Tech: DB2: an impressive new database release

This blog’s been slow because I attended a particularly good course on the new release of DB2 – called Viper on pre-release; now simply DB2 9. The main presenter was one
Boris Bialek, who was particularly knowledgable and entertaining.

Technology competitors are constantly leapfrogging each other. It’s hard to say that product X is consistently better than product Y when the following year, product Y brings out a new release that trumps competition.

Having said that, I think DB2 now has the edge that IBM’s competitors ( Oracle and Microsoft's SQL Server) will be struggling to match – Microsoft in particular, since they’ve only just released a major upgrade, SQL Server 2005.

In fact, IBM has made a realistic effort to keep Microsoft at bay with the release of an Express version, which is effectively free – but limited to 2 CPUs and 4 gigabytes of data. This is enough for them to prevent revenue leakage, but at the same time provide a small entry point for developers and small business.

In order of merit, the chief points about the new DB2 are:

• Native XML support – this is not the half-baked implementations of its competitors; it’s a true hybrid relational/XML database, storing XML documents intact – no shredding – and providing proper indexing to the XML fields. Each XML document is a field in a row.
• Autonomics – memory management is now largely automatic: just set the upper threshold of DB2’s entire memory needs, and its internal management will produce optimal results – among multiple instances – better than a DBA’s manual tuning efforts. It’s so good, IBM plan to get rid of all other memory management parameters.
• Free at entry level with Express versions
• Backup/restore – a host of improvements to handle partial/failure situations
• Range partitioning – ability to partition tables by key values
• Granular security – Label Based Access Control allows administrators to define access levels within tables
• granular compression – data compression can be defined up to the row level (note that this is not an exact equivalence to granular security
• other improvements – including capacity improvements at page level and below

Other databases will have some of these features already, but the true Native XML support is a first for a relational (non-specialised XML) database. The support for XPATH and XQuery structures is good – very good – as is XML schema support. All way better than anything currently on the market.

That hybrid model may cause some rethinking of the general concepts of relational databases. XQuery and SQL constructs can be embedded within each other, but you can’t precisely treat fields within XML documents as database fields – the document structures are too flexible. First Normal Form is instantly broken if tables and XML documents are treated as a continuum.

Interesting to see where this will all take us. And good to see the technology is there. Although this is invisible to most people, the world is already exploding with XML.