It's finally here - Informix 11.7 (Panther) has been launched!  See the press release here: 

The new Panther Launch press release 

(old info: 11.7 Panther Launch)

 

If you have not been involved with any of the Panther EVPs and have not been present in any of the Panther presentations here is a quick rundown of what it provides:

 

Focus areas: 

 

Flexible Grid 

Warehouse 

Ease of use/Embeddability 

Application Development 

Security 

Performance 

 

What is Flexible Grid? 

Flexible grid allows you to build a grid of servers using different hardware, operating systems, and versions of Informix.  Just name the servers you want in your grid and give your grid and name and you are up and running.  Administration is a snap, letting you add or remove servers as an online operations.  You can also do administration through any node, just designate one or more nodes as your administration node(s) and from there you can add tables, alter tables, add dbspaces, run stored commands on all nodes in the grid.  You also get the option of replicating just schema changes or schema and data changes.  In addition to administration the flexible grid also allows you to spread your workload.  Since often your grid will be spread across a wide area, perhaps across the country or world, you are given the option to define an ordered list of servers to attach to.  This way you can favor servers geographically close to you, and if that node fails have your workload go to a node that is farther away.

 

Warehouse 

For warehouse we have moved a number of the best features from XPS and moved them into Informix. For anyone that has known Informix for a while this is very similar to what we planned for our "arrowhead" project, only instead of moving Informix features to XPS we are going  from XPS to Informix.  For Panther we have move optimizations for star and snowflake queries to Informix.  We've also moved the multi-index scan over which allows our optimizer to take advantage of more than one index during query processing.  We've also completed our support for light scans and appends so that all datatypes are now supported.

We've also added on other warehouse features not in XPS.  These include the ability to define an interval to fragment your table by (such as monthly) and then Informix will automatically add new fragments as neccessary.

 

Ease of Use/Embeddability

We have leveraged the dbscheduler to add more tasks that can be scheduled to automatically run.  This means that Informix relies even less on DBA's to manage the day to day operation.  These new tasks do thing like monitor table usage to see if any can benefit from compression and look for user that have been idle too long.  We have also automated the tasks of adding space to tables.  You can now set a policy that tells the system when to automatically add space to a table or extend a chunk and where to get the space from.  We also now ship Timeseries support with Informix.  In addition there is no longer an extra step to register the Timeseries into your database, this will be done automatically on first use.  The same is true of BTS and Spatial, these datatypes can be used as needed without prior registration.  Another area we have worked on is self configuring.  Now it is possible to configure Informix so that it will automatically take advantage of whatever CPUs or memory is available on the machine it is being installed on.

 

Application Development

We wanted to make it easier for people to develop applications and decided the easiest way to do that would be to feel their pain.  We created a team to focus on open source products and to see how hard it was to get them to work with Informix.  From this effort we got two things - we got a bunch of open source products that now work very well with Informix such as Hibernate, Drupal, xWiki, and MediaWiki, and we also made a number of simple but powerful changes to Informix that made it much easier to port these applications.  These changes were mostly in the area of new SQL syntax, for instance the ability to create an object if it does not already exist, or drop an object only if it does exist, allow NULL columns, and expressions in aggregates.  The other thing we did was to add an SPL debugger.  This debugger will work with IBM Optim Studio and soon will work with Visual Studio.  Finally, we are now also bundling Mashup hub with Informix.

 

Security

In the area of security we have greatly improved our auditing facilities with selective row level (SRL) auditing.  With SRL you can greatly reduce the impact auditing has on performance as well as be much more selective as to what gets audited.  The result is better performance and less audit records to look through.  We have also added trusted context to Informix.  This feature is useful in a multi-tiered environment, for instance where you have an App server talking to Informix.  With 11.50 you would have to create a connection from the App server to Informix and multiplex all your users through that connection.  The result would be that all users would look the same to Informix.  With trusted context you are able to let the App server do the authentication and then send a message to Informix to say what user a particular query is being sent on behave of.  This way you can distinguish user 1 from user 2 from ... without shutting down and restarting the connection to Informix.

 

Performance 

We always strive to improve performance in general and 11.7 is no exception.  We have increased OLTP performance overall by 10% and warehouse queries have been improved by as much as 10 times! in the case of queries that benefit from star or snowflake join optimizations.  In other areas we have also introduced a new indexing method called Forest of Trees (FOT).  When you use a btree you can run into a problem with contention at the root node.  Even when all your applications are reading the btree each will have to get a read lock on the root node and this can lead to serialization in a heavily used system.  FOT reduces this problem by creating a hash table and then creating a btree under each has bucket.  This means that contention is reduced on the root node since there are now many root nodes, and it also reduces the depth of the btree which can lead to better performance as well.

 

These are just a few of the new features that have been introduced in 11.7.  There will be plenty of opportunities to learn more in the next few weeks.  We have already started doing our analyst briefings, a number of new articles will be getting published, lots of information and sessions will focus on 11.7 at the IOD conference in 2 weeks.  Finally, if you are interested in hearing more and don't know where to turn you can always send me a note!