features and enhancements from almost 200 developers. The database now lets businesses run applications that were previously only possible over large commercial systems. Computerworld catches up with one of its key Australian-based developers, Gavin Sherry.
What is the single biggest change from 8 to 8.1?
There were so many great improvements. For your everyday user, there has been a lot of work on performance. For business users, two-phase commit and table partitioning now allows them to deploy applications which were until now only run on large commercial systems.
Has there been a shift in focus over the last year?
There has been a strong focus on performance. The bulk loading optimisations mean that people can import data into PostgreSQL 30 per cent faster. The Symmetric Multi-Processing (SMP) performance improvements and the implementation of a new kind of table scan logic for large tables have yielded impressive performance boosts for applications processing a lot of data in PostgreSQL.
Table partitioning allows users to segment large tables into smaller sections. This is vital for data-warehousing applications on PostgreSQL.
Have you had an increase in the number of developers working on this version?
Yes. PostgreSQL has been attracting more and more developers. We are seeing contributions from all over the globe. Increasingly, support for the project is coming from large international software and hardware vendors. It's great that PostgreSQL, as a project, allows such a varied mix of people to bring their ideas to a problem.
Version 8 had over a million downloads. How many downloads do you expect with version 8.1?
The popularity of PostgreSQL exploded last year, with the release of 8.0 and a Windows installer. That is only going to grow with the release of 8.1. With the amount of coverage we're getting in the industry and with a lot of new companies, including startups like EnterpriseDB and GreenPlum, I hope we can get more than two million downloads this time around.
Another measure of our success, however, is how many organisations are packaging our software into their solutions. PostgreSQL is used by virtually every major vendor in the world, in at least one of their products.
Why do you think there has there been such an accelerating interest?
Well, PostgreSQL has most of the features developers and users are looking for. Compared with the commercial databases out there, it is remarkably cheap to set up, administer and maintain.
Over the last few years, we've seen the infrastructure side of things falling into place. Companies like Fujitsu, SRA, Pervasive, CommandPrompt and others are offering professional services and support. At the same time, there are a lot of talented consultants across the world with many years experience in the industry, providing advice and assistance to companies and recommending PostgreSQL before anything else. The other factor is that PostgreSQL has what we call a business-friendly license. With PostgreSQL, you're free to do what ever you like: distribute it with your application, distribute it in-house, and modify it if you like. It doesn't matter. That kind of freedom is very attractive.
You said last year that you expected more enterprises to start trusting PostgreSQL with increasingly large and important data? Has this happened? How is it being received and utilised in the enterprise space?
It definitely has happened. A lot of the companies I've come into contact with over the last year are running their entire business out of PostgreSQL. This includes everything from basic content management for their Web site, to payroll and accounting, through time management, CRM and ERP. With PostgreSQL, they have a company-wide solution.
What scale of transaction processing does this version enable?
There are no real fundamental restrictions in PostgreSQL. We have many users doing over a terabyte (TB) who are very happy to see the features in 8.1, as it will allow them to get much greater throughput.
GreenPlum, which is producing a data warehousing version of PostgreSQL which harnesses the features of 8.1 to their own distributed database technology, are talking about databases in the range of 250TB.
What was the greatest challenge in developing version 8.1?
Some of the SMP bottlenecks were difficult to locate and work around.
Our biggest contributor, Tom Lane, really got stuck into the problem and came up with a fantastic solution.
PostgreSQL now has an impressive list of sponsors. How many of these have come on board in the last year? Are you actively seeking any more?
It's hard to say how many have come on board in the last year. A lot of supporters and sponsors sit in the background for a while and get a feel for the community and the code, to see where they might be of assistance and what they can do with the database in their businesses.
We anticipate that many more big-name companies will really get behind PostgreSQL over the next year. The thing is, we aren't seeking these companies - they come to us [In breaking news Sun has announced it will now distribute PostgreSQL].
Why do you develop PostgreSQL - what is its main appeal to you as a developer?
PostgreSQL has a fantastic architecture and design. The code is clean and a joy to read and add to. A lot of what I've been doing over the last year is researching a means to add multimeter replication to PostgreSQL. The code is always making me think about good, clean, simple design. This is what attracts developers to open source, I believe.
When do you anticipate 8.2 to come out?
In about 12 months. Look forward to even more features.
Gavin Sherry is a software developer based in Sydney. He is a major contributor to PostgreSQL. He and his company, Alcove Systems Engineering, provide professional services to IDG Communications, publisher of this Web site, and to Fujitsu Australia Software Limited.