Want to know The Truth About CPM?

18 March 2013

Where in the world is Cameron, day 2, Australasian edition

Day 2, 19 March 2013, 8:55 am NZDST

It’s a good thing I am, seemingly, beyond embarrassment, as I feel asleep last night during dinner.

Actually, it’s a bit worse than that, as I actually fell asleep after the quite excellent NZOUG 2013 dinner, during the music quiz, in a room full of screaming people. With a live DJ. OTOH, I have never in my life awoken to the strains of “Wake up, Little Susie” sung just for me, so there is that. And before you go, “Ah, sloshed again. Get that geek dried out, asap,” I knew Cameron + tiredness + alcohol was a bad combination and very purposely held back from the truly excellent Mac’s Bitter. All to no effect – I might have well had more of that beer than I did for all the good it did me. I will plead the excuse that when I finally walked back to the hotel I was able to figure out that in fact, Cameron-time, it was about 4 am and I had just spent the entire night awake. My buddy Bambi Price, keeps on telling me, “Don’t think about what time it really is”, and my conscious mind doesn’t, but I fear that my body does.

Yr. Obdnt. Srvnt., Slay the evil of bad data in Essbase with ODI, 9:30 am to 10:20 am, NZDST

I really enjoy doing this presentation as I am a huge fan of ODI and of good data in Essbase databases. It is a surprising and mildly shocking fact that many, many, many Essbase/Planning implementations do not handle data quality at all, or rely on 100% manual data validation to tie out numbers.

I am writing this update a half hour before the presentation, so I have no idea what turnout will be. Also, it is quite possible that after yesterday’s Dodeca presentation the word has spread and I will be facing the New Zealand equivalent of a ghost town.

Update – nope, I had a decent number of people show up a wee bit late – was everyone toasting the sleepy American or just out partying? They aren’t telling me and I’m not asking. :)

Charles Naslund, Infrastructure Preparations for Hyperion EPM, 10:40 am to 11:30 am, NZDST

Many things in life elude me: why do we vote for politicians that lie to us and they know that we know that they are lying and yet we vote for them again and again; why, really, do all the things that taste so good end up being so bad for us; why is Oracle EPM infrastructure so hard?

Well, lying politicians (as far as I can tell this is true for all parties and all countries) and fattening-yet-delicious food will be with us (and me) forever, but could there be hope when it comes to the complex concept known as EPM infrastructure? Regular readers of this blog know that EPM infrastructure is a continuing challenge for me so I have great hopes for Charles’ session. Maybe my limited knowledge can be expanded. Maybe.

Charles (yea!, fellow Septic) is going through the architecture topology in nice simple to understand terms. Keep it simple Charles, please.

Now we’re onto the topic of virtualization – yup, it’s the same story as in the States – real boxes for Essbase, virtualize everything else, and it would be a really good idea if you went with Oracle VM.

SLAs (Service Level Agreements) – ah, these tend to be somewhat more honored in the breach than in the than the observance. This one particularly frosts my cookies when the Essbase or Planning server (or servers) go KABOOM and no one but no one in IT seems to own the servers.

  • Packet size – pre compression. Now I know why Smart View is faster than IE when it comes to forms. Here are the average network bandwidth requirements on a per form basis:
  • HFM 64 to 128 k
  • Planning 32 to 64 kb
  • Smart View 28k This is pretty darn amazing – the SV team has done some magic here.

SANs – Every client I’ve had in the last five years has wanted to run Essbase (and everything else) off of a SAN. Essbase needs fast disk to perform well, memory and CPUs be damned. How to do this – sort of be virtualized, sort of not, by using dedicated LUNs, CPUs, and memory. The data (PAG/IND or tablespace) is on the dedicated LUN – everything else can be on shared SAN resources. This strategy gives you a 20 to 25 percent performance boost.
All in all, a nice session. Maybe if I attend enough of these some infrastructure wisdom will rub off on me. Maybe.

Richard Philipson, EPM Case Study: Rank Group, 2:00 pm to 2:50 pm

Richard as always does an excellent job. Quite how he does Essbase, Planning, HFM, BI, infrastructure, etc., etc., etc. is a bit beyond me. Did I mention that he’s a talented graphic artist. <insert envy> Or maybe I’m just stuck in rut.

This is an interesting app – it’s architected so that the private equity firm (Rank Group) can bring in/drop companies really quickly. Unlike most other corporate systems, and because their business is so dynamic, they have both full public internet access (definitely not the norm) and four environments: dev, qual, prod, and archive. The last environment is used to snapshot their business at a given time so they can have a baseline to compare against. Again, that is not the way HFM is typically set up, to put it mildly.

The Teaser

This blog will be updated throughout the day (although looking at my laptop clock I realize it’s 9:30 pm EDST so how much bated-breath refreshing of this blog there may be is open to question) so stay tuned.

Be seeing you.

1 comment:

Anonymous said...

Since i'm only -6 hours different with your timezone, i check the your blog each morning for update. Btw, ANZ is not part of asian, so you might wants to use ANZ rather than Australasian. :)