Initial Opinions on OBIEE 11g?

[Mark Rittman]

Hi All,

Well, OBIEE 11g is now generally available, the NDAs are now revoked,
and probably most people have had a chance to try the installation,
use some of the new features, maybe try a migration from 10g. So, I
thought I'd start a thread to see what people's opinion is so far,
specifically

- How did the installation go for you? Within our company we've had a
few issues with newer versions of Windows Server, for example, and
with the "Simple" installation method. Also the installer tends to
have problems with systems with less than 3GB of RAM. Anyone else hit
any issues?

- Has anyone tried the Essbase integration? Does it solve the issues
we hit in 10g?

- What about new features such as Hierarchical columns? Do you think
this is a good solution, what with the distinction between these and
regular attribute columns?

- How are you getting on with WebLogic, and Enterprise Manager?

- Any other initial observations?

We may end up breaking this into separate threads, but I thought it'd
be interesting getting feedback from the "expert practitioners" on
this board.

Mark

[Michael Wilcke]

Mark,

my first experiences with 11g so far:

- win32 installation => just too little memory (max. 3.x GB) to have
fun working with
- win64 installation => works fine with 4-5 GB

Good news: I got a nice justification for a new 8 GB MacBook Pro
Bad news: you need a second VM with a (complete! and then disabled)
win32 install to get admintool and catalogmanager working

- I got win32 admintool nearly working in win64 install (just "view
data" doesn't work, because of ODBC 32-64? issues). But I'm not sure,
whether this is a solid solution. Any comments?

- simple install doesn't work on win64 (admin credentials not
accepted), but enterprise install works fine

- some research was required to figure out, how to start up an
installation after reboot (NodeManager, AdminServer, ManagedServer,
opmnctl,... UFF)

- couldn't get the system (win32, win64) start up automatically with
server boot (without logging in and having cmd-windows on my task-bar)

- couldn't get the Session Monitor to show physical SQL (have to
manually navigate to end of nqquery.log)

- changing Server-IP (this happens with my laptop-DHCP setup) may
require reboot/restart

- experienced some strange security issues after wake-up of VM (could
not log-in as weblogic any more, required restart)

WebLogic: eats all that RAM - cluster-ready-install is overkill (I
don't have a single "cluster customer" - and also want a lean demo
system)

EM: NICE - heading in the right direction. Needs to be completed over
time (most configs still to be done in config-files, that are well
hidden in a
large directory tree).

Essbase/new features: couldn't get hands on this so far, because all
these install-issues kept me busy.

Michael

[ravi kulkarni]

Hi,
We are in situation where our DW is running on 10g Rel on Linux and might take a while to have parallel environment to test migrate to 11g.
1. But was able to install OBIEE successfully on linux (64 bit). Irrespective of warning, I used 10g rel 1 database for repository and was able to create repository successfully. (I am not sure the problems that I faced are because of this)
2. Then installing admintool on windows was kind of painful. I tried on windows 2003 on VM. Documentation suggested, we should choose simple install and then shut down all services. But for some reason, after simple installation, configuration starts failing around step, where it tries to start OPMN, couldn't find any resolution. In fact couple of guys posted similar questions on Mark's blog, asking how they should proceed. Then I aborted installation and found that all necessary binaries were installed (like admin tool, catalog manager etc) I dont see it windows menus, but after navigating to ORACLE_HOME/bin folder, I do see admintool executable.
3. Then was able to configure system DSN to connect to BI server installed on Linux and was able to connect to it.
4. Using DOS upgrade utility, I was able to upgrade 10g repository to 11g and see all of it.
But now stuck there, because I can see repository but for some reason, repository is not able to comminucate with DW or I even don't know, how do I configure BI server to talk to DW server. Remember using system DSN on windows we connected from client running on windows to talk to BI server running on Linux.

There is lot of documentation available, but I couldn't find one consolidated document, which describes step by step tasks in order to migrate from 10g to OBIEE 11g. In my case my OBIEE 10g is running on windows and I tried to migrate to 11g running on Linux. May be I have to take one step back and just continue to use it windows to make it work and then may be try to move BI server to Linux.


Again, for all of this, I must of have spent, not more than 15-20 hours.

I will try it one more, may be over long weekend. I have attended so may presentations, about features and enhancments incorporated 11g, now it's available, but I am still not able to see my own dashboards, kind of frustrating, but I guess, it will take a while to understand product well and then to migrate.

Anybody else was able to successfuly and completly migrate to 11g?

--
You received this message because you are subscribed to the Google
Groups "OBIEE Enterprise Methodology Group" group.
To post to this group, send email to
obiee-enterpri...@googlegroups.com
To unsubscribe from this group, send email to
obiee-enterprise-met...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/obiee-enterprise-methodology?hl=en

All content to the OBIEE EMG lies under the Creative Commons Attribution 3.0 Unported License (http://creativecommons.org/licenses/by/3.0/).  Any content sourced must be attributed back to the OBIEE EMG with a link to the Google Group (http://groups.google.com/group/obiee-enterprise-methodology).

[Michael Wilcke]

I had similar problems connecting to the db.

As a workaround copy your tnsnames.ora to <..>/Oracle_BI1/network/
admin

Worked for me ...

/Michael

> On Mon, Aug 30, 2010 at 11:19 PM, Mark Rittman <mark.ritt...@rittmanmead.com

> > obiee-enterprise-met...@googlegroups.com<obiee-enterprise-­methodology%2Bunsu...@googlegroups.com>

> > For more options, visit this group at
> >http://groups.google.com/group/obiee-enterprise-methodology?hl=en
>
> > All content to the OBIEE EMG lies under the Creative Commons Attribution
> > 3.0 Unported License (http://creativecommons.org/licenses/by/3.0/).  Any
> > content sourced must be attributed back to the OBIEE EMG with a link to the
> > Google Group (http://groups.google.com/group/obiee-enterprise-methodology

> > ).- Zitierten Text ausblenden -
>
> - Zitierten Text anzeigen -

[nge...@gmail.com]

Mark,

This is a very good post that I had actually planned on throwing out
there, but I am glad you beat me to it.

I've had mixed results but mainly success working with the GA of OBIEE
11g R1. It is no where as buggy as the Beta releases though.
It almost appears that the development team really favored the *Nix
environments for this release over Windows which is okay by me.
Ultimately, we've had success on 32-bit Windows and 64-bit RHEL.

The Essbase integration is really sweet but I found that you must have
the 11.1.2 client installed, in addition to the new configuration
settings, in order for connectivity to any version of Essbase (i.e.:
9x, 11x) to work. The 11.1.1.3 client doesn't work. I am quite
certain this is a fusion middlware thing and I have my questions out
to my Oracle guys on that one.

WebLogic and EM are awesome and we've had a lot of experience with
Weblogic since it was BEA's so that is no problem.

Observations I would make for those getting ramped up with OBI11g are
that you really need to understand Weblogic and Fusion Middleware.
What is Node Manager? What is OPMN? etc. I would recommend anyone
conducting an install of OBI 11g take a step back for OBI itself for a
second and spend a few days really just learning the Fusion Middleware
stack and how that really works. Then, jump back into the OBI piece.
You'll be happy that you did.

I am actually part of a free webinar on 11g today (8/31/10) if any
wants to sign up you can do so here (http://biconsultinggroup.com/
aboutus.asp?CategoryID=373). We'll be doing a quick run down of what
we like about the release.

Has anyone attempted to create new maps and/or layers for OBI
MapViewer?

Has anyone had any problems with the upgrade migration from 10g to
11g?

Cheers,
Christian
http://www.artofbi.com

[ravi kulkarni]

Thanks Michael,
I have all necessary entries in ORACLE_BI1/network/tnsnames.ora  I can tnsping to DW database from there.

When I checked tns log on database, I do log entries both for BI server running on Linux and admin tool running on windows, with success status. But when I use admin tool, open repository and right click on table and select "update row count" I get "connection has failed" message

I guess I am missing something here.

[chet justice]

I can't say I've had any success installing the software.  So far I've tried on XP Pro (VM on laptop) and Server 2008 (AWS instance).  I've chosen the Enterprise install each time.  

The laptop I can understand, sort of, as it seems to be severely underpowered.  3 GB RAM total for the machine.  I understand the integration with WebLogic/Fusion, but it makes it that much more difficult for those starting out to get it up and running on their personal machines.  I can still install the database on my laptop; in fact I have 3 or 4 versions (in VMs) on my machine right now.  That's nice.

A couple of days ago, I fired up a cloud Windows Server 2008 server with 18 GB RAM.  Installed the database.  Created the BI metadata (rcu).  Finally began the OBIEE install (Enterprise).  It failed finally on one of the configuration steps (12 of 13), I believe it was one of the opmnctl processes.

That said, I haven't read the install documents from front to back yet.  I have, however, installed a lot of Oracle software over the years, including the EM, OAS (10g), etc.  I can usually get the software up and running.  This is a different beast though.

I'll eventually read the docs, sooner rather than later.

On the bright side, installing the latest 11g forces people to learn a bit more about the interactions of the database, EM, WebLogic, etc.  That's a good thing.

My 2 cents.

chet

chet justice

http://oraclenerd.com
http://www.linkedin.com/in/chetjustice

[Alexander Hansal]

Hi Mark et al.

Due to the lack of transportable computing power I opted for a VM on a
SSD with 2 GB RAM reserved for the VM.

All through the beta phase, this was a nightmare and I have seen the
post-install configuration and services fail quite often.

Now I have installed the GA release on my Windows XP laptop directly
(3.7 GB RAM) with all the components (DB, WebLogic, BI) installed on
the SSD Drive.

The install ran flawlessly and it works like a charm.

I have yet to find/develop a good way to stop/start the "beast" with a
single click and without command boxes floating on my desktop.

The blog posts of Venkat, Mark, John and Erik Eckhard have been most
helpful.

Started reading the docs some weeks ago but not yet finished...

No upgrade from 10g done yet, this is next on my list.

have a nice day

@lex

[Mark Rittman]

Thanks everyone for the feedback so far. It's very interesting.

For me, the bit that's surprising me is how many problems people are
having with the installation. We were on the beta since about May, and
pretty much all of the beta releases (except one that had a known
issue) installed fine for me, with none of the issues that have been
reported. Others within RM have mentioned a couple of the issues, but
not on the scale that have been reported here and elsewhere on the
web.

I thought it'd be worth outlining how I set up OBIEE, at least so I
can give a reference point.

1. I installed using both the simple install, and the enterprise
install. I've also performed the enterprise/scale out BI system
install with no problems.
2. The OS I installed on was Windows Server 2003 SP2, 32-bit.
3. I assigned 4GB of RAM to the VM, which sometimes was taken down to
3GB after the install. I'm not sure if the full 4GB was even used
really, as it's 32-bit Windows Server.
4. I installed it using VMWare Fusion 3, on the Mac, with the Mac host
having 8GB of RAM.
5. For the Windows Server environment, the only pre-install
configuration I performed was to install the Microsoft loopback
adapter, assign a static IP address to it, install the VMWare tools,
and install Oracle Database 11gR2 with the data warehouse install
template.
6. The install of OBIEE would always go flawlessly, without requiring
restarts of OPMN etc.
7. After the install, I'd need to add the entries under C:\Middleware
\Oracle_BI1\network\ before the BI Server could see my database

I've also carried out an install using OEL5 and the 64-bit version of
OBIEE 11g; again, there have been no problems, with a 4GB VM and an
8GB host system.

I suspect most problems people are having are falling into four
categories:

- trying to install into an environment without enough RAM - this is
causing processes to fail to start up, affecting the post-install
config steps
- using marginally-supported OS versions - Vista, XP etc - I haven't
looked carefully at the supported systems doc, but I suspect these are
not supported. I also always try to use Windows Server 2003 as (a)
it's a server OS and (b) it doesn't have all the user access control
guff that came with Vista/Windows Server 2008
- problems around networking - IPV6 as someone else mentioned; not
setting up the loopback adapter, other other misc. networking and/or
host naming issues
- Trying to do things like installing just the admin tool on its own,
trying to run with certain service shut down etc - basically, the
install process in 11.1.1.3 is pretty basic, can't really be deviated
from, I've not tried these but i'm not surprised they don't work

What I will do now is try and collate together, from this thread, the
comments on this blog, and other blog posts by the likes of Chet etc
to see if we can start to categorize these standard install issues,
and come up with a set of troubleshooting steps, and common
resolutions, for them. I'll post the results of this on here, and on
our blog, once I pull them together.

thanks

Mark

On Sep 1, 4:00 pm, Alexander Hansal <sheepfarmer9...@googlemail.com>
wrote:

[Michael Wilcke]

Mark,

thanks for your summary.

One point left unmentioned though, and this one is important for me,
since I'm about to install BI11g at a customers site...

Did you manage to get a REAL server installation, i.e. an installation
that starts up completely after e.g. booting the server (not requiring
login by some admin)?

My plan was to try "Java Service" (that we used to start OC4J on
servers with 10g)

/Michael

[Mark Rittman]

Hi All,

Following-up from my email, I've set up a public Google Docs
spreadsheet where we can collate the install experiences we've had,
and try and work out if there's a pattern to what works, and what
doesn't work, and if there are categories of problems that people are
hitting that we can try and find answers for collectively.

I've produced an input form for the spreadsheet, which is here:

https://spreadsheets.google.com/a/rittmanmead.com/viewform?formkey=dDJhVTFfMmc3ekpYczVQWW45ZGNJNkE6MQ

It would be great if anyone who's done an install can complete this,
as it'll allow us to relate the successes or not with the host OS,
amount of RAM, etc etc.

The spreadsheet itself can be accessed from here:

https://spreadsheets.google.com/ccc?key=0Alsb7uc3oet7dDJhVTFfMmc3ekpYczVQWW45ZGNJNkE&hl=en

The spreadsheet is open to the public, and editable, so be careful
(but also record anything that's of use).

Once we get sufficient responses in, we can try and draw some
conclusions

thanks

Mark

[Chris Marais]

Hi All,

Let me break this into 3 sections, installation, upgrade and general
impressions.

Installation experiences:

I was an eager beaver so when it became available I downloaded
straight away. I am the type of person that jumps in straight away so
I ran my initial install on my laptop which has an existing 10g
install.

First thing I noticed is how much longer the install takes and the
fact that I needed a database to complete the install. My laptop is
hopelessly under resourced for the job (XP 2G ram). To get past the
database requirement I installed XE (was a complaint about not being
supported but it seems to work fine).

I kicked off my install around 8pm and after getting through all of
the screens and letting it go into auto pilot I went to bed (about 3
hours later). When I woke up in the morning I noticed that the last
step had failed I hit retry and it worked. I was quite fortunate that
first up I had a working environment on my laptop. It is a little
fragile when you first kick off a publisher report but after that it
becomes fairly responsive.

I have since re-installed a few times on my laptop with various
failures at different stages in the process as well as a Windows 2003
server (now running 10g and 11g side by side - took a little bit of
ferreting around to get it working as I was not familiar with weblogic
or enterprise manager - that has changed now)

My upgrade experiences:

Other than the actual upgrade taking quite a while it seems to have
worked for the most part. Our back end database is sql server 2005 and
the year ago measures seem to be having a few issues - have not put
too much troubleshooting time into it so I am sure it is solvable.

From the web catalogue perspective some of the more complex
reports(read insanely complex) have had issues as part of the upgrade
and do not actually run post upgrade.

General impressions:

Analysis services cubes are much faster to query, 10G was creating
some mad MDX for some queries causing anaysis services to pretty much
return all measures at all dimension levels. I might re-enable access
to AS cubes once we upgrade after some more thorough testing (the
hierarchical column certainly makes it more intuitive to browse).

BI publisher might actually get some playtime now that it is better (I
was going to say fully but let's not get ahead of ourselves)
integrated into OBIEE.

The tables being interactive on the dashboard is a nifty feature and
certainly will save on some clicks into answers. Overall the UI is
much improved and I think it will be positively accepted by my user
community.

Using weblogic and EM has been a bit of a learning curve for me and I
wasn't too sure how easy it would be to use but I must say that I am
starting to grow fond of the web console (certainly saves me remoting
to servers)

That's it for now. Apologies for the poor formatting as I am typing
this up on my blackberry in a Gmail client.

Regards

Chris Marais

> --
> You received this message because you are subscribed to the Google
> Groups "OBIEE Enterprise Methodology Group" group.
> To post to this group, send email to
> obiee-enterpri...@googlegroups.com
> To unsubscribe from this group, send email to
> obiee-enterprise-met...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/obiee-enterprise-methodology?hl=en
>
> All content to the OBIEE EMG lies under the Creative Commons Attribution 3.0
> Unported License (http://creativecommons.org/licenses/by/3.0/). Any content
> sourced must be attributed back to the OBIEE EMG with a link to the Google
> Group (http://groups.google.com/group/obiee-enterprise-methodology).
>

--
Sent from my mobile device

[Stewart Bryson]

@ravi kulkarni

Try clicking the "Require fully qualified table names" in the connection pool properties dialogue box and see if that solves the issue.

Regards,

Stewart Bryson

[Kevin McGinley]

I won't repeat everything that has been said here, as I have
experienced both problems and the simplicity that Mark has described.
I've probably done over 50 installs (mostly on Windows) across a
variety of laptops, servers, local VM's, VM's running in VSphere, and
AWS and here are my quick thoughts:

- Networking can throw the install for a huge loop and cause specific
tasks to fail, namely the cluster controller. You can always skip
past the failed components, disable the cluster, then bring everything
up post-install.
- Loopback helps, but so does editing the hosts file, as the install
hiccups when it detects an IP address only and not a host name.
- I've done a successful simple install on a Windows 2003 32-bit
machine with 1.7gb RAM (AWS), but the database has to be elsewhere.
It's slow to perform anything in EM/WL, but SampleApp Lite is fine.
- Had the same problems as others with TNS after an upgrade. I will
try Michael's solution.
- Scripting the Admin and Managed servers to start after boot is on
the to-do list. I hope an OC4J-as-a-service solution presents itself.
- I have also had problems getting the Admin tool to run on 64-bit.
- I hope the client-only install option is addressed soon. We
discussed this during the beta, but I've yet to hear anything else.

General comments about 11g overall:

- I'll be curious to see what existing customers think of the new UI
for Dashboards/Answers. I can see love/hate attitudes developing.
New customers should be fine.
- EM is a leap forward, but it's not without a learning curve. New
admins will be fine, but existing admins will be pulling their hair
out for a while.
- One area that isn't discussed as much online is security. I will be
very interested to see what it's like to design a brand new security
model from scratch in a platform-only environment.
- I am anxious to explore the charting engine and see what, if
anything, can be tweaked. Now that we're on an Oracle charting
engine, I hope the ability to build new chart types follows.
- The overall integration of Oracle FMW technology is great. It's
easy to see, though, how OBI itself will become similar plumbing in
other technology someday.

Kevin

On Sep 1, 11:11 am, Stewart Bryson <stewartbry...@gmail.com> wrote:
> @ravi kulkarni
>
> Try clicking the "Require fully qualified table names" in the connection
> pool properties dialogue box and see if that solves the issue.
>
> Regards,
>
> Stewart Bryson
>
> On Wed, Sep 1, 2010 at 9:19 AM, Mark Rittman

> <mark.ritt...@rittmanmead.com>wrote:

>
>
>
> > Hi All,
>
> > Following-up from my email, I've set up a public Google Docs
> > spreadsheet where we can collate the install experiences we've had,
> > and try and work out if there's a pattern to what works, and what
> > doesn't work, and if there are categories of problems that people are
> > hitting that we can try and find answers for collectively.
>
> > I've produced an input form for the spreadsheet, which is here:
>

> >https://spreadsheets.google.com/a/rittmanmead.com/viewform?formkey=dD...

>
> > It would be great if anyone who's done an install can complete this,
> > as it'll allow us to relate the successes or not with the host OS,
> > amount of RAM, etc etc.
>
> > The spreadsheet itself can be accessed from here:
>

> >https://spreadsheets.google.com/ccc?key=0Alsb7uc3oet7dDJhVTFfMmc3ekpY...

> > obiee-enterprise-met...@googlegroups.com<obiee-enterprise- methodology%2Bunsu...@googlegroups.com>

[ravi kulkarni]

Mark,
About #5
You did it because of DHCP?
I have opened TAR, because during configuration it's failing during Step Creating ASInstance. Oracle rep asked 2 questions.
1. is it on physical box or VM
2. DHCP

To rule out VM issue, I tried installing on physical box, but still configuration fails at same point.

In our case, server is already configured with static IP, do I still need to create loopback adapter and assign static IP to it? I will try it out anyways.

--
You received this message because you are subscribed to the Google
Groups "OBIEE Enterprise Methodology Group" group.
To post to this group, send email to
obiee-enterpri...@googlegroups.com
To unsubscribe from this group, send email to
obiee-enterprise-met...@googlegroups.com

[Adrian Ward]

Let me start by the positives.

Having seen the initial demos over a year ago I was looking forward to the product being released.
Hierarchical columns, better Essbase, integrated BI Publisher, move to a meta data repository = all good news
When we worked on the beta we had a 3 day workshop with Oracle (there were about 16 customers and consultants there) in Reading and it was very impressive. We installed the product from start to finish - RCU, install and up grade. There were a couple of work arounds they gave us to install but for me the process worked really well. The new features were impressive and daunting at the same time.
Speaking to clients they love the ability to patch the rpd with 'xml', in theory that gives them back source control.

Now we have the GA available.

I have not been able to install a clean version without a failure along the way.
Tried hard on several types of VM - Windows XP, 2003 Server, 2008, Win 7, Win 7 64 bit, Win 2008 64 bit.
Where an OS was not properly supported the install would notify and stop. IE 6 was unsupported.
I was using an Oracle Virtualbox VM, with fully patched OS's (this took days!)
After dozens of attempts I have given up on the VM route - I did get it working once I had fiddled with various settings on Win 7 and 32 bit others but it ran like a dog.
(I have 4 GB ram on the laptop and had the DB installed on the VM's too)

Both Oracle 10g and 11g database seemed to work fine.
The RCU works well.
I have managed to install on a proper PC (windows) and get up and running. Windows XP Professional, 11g database, IE 8. Although the odbc driver appears to have issues (not sure if that is fixed by Oracle yet?)

All my install experiences had a common theme - the opnmctl failed. If you ignore this and run the em and control panel to start the services you can get it to work - but sometimes not (I have not had time to figure out why it sometimes fails)

If you forget the password on an rpd (or do not know it in the first place) then there does not seem to be any way to recover it. (Anyone tried to crack the rpd?)

Overall

As a MAC user would say about their toy, "I love this thing, it just works!" - this cannot be said of 11g OBIEE. True for some, but mainly not. You can get it working - I have a client machine running, but you have to fiddle or ignore messages.

I would like to see a developer install (i.e. client tools only) as soon as possible. The current 10g version is simple to package and a developer can be up and running in less than an hour. None of the sites I have worked at for the last few years had anything like the type of PC required by developers now. Often you have centrally controlled software distribution, and no downloading or usb sticks, so the easier to package the better (MOD sites will really struggle here). All the offshore developers will also struggle, so will need probably have to remote to a machine in the UK to do their work.

I would also like to see a non FMW integrated version (maybe that is possible today?). For sandboxes, laptop demos etc I really do not want all the extra hassle of 'managing the service'

We manage large platform implementations which have up to 20 instances per box. All instances can use their own method for security, with their own settings - this is going to take some serious thinking to figure out how to implement in 11g.

I still love the new front end and integrated BI Publisher, and look forward to hierarchical columns. It will take a while to get used to the change in the way you write a request, as the old method was really simple - the new method is very similar to the Siebel Marketing Analytics module.

I think the new customers will love the product, old customers will really need a good excuse to go through the upgrade.

But there again it takes a while to teach an old dog new tricks :)

Adrian Ward

PS How do you turn off those annoying chart 'growing' actions?

(btw - these views are my own and not that of any company or client that I am associated with)

[Jeff McQuigg]

I don't have the time to play around with it, and probably won't for a
while, so I'd like to hear some more about core functionality. So far
all people have talked about it seems is installation and some non-
core items such as Essbase and BIP.

Can people comment on the core functionality? Is the UI any better?
Is dashboard prompting actually usable by humans? Any good modeling
features? Does it solve any problems/bugs/limitations of 10.x? Does
it generate better/worse SQL?

Thanks for replies.

Jeff M.

On Aug 30, 11:19 pm, Mark Rittman <mark.ritt...@rittmanmead.com>
wrote:

[Mark Rittman]

Good questions Jeff. That's what I originally intended with the
thread, but seeing as people were having so many problems with the
install, it sort of got diverted.

Here's some thoughts on the repository modeling (RPD) changes in
11gR1.

1. There are some interesting extensions/improvements in 11g.
Specifically, the double-column (ID/description) feature that allows
you to filter on an ID, when showing a description in a dashboard
prompt. This sort of falls into the "why wasn't it there at the start"
category, but it's still welcome. Venkat's written this up here:

http://www.rittmanmead.com/2010/08/18/oracle-bi-ee-11g-handling-double-columns-iddescription-interoperability/

2. There's also the concept of lookups, both physical lookups and
logical ones. Physical ones allow us to use a LOOKUP function to reach
out to lookup tables in the physical layer, and add additional columns
to tables in the logical layer. Not sure why you'd use this rather
than just a standard join in the physical layer, though it might be a
way of avoiding unwanted outer joins in an LTS definition. Where it
gets more interesting though is applying this same function in the
logical layer, as the BI Server will do a separate SQL query and then
join the results at the BI Server layer, which may be useful from a
performance / separation of queries perspective. Venkat has written
this feature up here:

http://www.rittmanmead.com/2010/08/17/oracle-bi-ee-11g-lookup-tables-sparse-and-dense-lookups/
http://www.rittmanmead.com/2010/08/20/oracle-bi-ee-11g-reporting-on-clobs-lookups/

3. There's also the ability now to define ragged and skip-level level-
based hierarchies, and parent-child hierarchies. I guess we all know
the potential/advantages of this, but one thing I've noticed in
playing with it is that there are some non-obvious modeling steps
you'll need to take in order for it all to work.

4. The other major thing is hierarchical columns, where we drag and
drop hierarchies from the BMM layer into the physical layer, and then
users can perform "in-column drilling" within Answers. Where this gets
interesting is in the SQL that's being developed - in generally, with
11g, we see a lot more subquery factoring going on (the WITH clause)
even for simple tabular reports. With hierarchical columns, the SQL
generated can be immense, with each indentation of the query requiring
its own SQL statement, which are then unioned together to get the
results set. On a single user, test system on your laptop, it all runs
fine, but on a real system with 1000's of users, this could get
"interesting". Same goes for hierarchical column queries against
Essbase/MDX sources - for each indentation, a separate MDX query is
fired, even though in reality, a tool such as WebAnalysis would fire
just a single MDX query. I think, when we look back at 11gR1 when
we're at 11gR2, we might well possibly say that Oracle ended up doing
a lot of optimization work to get their hierarchical column queries to
generate efficient SQL.

Anyone else got to the point of trying out these new RPD features?

Mark

[Kevin McGinley]

Agreed on the hierarchical columns SQL. Another interesting point
there is that cache is supposed to work across all of those
subqueries, but like all cache hits, it can be tricky.

To comment on your front end UI questions, Jeff, I think the UI
'improvements' are going to be open to vast interpretation. The
overall UI change is the biggest in a long time, so it's hard to be
objective about it. My very first reaction was !()@$*@# but since
then, I've grown accustomed to it. That's not to say I think it's
better (or worse), but there is an adjustment period. The prompt UI
is very different as well, and this area is a mixed bag. On the one
hand it's nice to have new form controls available (radio, check,
etc), but on the other hand, the carry-overs from 10g (drop-downs and
multi-selects) feel more cumbersome. The UI for building a prompt has
changed, too. Again, there are elements of that work well, and others
that feel clunky.

I really like the home page (vs. launching directly into a dashboard)
and while some users might complain about the extra clicks to get to
their default dashboard, the login process is cleaner and you don't
have to build an empty dashboard page if you want to start firing
queries right away. I also really like some small touches on the
dashboard: re-pivoting tables/pivot tables, auto-scrolling legends,
zoom options for charts - all of these things fix little challenges
one had with the UI in 10g, like drilling down on a chart that could
explode to a large number of series, rendering the chart illegible.

The one area that I haven't had much time to play in, but I had a very
positive initial reaction to, was the new web layout editor for BIP.
It's very easy to jump in and start using it right away. Time will
tell how well it handles more complex BIP reports.

There are so many areas of the tool to talk about, we almost need to
start discussion topics on a more granular level. That said, I'm not
sure if this forum should turn into an expose on 11g! :) We might
consider having conversations around what's going to be the best way
to implement all of this newness in a real world environment (versus
the sandboxes that most of us have been playing in). An upgrade
discussion topic might be a good place to start for those of us doing
a real world upgrade.

Kevin

On Sep 2, 11:38 pm, Mark Rittman <mark.ritt...@rittmanmead.com> wrote:
> Good questions Jeff. That's what I originally intended with the
> thread, but seeing as people were having so many problems with the
> install, it sort of got diverted.
>
> Here's some thoughts on the repository modeling (RPD) changes in
> 11gR1.
>
> 1. There are some interesting extensions/improvements in 11g.
> Specifically, the double-column (ID/description) feature that allows
> you to filter on an ID, when showing a description in a dashboard
> prompt. This sort of falls into the "why wasn't it there at the start"
> category, but it's still welcome. Venkat's written this up here:
>

> http://www.rittmanmead.com/2010/08/18/oracle-bi-ee-11g-handling-doubl...

>
> 2. There's also the concept of lookups, both physical lookups and
> logical ones. Physical ones allow us to use a LOOKUP function to reach
> out to lookup tables in the physical layer, and add additional columns
> to tables in the logical layer. Not sure why you'd use this rather
> than just a standard join in the physical layer, though it might be a
> way of avoiding unwanted outer joins in an LTS definition. Where it
> gets more interesting though is applying this same function in the
> logical layer, as the BI Server will do a separate SQL query and then
> join the results at the BI Server layer, which may be useful from a
> performance / separation of queries perspective. Venkat has written
> this feature up here:
>

> http://www.rittmanmead.com/2010/08/17/oracle-bi-ee-11g-lookup-tables-...http://www.rittmanmead.com/2010/08/20/oracle-bi-ee-11g-reporting-on-c...

[Mark Rittman]

Hi All,

We've had a number of responses now to the survey (17 in total), and
whilst there's no obvious culprit, here are some observations

- most people actually seem to have got it installed OK, with no
problems
- those that haven't, seem to have hit problems with a couple of
things
---- installing on Linux and therefore not getting the admin tool
---- installing on Windows 64-bit and having to install again on 32-
bit to get the Admin tool, Catalog Manager etc
---- if using DHCP, either not installing the loopback adapter, or if
they have, not setting a static IP for it
---- trying to install with under 3GB of RAM

Also the Simple install seems more problematic than the Enterprise
install, which correlates with our findings as well.

So in summary, it seems that the most guaranteed way to do the install
is

1. Either use a static IP, or if using DHCP and Windows use the
loopback adapter and assign it a static IP address
2. Have 4GB or more of RAM available
3. Be prepared, if using a 64-bit version of Windows, to have to
install the admin tools on a 32-bit environment
4. Similarly, if using Linux, be prepared to install the admin tools
on a 32-bit windows environment
5. If there is an issue, it's usually because (a) you're using an
exotic OS or (b) you've not got enough RAM or (c) your networking is
not set up as specified in the docs - static IP, and loopback adapter
if you're using DHCP.

There may be other issues hit by people, but it's not clear from the
responses if there's a pattern, which would suggest it's specific to
their environment and their setup. Memory and networking seem to be
the main areas in which things can go wrong though (and problems with
the Simple Install)

Mark

On Sep 3, 5:38 am, Mark Rittman <mark.ritt...@rittmanmead.com> wrote:
> Good questions Jeff. That's what I originally intended with the
> thread, but seeing as people were having so many problems with the
> install, it sort of got diverted.
>
> Here's some thoughts on the repository modeling (RPD) changes in
> 11gR1.
>
> 1. There are some interesting extensions/improvements in 11g.
> Specifically, the double-column (ID/description) feature that allows
> you to filter on an ID, when showing a description in a dashboard
> prompt. This sort of falls into the "why wasn't it there at the start"
> category, but it's still welcome. Venkat's written this up here:
>

> http://www.rittmanmead.com/2010/08/18/oracle-bi-ee-11g-handling-doubl...

>
> 2. There's also the concept of lookups, both physical lookups and
> logical ones. Physical ones allow us to use a LOOKUP function to reach
> out to lookup tables in the physical layer, and add additional columns
> to tables in the logical layer. Not sure why you'd use this rather
> than just a standard join in the physical layer, though it might be a
> way of avoiding unwanted outer joins in an LTS definition. Where it
> gets more interesting though is applying this same function in the
> logical layer, as the BI Server will do a separate SQL query and then
> join the results at the BI Server layer, which may be useful from a
> performance / separation of queries perspective. Venkat has written
> this feature up here:
>

> http://www.rittmanmead.com/2010/08/17/oracle-bi-ee-11g-lookup-tables-...http://www.rittmanmead.com/2010/08/20/oracle-bi-ee-11g-reporting-on-c...