The AC - DC Railroad

A few weekends ago, service on Metro-North and Amtrak was thrown into chaos when two trains ripped down portions of the overhead caternary (power line). Trains were cancelled, weekend riders stranded.

Metro-North’s service in Connecticut is made all the more challenging by a technological quirk of fate. Ours is the only commuter railroad in the U.S. that operates on three modes of power… AC, DC and diesel.

On a typical run from, say, New Haven to Grand Central, the first part of the journey is done “under the wire”, the trains being powered by 13,000 volt AC overhead wires, or catenaries. Around Pelham, in Westchester County, the pantographs are lowered and the conversion is made to 660 volt DC third-rail power for the rest of the trip into New York. Even diesel engines must convert to third-rail, as their smoky exhaust is banned in the Park Avenue tunnels.

And there’s the rub: Connecticut trains need both AC and DC, overhead and third-rail, power pick-ups and processors. That means a lot more electronics, and added cost, for each car. While the DC-only new M7 cars running in Westchester cost about $2 million each, the dual-mode M8 car designed for Connecticut will cost considerably more.

So, some folks are asking… “Why not just use one power source? Just replace the overhead wires with third-rail and we can buy cheaper cars.” Simple, yes. Smart, no. And here’s why.

 There’s not enough space to lay a third-rail along each of the four sets of tracks in the existing right of way. All four existing tracks would have to be ripped out and the space between them widened. Every bridge and tunnel would have to be widened, platforms moved and land acquired. Cost? Probably hundreds of millions of dollars, years of construction and service disruptions.

 Even with third-rail, the CDOT would still be required to provide overhead power lines for Amtrak. That would mean maintaining two power systems at double the cost. We’re currently spending billions just to upgrade the 80-year old catenary, so why then replace it with third-rail?

 Third-rail AC power requires power substations every few miles, meaning further construction and real estate. The environmental lawsuits alone would kill this idea.

 DC-powered third rail is less efficient. Trains accelerate much faster using overhead AC voltage, the power source used by the fastest trains in the world… the TGV, Shinkansen, etc. On third-rail speeds are limited to 75 miles an hour vs. 90 mph under the wire. That means, mile for mile, commute time is longer using third rail.

 Third-rail ices up in bad weather and can get buried in snow, causing short circuits. Overhead wires have problems sometimes, but they are never buried in a blizzard.

 Third-rail is dangerous to pedestrians and track workers.
The idea of conversion to third-rail was studied in the 1980’s by consultants to CDOT. They concluded that, while cumbersome and costly, the current dual-power system is, in the long run, cheaper and more efficient than installing third-rail. This time, the engineers at CDOT got it right.

Not satisfied, some of the third-rail fans tried pushing bills through the Legislature in 2005 to study the replacement scheme yet again. More studies would have meant years of delay in ordering already overdue car replacements. Fortunately, the Legislature dispensed with these nuisance proposals quickly.

Doubtless, we’ll have further “wires down” problems in the years to come. Ironically, Metro-North’s 97% on-time record has made us come to expect stellar service, despite our ancient infrastructure. But in the long run, service will be faster and even more reliable by sticking with our dual-mode system.

Mortal man can do it

Install 11.1.2 that is.

Introduction

This is thankfully going to be a relatively short one, both for you and for me, as I’m busy working on other (hopefully) cool things to blog about.  Dodeca is coming back as are a series of Planning posts.

 

This is not a post on how to install EPM 11.1.2 as John Goodwin (and others) took care of that.  Nor am I going to review all of the good things 11.1.2 brings to Essbase geeks, as it’s been more than adequately covered here, and here, and here, and goodness knows how many other places.  Besides, you all went to ODTUG Kaleidoscope 2010 and know all of this already.

 

What follows is just my general impression of the 11.1.2 install from a non-infrastructure consultant’s perspective and a few tips and tricks I picked up along the way.

The last one just about killed me

I had sort of a love/hate relationship with my just-on-my-laptop VM 11.1.1.3 installation.  I don’t know if I was having the mental block of the ages, was plain stupid, or just unlucky, but that install was hellacious, especially getting EPMA to work.  I probably reinstalled that product eight or nine times.  If VM Ware didn’t have a snapshot feature for me to restore to a clean pre-install view of the VM, I’d likely still be installing 11.1.1.3.

Heavy Lifting

Well, the 11.1.2 install is still big.  The install footprint of 32-bit Foundation, Essbase, Planning, ERPi, Financial Data Quality Management, Financial Reports, and Hyperion Profitability and Cost Management is 8.37 gigabytes.  All I can say is downloading the install files on my meager DSL line took a looooong time.

 

But we like big in America, right?  (Some day I will bore you all with my housewares-are-built-for-aircraft-carries-but-alas-I-live-in-a-submarine analogy about old urban houses, but not today.  Lucky you.)  So big is in, big is bold, and as far as Oracle EPM is concerned, big is here to stay.

 

Memory requirements are big as well.  11.1.2 uses significantly more than 11.1.1.3.  Where the four gigabytes available to Windows 2003 Standard Edition 32 bit was more than adequate for 11.1.1.3, it isn’t really for 11.1.2.  It’s time to move on up to Enterprise Edition, or in my case, a 64 bit version of Windows 2008 Standard Edition.  I didn’t use the 64 bit OS for this first install, but for all of you consultants (no one puts everything onto a single box in the real world, right?) out there with 32 bit machines – ask not for whom the bell tolls, it tolls for your x86 box.

My “full stack” on 11.1.1.3

11.1.1.3 free memory

My “full stack” on 11.1.2

11.1.2 free memory

Quite a difference there.  Oh well, 64 bit is the future anyway.

Didn’t hurt a bit

Keeping memory considerations in mind, here’s the really important thing – after the tedium of expanding and combining the 11.1.2 installation files into a single directory, the installation was painless with the exception of two issues.

One tiny snag

After installation was complete, I ran the EPM Configurator.  It all worked (slowly, but hey, this is a VM) without a hitch, with the exception of one thing – ERPi.  My VM is Windows Server 2008 SP2 and SQL Server 2008.  The configurator threw an error when it got to ERPi as that only works with Oracle’s own database.  (What, they make something other than the world’s most awesome EPM stack?  Apparently, there’s this product called Oracle 11g.  Who knew?)  That is a change from 11.1.1.3 – ERPi worked just fine with SQL Server 2005.  So, for the time being at least, ERPi isn’t on this box.  Maybe later.

One other snag, not so tiny

If this one is in the documentation, I couldn’t find it.  It’s valid at least for SQL Server installations and if you want EPMA to run, it will behoove you to read on. 

 

I used one great big single database for everything – remember what I wrote about real world considerations not exactly entering into my install?  This is just for testing/education purposes so please don’t use my example as the way to properly install Oracle EPM 11.1.2.

 

Having stated that caveat, when I tried to run EPMA, I got a lulu of an Event Viewer error message, telling me that my SQL Server 2008 database needed to be in ALLOW_SNAPSHOT_ISOLATION and READ_COMMITTED_SNAPSHOT mode.

 

I have read around and am slightly alarmed by the comments here and here about what happens when these settings are turned on.  However, my actual usage will be pretty light and if things get too slow, I’ll reevaluate the settings.  This is probably not a great approach in your real world settings.  I will also caution that I am fairly SQL-stupid, so this may not be a big deal.

 

Using the all-in-one database HYP_EPM, here’s the SQL query I had to run to apply the settings:

ALTER DATABASE HYP_EPM

SET ALLOW_SNAPSHOT_ISOLATION ON

 

ALTER DATABASE HYP_EPM

SET READ_COMMITTED_SNAPSHOT ON

 

VM Ware came to the rescue again, as I had oh-so-cleverly created a VM Snapshot of the bare OS.  Having reverted to that, recreating the HYP_EPM database, and then running the above query, the install (and EPMA) worked, so I’m keeping my fingers crossed on this setting not having any really bad impact. 

 

You can read people who actually have a clue about installations discussing this very subject in this thread on OTN, so it wasn’t just happenstance that I decided to go down this route.

Errata

The word “Hyperion” is almost gone.  It’s not in the directory path.  It’s not in Windows’ Start Menu.  It is however, amusingly the name of the configurator utility. 

Eh, I always preferred Arbor Software anyway, and yes, ARBORPATH lives.

Look Ma, no cavities!

As this is a testing environment only (my VM, that is), I had no fears about installing all of the components into a single SQL Server database.  And that’s it – it installed and configured on the first second go round.  Who’d a thunk it?  Maybe you all had unalloyed success with 11.1.1.3, but that wasn’t my experience. 

Service startup

11.1.1.3 had a service startup script, sort of, that never really worked; 11.1.2 has one and it works.  No more trying different start orders, or reading the official order and knowing that it doesn’t really work for you, or casting the chicken bones to figure out how to get the services running.  Someone at Oracle heard the huddled masses yearning to breathe free and cut The Gordian Knot of service startup with this service start script.  (I hope there’s some sort of award for mixing metaphors, ‘cause I think I have at least a shot at honorable mention with that last sentence.) 

 

Why do I get these warm and fuzzy feelings about Oracle?  They just get it in a way Hyperion never did.  What’s really weird is that so many of the players are the same – what on earth was holding them back?  No matter, Oracle has set them free and we all benefit from it.

Conclusion

• Oracle EPM 11.1.2 is resource intensive.  Now would be a good time for us my-world-is-on-my-laptop tinkerers to go 64 bit.  I am downloading the 64 bit binaries as I type.
• At least on SQL Server installations, the EPMA relational data store needs ALLOW_SNAPSHOT_ISOLATION and READ_COMMITTED_SNAPSHOT set to ON.
• A plain Jane install on a single Windows 2008 Server box can be accomplished by mere mortals, like Yr. Obdt. Srvnt.  I have no idea what it’s like to install in a distributed environment. 

 

Once installed, you’re off to the cool new stuff in Planning and especially those ASO allocations through Calc Manager and MaxL.

 

Happy Essbase hacking until next time.

Paying for Transportation: Let's Not Be Fuelish

My father taught me that “there’s no such thing as a free lunch”. I’m trying to teach my daughter that there’s no such thing as a free ride.

With a newly minted driver’s license, she’s found new freedom behind the wheel… until I ask her to fill the gas tank. Now, as we drive past local gas stations, she’s taken new interest in comparison pricing.

A large part of the cost of gas is taxes that subsidize our transportation system. That’s why gas is so much cheaper in New Jersey (no taxes)… but also why they have expensive tolls on their highways. Remember… “no free ride”.

But as I have written about for five years now, gasoline prices in this country are still too low because they don’t ask us to pay for the true costs associated with driving.

How much would gasoline cost if the pump price included the billions we spend each day fighting to keep access to Middle East oil? Or, what if drivers were asked to repay the 40% of all police costs associated with patrolling the roads? Or how about asking motorists to pay at the pump for the land lost to taxation by turning it into highways?

But it gets worse.

In Connecticut we are facing a major funding crisis in transportation because of our over-reliance on the gas tax.

Not even the relatively expensive fares on Metro-North reflect the true cost of running that service, hence the operating subsidy by the state from the Special Transportation Fund (STF). Created in 1984, the STF was to be a “lock box” of money to invest in our state’s transportation infrastructure and help cover the cost of transit operations.

But 40% of the STF comes from gasoline taxes. So in 1997 when Governor John Rowland got the legislature to cut the gas tax from 39 cents to 25 cents a gallon, he doomed the fare-paying bus and rail passengers in our state.

Every penny of gas tax raises $15 million in annual revenue. So Rowland’s tax-cutting stunt cost the STF $210 million a year, or almost $3 billion to date. That’s money that wasn’t available to be used to repair our roads, buy new rail cars or buses or keep fares affordable.

And going forward, with automobiles becoming more fuel efficient (i.e. using less gas) and eventually becoming all-electric (using no gas), the STF will have even less funding to maintain the highways and keep mass transit affordable.

And that’s not even considering the legislature’s recent diversion of $10 million from the STF into the general fund to balance the budget. So much for “lock boxes.”

What’s the alternative?

Well, tolls would be a good start. But the state’s Transportation Strategy Board spent a million dollars to study tolling, only to reject eleven options cited by the consultants. And there are precious few lawmakers in Hartford or Washington willing to suggest tolling our “free-ways”, especially in an election year.

Or how about using a VMT (Vehicle Miles Traveled) tax? Motorists would pay for the number of miles they drive: a highway use fee. Of course, the use of GPS to track their travel patterns and charge by time-of-day and congestion probably leaves the libertarians shaking… as if Big Brother doesn’t know your every move already by tracking your cell phone.

Or how about what they’ve done in Portland Oregon, a payroll tax of .07% with proceeds going directly to the transit system, by-passing sticky-fingered legislators.

Someone, somehow is going to pay for our roads and rails. If not in a gas tax, then in some other revenue raising-mechanism. Because, as we must all understand, there is no free ride.

What I learned in ODTUG Kaleidoscope Summer Camp

Oh, you missed a good one

I think I’ve finally recovered from 2010’s Kaleidoscope conference.  That is one fun conference. 

Why weren’t you there?  Oh, you were?  How does it feel to be, oh, 30% smarter because of 4.5 days of conference?  Pretty awesome for what is a small amount of money and time.  You know where you should direct your effort and career for at least the next year?  Yup, it’s hard to believe the content is all done by volunteers.

Let’s review it by days.

Saturday

Volunteer time at the Ronald H. Brown Middle School.  I find it revealing that ODTUG sponsors a volunteer activity.  Think about who your fellow ODTUGers (ODTUGians?  ODTUGanese?  ODTUGs?  You choose, and please let me know.) are – they’re people who freely share their knowledge, expertise, and experience so that others may benefit.  This philosophy of giving carries beyond the boundaries of Oracle. 

I thought I would be “clever” and do the inside library work instead of painting outside.  Note to all – when it’s stiflingly hot, fresh air might be better than stale air.   Next time…

Sunday

Did you go to the Symposium on Sunday?  The one with all of the cool stuff that I’m not allowed to blog about?  Yeah, me too.  For the rest of you – what, are you crazy?  Show up next year.  You too will then have The Knowledge.

I will write three things that (hopefully) won’t get me in trouble.

1) Essbase is going to be at the heart of a *lot* of Oracle products.  It’s kind of scary to see how valuable good old Essbase is and will be.  Think that hoary old film, Red Dawn, but instead of knocking off foreign invaders and yelling, “Wolverines”, yell “Essbase!”  The name of this company ain’t Hyperion Solutions, that’s for sure.

2) In the not too distant future, I will actually want to drop the Classic add-in.

3) Essbase Fan, Toufic is looking for you.  He’s not so crazy about ScaryView.

And one thing that I won’t get into trouble for writing – Oracle EPM 11.1.2 is here, and it’s got lots of new features and functions.  Check it out – it’s coming to an installation near you.  The Planning features and ASO procedural calculations alone ought to occupy my spare time for the next six months or so.

Monday

Sessions

There was sort of a theme to Monday – can you spot it?

ASO – Glenn Schwartzberg’s Practical Coding Examples.  Gary Crisci, president of the ODTUG Hyperion SIG, was unable to attend, so I drove whilst Glenn talked.  Gary would have been better but Glenn ran over anyway.  Hmm, maybe Gary would have cracked the whip and kept Glenn on schedule.  :)  Regardless, it was an honor to click for Glenn.

ASO – Gary Crisci’s Essbase ASO:  The Best Thing Since Sliced Data – I believe this was given by an Oracle employee, Dave Collins.  If I have that wrong, someone please drop me a line.  It was an excellent presentation.

ASO – Edward Roske gave his always popular ASO Tuning and Optimization presentation.  I didn’t attend this, but instead attended a Leveraging Essbase Data Types class given by Mike Nader.

ASO – Buland Chowdhury of Orcale gave a presentation on implementing efficient ASO databases.  I have no idea where I was – my notes (and brain) are blank.  I think I had ASO spilling out of my ears because my brain was full so I went back to my room to obsessively rehearse my presentation.  Fun times, fun times.

Midnight (but not actually at midnight) Monday Madness

You know, for someone who has never actually watched an entire episode of Jeopardy, I think I did okay on the answers (I submitted 40+, saw five that I recognized) and being a judge was fun.  Someone (Glenn?) made the comment I was a tough judge.  Happily for the contestants, I didn’t wear a funny looking wig and robe and get to say things like “You shall be taken to the place from whence you came, and from there be taken…” 

Tuesday

A slightly more diverse day than Monday (hey, you have your interests, I have mine).  I checked out the ever-entertaining Tim Tow with his Dodeca presentation (Why doesn’t everyone just use this product?  It’s so obviously the best of the best.  Yes, there are many questions in life I never get satisfactory answers to.) 

My long time friend and colleague Dave Farnsworth gave a presentation on hacking Planning tables.  There have been unofficial and Hyperion-only guides (I saw one, once) that explain the schemas, but little direct documentation of what the tables do, and how to pull information you can actually use out of them.  Go check out his whitepaper if you didn’t download it during the conference.  For those of us who use Planning, this alone was worth the price of conference admission.

I attended two of the training courses – both Studio related.  I’ve had training on Studio before, so I was hoping these would be a quick refresher.  Unfortunately, there was so much material to cover that the training went at double the legal speed limit, and if there was a problem you (that would be me, actually) were lost.  I sat and watched so it wasn’t completely lost on me.

Wednesday

I didn’t attend the ODI 11g presentation so I could be the ambassador to Matthew Holt’s Dodeca presentation.  His was a really interesting session because although it focused on Dodeca the lessons learnt that he related were germane to any technology center of excellence.

Then ASO again, with Glenn Schwartzberg, this time Studio tips and tricks.  He was excellent as always.

I went to a really interesting MDX presentation, but it was sort of counter intuitive.  Dan Pressman introduced a methodology for analyzing the ASO cache so that MDX is not used, or at least is used as little as possible.  I’m sure one day I’ll be smart enough to understand it.  I hope.  :)

HAL is dead, DIM is dying, ODI is rudely healthy

Finally, my ODI presentation with Markus Shipley.  I have to say that I work myself up into way too much of a lather over this stuff.  This presentation was recorded, and I have the recording on my laptop; I am listening to it as I type.  I have figured out who I sound like.  William Holden?  I wish.  Cary Grant?  It seems incredibly unlikely.  Jimmy Stewart?  Getting closer.  Ira Glass of NPR’s This American Life?  I sound and look like him.  Ouch.

ODTUG will soon post this and other presentations for you all to hear.  Enjoy the moments when I say things like “Right here” and “As you can see”.  Were you in the room, such comments might even make sense as you would be able to see my ever dancing hands.  The lack of this visual aid will add an air of mystery to the presentation when webcast.  I think it’s good to have intrigue in one’s daily humdrum life.

I then attended my co-presenter, Markus Shipley’s, Calc Manager session.  When I’ve used Calc Manager in the past I’ve just copied and pasted in Essbase calc script code.  Next time I’ll actually give the graphical approach a try.

Finally, I saw my ex-colleague Tracy McMullen do her always outstanding best with Charles Chew re figuring out if EAS, EIS, or Studio is the right choice.  It was an interesting exercise in product feature comparison.

Just a side note about two of Wednesday’s sessions – both the Dodeca and the EAS, EIS, or Studio presentations were ostensibly about their specific technologies.  Not entirely true.  Both presentations were really a lot more valuable than that because they addressed wider ideas such as what it takes to successfully set up a technology center of excellence and how does a company go through the technology selection process.  I’m not sure how we (that would be ODTUG) should highlight presentations that have this kind of scope but this is the sort of stuff that management consulting companies charge big bucks for.  We (that would be the Kaleidoscope attendees) got it for free along with the tool specific content.  That’s extraordinary value for money and another reason why Kaleidoscope is so awesome.

Thursday

Awww, I was getting sad, the conference was rolling to an end. 

Matt Milella kicked off with 11.1.2’s SmartView Extensions.  Oracle put a lot of work into SmartView to make it less painful in Excel.  Just keep going guys till you reach the nirvana Toufic demoed on Sunday.  I’m waiting with fingers crossed.

I then didn’t leave the room (so convenient) and sat in on Robert Gideon’s SmartView VBA automation session.  Gads, I love sessions with code samples.  I am sad, but there it is.

And the last was like the first, Glenn again, with his Load Rules Advanced To Basic session.  I was his ambassador (it’s just a chance to mock the person you’re introducing, and you get these cool pins for your conference badge) and led the room in a round of applause of appreciation.  Glenn has given so much to ODTUG, even though he’s not on any board.  I am really looking forward to working with him this year on the Hyperion SIG board.

The End, until next year

And that was it.  All that hard work (okay, all I did was help review presentations, but I have a little sweat equity) by ODTUG, and Your Conference Connection really paid off.  I have been and remain incredibly impressed by the time, energy, and enthusiasm everyone brought to the conference.  It showed in the high caliber of the presentations, the roadmaps from Oracle, the training sessions, and the buzz amongst the attendees.  If you can’t tell, I loved it.

Remember, except for YCC who run the business side of the conference, everyone who participates in Kaleidoscope does so voluntarily.  The ODTUG board members and the Kaleidoscope Conference Co-Chairs (hello, Edward, 2011 Conference Supremo) spent countless hours making sure that this conference delivers value.  They have my heartfelt thanks for doing such a great job.  I can’t wait till next year.  You will be there in Long Beach, won’t you?