I guess I should explain the versioning (11.1) a bit more before I continue. In case you didn't get it before, Hyperion 9.5 (AKA "Code Name Kennedy") is officially dead. All of the products that were going to be released under the 9.5 name are now going to be part of version 11.1. The collective name of almost all of the former Hyperion products is "Oracle Hyperion EPM System 11" the first version of which will be 11.1. All of the apps will be released under this name just with EPM being replaced by the Hyperion product name. For instance, Hyperion Planning will now be "Oracle Hyperion Planning System 11" or possibly "Oracle Hyperion Planning 188.8.131.52.0" (more on this later).
Essbase is officially dropping the Hyperion name (Arbor Software would be so proud). It's new name will be Oracle Essbase 184.108.40.206.0. Before anyone gets excited that calling it 11.1.1 makes it somehow get patched on the day of its release, it doesn't. 220.127.116.11.0 will be the most major release of Essbase since 7.1.2 (when they added Aggregate Storage) but it will be a completely unpatched version of Essbase. My advice is to wait until the last digit starts counting up from 0 before anyone tries to implement it. I know people who were in the "Oracle Essbase" and "Oracle Hyperion Planning" betas (check me out: I'm using the new names), and every one of them said that they got the products to crash multiple times during the betas. Definitely delay implementing until a real patch comes out.
One of the things that Al mentioned is that when "Oracle Hyperion EPM System 18.104.22.168.0" is released tentatively in June of 2008, it will be English-only and astoundingly, Windows-only. This had been mentioned at Collaborate among other places, but it still bewilders me since a large portion of the client base is on Unix. I heard a rumor from one of the System 11 beta testers that further, the initial release will not work on any web application server but Oracle Application Server. Considering Oracle now owns BEA, it would bowl me over if the initial release didn't run on BEA WebLogic, but in case this rumor is true, I had to pass it on.
The next release (22.214.171.124.0) is tentatively scheduled for Q3 of 2008 (late September, in other words). It will support other operating systems, other application servers, and include a few minor enhancements Oracle couldn't get into 126.96.36.199. The localized version with languages beyond English should be out by the end of 2008. Code named "Dickens," it will be numbered 188.8.131.52.0. The version beyond that is code named "Talleyrand" and should come out sometime in 2009.
Al discussed the new installation procedure in System 11. Oracle has been getting an earful over how difficult System 9 is to install, so System 11 will have a single installation and configuration program. This program will begin by performing a platform preinstall check to confirm that everything is correct before the installation begins. If your operating system is the incorrect version, for instance, it should notify you that you need to download a service pack or upgrade your entire server. The new install will be organized by tier and should install across multiple servers at once (I'm not sure how that will work). The new installation program will also install every Hyperion product you own in the correct order. This should take the guesswork out as to the order products need to be installed. (Can I hear a hallelujah?)
The "What's Coming in Oracle Essbase 11" portion of Al's talk went into excellent detail. I'll hit the high points here. If you want a thorough discussion (with a few screen shots) of Essbase 11, my fellow Oracle ACE Director in the Hyperion space, Tim Tow has 2 detailed entries: the first is on the Collaborate Essbase New Features presentation and the other entry (which gets a bit API-geeky towards the end) is specifically about the new date and text measures in Essbase cubes. Here are the new Essbase things worth mentioning:
- Varying attributes. Right now, when you change the attribute (on an attribute dimension) for a base member, Essbase treats that member as if it always had that attribute. Let's say you had a product dimension and attribute dimension (boolean) called "Discounted". Product XYZ is not currently discounted, so it has a boolean on the Discounted dimension of false. Next month, it's being discounted, so it gets a boolean on the Discounted dimension of true. What this does is to restate all of history (and all forward forecasted months too) as if that member was always discounted. The new "Varying Attribute" functionality will allow that member's Discounted status to change over time without restating of history. I think this will be implemented the way Planning already does it: through Smart Lists. It will just be done at the Essbase server level so Smart Lists will be expanded before the results are ever returned to the client.
- Non-numeric measures. This will allow cubes to finally contain text and date values as measures. (Can I hear another hallelujah?) With this change, Essbase can finally replace many of the data marts that have required relational databases to this point. It's my understanding that these measures will be stored as numerics behind the scenes and replaced with text/dates at query time (again, like Smart Lists). Tim has an entire blog entry on this if you want more detail.
- Operational logging, backup, and recovery. Al didn't explain much about this one, but I've heard that all changes to the database will now be stored in a separate audit log. This will include things like outline changes, data loads, calcs, sends from Excel, sends from Planning, and so on. Further, if a database crashes, an administrator will be able to replay the activities in the log from the time of the backup to get the database back up to the point where it crashed. Say that since the last database backup, two data loads and two business rules have run. The database crashes and has to be restored from the backup. Once the backup is restored, the two loads and rules can be replayed in the order they were executed to get the database back up to where it was. If it works this way, this will be very cool. It certainly has to be the almost useless transaction log that Essbase maintains at present.
- Time Intelligence in BSO. In 9.3, ASO added support for built-in, time aware dimensions. Essbase 11 brings these to BSO (Block Storage Option) databases.
- ASO Enhancements. The three that I know about are ASO as the target of a transparent partition, the ability to clear a portion of an ASO database, and user submission of data to level-0 of an ASO cube. It's obvious to me that ASO is the future of Essbase (at least so far as Oracle is concerned) so look for ASO to quickly be brought up to the capabilities of BSO.
- Performance enhancements. The ones mentioned included faster restructuring of cubes, optimization of queries, and "allowing an on-line feed into a persisted cube while allowing read/write to continue." I'm not sure what that last one means, but feel free to discuss it in the comments to this entry if you have any idea. It might be an ASO thing.
- OBIEE Integration. You'll be able to use OBIEE as a data source for an Essbase cube. I'm curious to see just how automated this will be. For instance, will you be able to just press a button and an Oracle OLAP cube gets persisted into an Essbase BSO cube? I guess we'll wait until late June to find out.
- Essbase Studio. The long-term vision for the Essbase Studio is to replace both EIS (Essbase Integration Services) and EAS (Essbase Administration Services). Let me tell you that it's not there yet. The first release of this will need to be used in conjunction with EAS, but it may be able to replace most of the EAS functionality. It will have the ability to do snowflake and star schema builds into Essbase cubes like EIS. It will have the data load and dimension build editors from EAS. It also adds some nice features like the ability to have a dimension library and push one dimension to multiple cubes. Further, it will be able to tell you that if you make a change to X dimension, that change will impact the following N cubes.
- Drill Anywhere. Utilizing the Essbase Studio, the ability to drill to detailed information will be expanded. The sources into which one can drill now include relational, OBIEE, URL, FDM, Essbase (linked partition), and intriguingly, a Java method (basically, it runs Java code). In old Essbase, drilling into transactional data was only allowed on data, but drilling in Essbase 11 will also be supported on metadata. I can't think of how I'll be using that yet. Maybe if you double-click on a customer name, it will take you to a report showing you the contact information for that customer? I'm anxious to try it, at least.