Normally I don’t write blarticles related to the news. However, I could not resist writing something about Oracle’s dog and pony show last Wednesday at San Francisco’s City Hall. Oracle put on an event called “Half Way to Fusion”. It was meant to be sort of a round-up collection of what they’ve accomplished in the last year and how much closer they were to completing their vision of Fusion. In many ways, this is exactly what they presented. However, the real story behind this event was more in what Oracle did not say than in what they did. I was very disappointed in the follow up reporting of this event, as I think most accounts completely missed the realities of what was being announced. So let me read between the lines a little bit and provide a technical explanation of what Oracle was really saying. I think you’ll find some pretty amazing statements in there…
I Pity the Fool
I am convinced the executives at Oracle have been watching A-Team reruns. For all intents and purposes, last Wednesday’s highly-scripted “Half-way to Fusion” event should have been lauded as a recently discovered episode. For those who have not seen the A-Team in quite some time, let me start with a refresher on the basic episode plotline and the cast of characters we all came to love (courtesy of the A-Team Shrine):
Here’s the basic plotline:
The A-Team stumbles across someone in distress. They take them under their wing only to find themselves in a pickle for doing so. To get out of this pickle they end up building some vehicular monstrosity out of spare parts, broken appliances, and slightly chewed bubble gum. Relying only on their ingenuity and chutzpah, then end up prevailing against all odds.
Does this by any chance sound strikingly familiar to Oracle’s game plan of late?
For both Oracle and the A-Team, each episode centers on a motley cast of characters, fusing together their different styles to their foe’s eventual demise:
Charles Phillips, President, a.k.a.
Colonel John "Hannibal" Smith - The leader of the A-Team was forever "on the jazz" and just loved to pick fights with the head honcho of the bad guys.
Charles Phillips is the consummate showman. His comments are never to far from a jab at SAP.
Thomas Kurian, SVP Oracle Server Technologies Development, a.k.a.
Captain "Howlin' Mad" Murdock - The even-more-mentally-imbalanced-than-the-others pilot seemed to be proud of his "condition".
While Thomas is far from “Howlin’ Mad” (his presentation was by far the best of the event), he clearly is the hot glue gun of the crew. He leads the infrastructure, tools, and technology team for Fusion and his team put on a pretty impressive demonstration of what they have been working on. The only arguably “Howlin Mad” comment made by Thomas and his team was, “JDeveloper is completely declarative, so it’s pretty easy for a PeopleSoft developer to switch to.” But I am getting ahead of myself here.
John Wookey, SVP of Applications Development, a.k.a.
Sergeant Bosco "B.A." Baracus - Mechanic and general hard bloke. Nicknamed B.A. on account of his bad attitude.
John Wookey may actually be the hardest (working) bloke in Silicon Valley right now. Business Week wrote a thoughtful puff piece on him that tells the story of an affable, egalitarian, well-liked leader who works pretty much round the clock to get things to happen. So maybe John’s not B.A., although he did seem pretty perturbed at his demo-giver on stage.
So with our plotline and general cast of characters, let’s get the episode started. As we open, fade into San Francisco’s City Hall, replete with red carpets, huge customer-praising banners, and thumping techno-music shaking the walls. The excitement is palpable as people chatter on the edges of their seats. I actually think this was a well constructed marketing ploy – pack the seats so tightly that the only way for everyone to squeeze in is to alternate edge of the seat/back of the seat. Brilliant marketing maneuvers set aside, as the music dies down and the crowd comes to a hush, Colonel “Hannibal” Phillips strolls up on stage.
“We’re half way there!”
As the night marched on, presenter after presenter, I have to say I could not believe my ears. As a quick Google search will tell you, Oracle officially closed the PeopleSoft transaction on January 10, 2005. Let’s do some quick math. This means that whatever Oracle is halfway to, given the elapsed time of one year since the PeopleSoft deal, by Jan 18th, 2007 they should be completely there (wherever that is). This reminds me of an old riddle:
Sphinx: “How long is a piece of string?”
Adventurer: “Twice the distance from the middle to one end.”
You see, it’s all pretty ambiguous. And strangely enough, every single timeframe slide that the Oracle team put up that night, still has 2008 as the delivery date. I’m guessing they are still working out the kinks in Fusion’s calendaring application.
After that, things only got more interesting. For the whole event, I was sitting next to my business partner (and CEO of Newmerix), Ed Roberto. Like children at a school assembly, we kept nudging each other saying, “Did they just say what I thought they said?” As I mentioned before, many times it was not so much what they did say, but the implications of what they didn’t that left our jaws on the floor.
When I got up on Thursday morning, I was fully expecting a plethora of news reports, all unleashing the dogs on Oracle for what they had really announced. I was shocked to read story after story, analyst after analyst essentially regurgitating the episodic sound bites and catch phrases that, I am sure, Oracle had carefully constructed.
So I sort of feel it is my duty to try and give the contrarian view to the press's cursory analysis. My context is that I am the CTO of Newmerix and Newmerix builds change management products for packaged applications (like PeopleSoft and EBS). Because of this experience, I have become technically intimate with PeopleSoft and Oracle EBS, and somewhat with SAP, J.D. Edwards and Siebel. It is only by looking deep into the technical tea leaves that the ramifications of what Oracle was announcing truly become clear. I hope you find my background and commentary a useful counterpoint to the general reporting of the event.
Let me start by paraphrasing the realities of what was announced at the event.
“We are totally abandoning all the applications we own (EBS, PeopleSoft, JD Edwards and Siebel) and rewriting the whole thing. Well sort of, because it’s all based on the next generation of EBS and we tried to integrate as many of the good ideas as possible from the other toolsets. As for the applications, we took the best requirements from the 16 Billion dollars of applications we just bought and decided to totally rewrite them all on top of our next generation application platform. Don’t worry though. The end result will look a lot like what you have now. And don’t panic, the technology market is used to this 7-10 year upgrade cycle. We’ve been doing this with our customers for years. Remember moving from client/server to the web?”
To expand on how I got to this somewhat pithy and dramatic summary, I think the best way to navigate through the evening is by highlighting specific quotes directly from the execs. I believe these quotes (and the implications of them) reveal the underlying truths of what Oracle was announcing. I tried my best to copy the quotes down as correctly as possible, but I was scribbling like Bukowski running out of gin. So while I may deviate slightly from the wording, I’ll do my best to keep the spirit of the comments intact.
Oracle – 1, Esperanto - 0
Charles Phillips: “Let me dispel the myth that we have 4 different languages to integrate – we’re not doing that.”
This was the first time Ed and I turned to each other and sort of said “Wah?” One of the things I have written about a few times is that each application Oracle acquired uses a different (and sometimes proprietary) programming language for implementing business processes and code customizations. PeopleSoft has PeopleCode, Oracle EBS uses PS/SQL and Java, JD Edwards uses BFL (Business Function Language – a C derivative), and Siebel uses Siebel VB. Part of the unknown of how Oracle would get to Fusion has always been the question of how they would integrate these disparate languages together. The reason this is so important is that all of their customers’ business idiosyncrasies exist in millions of lines of customization code they have written in these languages. I actually have a half-written blarticle proposing how Oracle could do this integration elegantly. According to the Colonel, they are taking the easy way out; everything in Fusion is going to be in Java. Good thing I only wrote half of it.
Well it is one thing for Oracle to say “how lucky are we that we don’t have to do this”. It’s another thing for the customers to say “What about the thousands of lines of customization code that will have to be ported to your new Java based Fusion toolset? And what about all my highly paid PeopleSoft developers? They spent years learning the quirks of PeopleCode. Should I just show them the door?” Well, I hope you have some spare Java programmers lying around or a good training budget because you’re going to have to retool your whole application team.
In an attempt to be fair to Oracle on this point, SAP has actually been heading in the same direction for quite a while with their new NetWeaver toolset. Web DynPro now offers an all Java based platform for extending SAP applications. However, there is one major difference. SAP has been considerate enough to allow ABAP (their legacy proprietary programming language) to be used along side with Java. It’s a best of both worlds strategy. It lets SAP customers incrementally move their code base and team skill set towards an inevitable Java-based future. And without totally abandoning the existing ABAP code base they paid so handsomely to develop. I am really surprised Oracle has not taken this approach.
As a side note, there is also some rumblings that SAP will support the .NET language framework along side Java. While it’s hard to find much mention of this (it’s usually a small icon buried deep in their slide decks), the likelihood of finding a .NET logo in an Oracle deck is about as high as finding a pork chop at a bar mitzvah.
I’m Guessing Oracle’s Not a Vietnam Vet
John Wookey: “We are leaving [Oracle EBS] Forms and PeopleTools behind.”
This was probably one of the two most relevant and revealing statements made during the event. And to be clear, this statement was not made as some sidebar parenthetical or mumbled under his breath. Wookey was as declarative as declarative can be. Think about this for a second. Oracle is going to EOL (end-of-life) the tools you currently use to build your applications and move you to something totally new. In other words, you will have to completely forget how you build applications now and learn a whole new toolset for building in Fusion. Well, unless you’re an EBS user that is.
See, the dirty little secret here is that Fusion is really just the next generation of EBS. Here are two more quotes that support this assertion:
Thomas Kurian: “We are using a completely new toolset based on JDeveloper, BPEL, and open standards.”
Thomas Kurian: “Our next generation of applications is based on the Oracle EBS data model.”
The truth is that Oracle has been working on the next generation of EBS well before they ever picked up PeopleSoft or Siebel. At least a year ago, Oracle put out the beta 11i release of JDeveloper. JDeveloper was always intended as a migration path for Oracle EBS and Forms. That release (pre-PeopleSoft), already had most of the features that Oracle touted last Wednesday as part of the Fusion technology stack. BPEL Manager, an all Java based framework, drag and drop design time interfaces, metadata structures for MVC modeling. It was all there. I know this because I downloaded it and played with it. I even went and took a BPEL Manager class at Oracle. Almost everything they showed on Wednesday was pretty much there in the original beta. While it’s a very robust toolset, I guarantee you Oracle started working on it long before they decided to get Conway’s panties in a twist.
So what’s the big deal? Well, while Oracle is still driving in the same direction they’ve been describing for a year, they just took an unexpected hard left in asking people to learn a new toolset. This is your captain speaking - I’ve just turned on the seat belt light.
Ask anyone who has been developing on PeopleSoft or J.D. Edwards for a while and they will tell you it takes a long time to get to know the applications well. How does security fit in? How does multi-language support work? How does custom code get tied into to application? It’s all very complicated and specific to the architecture or whatever application you are talking about. While there are common concepts in each application (grids, security groups, textboxes, search capabilities, reporting, etc..), each of them works a bit differently. Developing in PeopleSoft or J.D. Edwards is not for the faint of heart. We’ve been doing it at Newmerix for three years now and still can’t claim to know everything. As an exercise, scan Monster.com for “PeopleSoft” and you’ll see two things: First, a whole bunch of resumes from people who call themselves PeopleSoft Developers and second, a whole bunch of Job Openings for people who call themselves PeopleSoft Developers. Moving from PeopleSoft to J.D. Edwards is not like moving from Java to C# (which a number of Newmerix developers have done). It’s a complete application operating system that you have to learn top down, not just a few language concepts and some annoying new syntax.
Approaching Reality on a Case by Case Basis
John Wookey: “For customers who customize, the reality is, on a case by case basis, this may cause some challenges.”
You think so? In saying this, Oracle carefully forgot to mention that about 80% of their customer base uses a little to a lot of customizations. I have to believe Oracle is coming from the EBS perspective here where customers are generally less customized. With PeopleSoft, J.D. Edwards and Siebel though, customization is a way of life. And to make things worse, customizations don’t just occur in one place in the application, they are totally pervasive across all the bits and pieces of it.
Thomas Kurian talked about a tool they were building that would look at the customizations you have made to your data model and help migrate them over to the new Fusion data model. While that’s useful, that’s about 5% of the actual problem. It’s sort of like saying you’ll get your personal finances in order by recategorizing your Halloween costume purchase from “Entertainment” to “Clothing”.
While its not hard to look at a PeopleSoft EMPLOYEE table and a Fusion EMPLOYEE table and figure out what columns need to be added, what about all the other tables that are custom created by each customer. Remember that when adding a customization in the first place you have two choices: extend what PeopleSoft provided you or totally replace it with something of your own. What if you replaced a core table with your own over time? How could Oracle possibly know how to fit these back into their data model?
To make matters worse, data model changes drive customizations in every other aspect of the application. For example, SQRs, the main report generation tool used in PeopleSoft, reference these customizations directly. While Oracle said they will offer a tool to convert your SQRs to their new XML based reporting engine, it’s just plainly not that easy. You see, an SQR is not just a set of SQL queries with a dash of output formatting added in. SQR is really a query-based programming language. If you look at an SQR, you’ll see mixed into all those queries and the programming code, the custom data model extension you have created or extended. So while Oracle may be bullish about their ability to do this, converting one language to another (both traditionally and from the programming sense) is just not that easy.
Here’s a simple example to illustrate my point. Go to Babelfish, pick English to French and type in “bait and switch”. Now take the resulting French and put it into the translation box and pick French to English. In all seriousness, this was the first phrase I played with for this blarticle. I am still laughing at the irony of the results. But the point is obvious – language translation needs context and in the PeopleSoft world the customizations are the context. To extend the analogy into the traditional programming world, if this was so easy, Microsoft would have written a Java to C# converter a long time ago and called it a day.
In addition to reporting, all of the custom PeopleCode you have written as well as the UI controls on each page directly reference those customized tables. Oracle will never be able to parse apart PeopleCode (yet another language) and turn it into Java with a simple find and replace. So you’re going to have to totally rewrite you code base from PeopleSoft to work with whatever the final “merged” data model is in Fusion. For some customers this means they will have to go through hundreds of thousands of lines of code. And I forgot to mention, the event model which fires custom code snippets works totally differently in PeopleSoft, EBS, J.D. Edwards, and I would expect, Fusion. Just understanding the sequencing of events in PeopleSoft is hard enough, writing the code might even be the easy part.
How to Get Your Customizations onto the Bus for Half Price
John Wookey: “We recommend you retire customizations.”
This is sort of like saying “We recommend you get rid of all fonts, bullet points, and coloring in every Word, Excel and PPT document you own. Just use our standard templates for everything”. I could expand on this topic for a long time but the net-net is that this is an unrealistic strategy. There is a reason why customers customize – it’s because everyone’s business is a little bit different. Oracle, it’s the whole reason they bought your software in the first place. Unless everyone gets together and agrees on how to run their businesses similarly, this will never, ever happen.
The packaged application vendors have been going on and on about this for as long as I can remember. I really wish they would just stop it. They need to embrace the fact that customizations make their world go round. As I have mentioned before, Newmerix exists today because of this customization phenomenon in the customer base (and the challenges it introduces). Every time Oracle or SAP have tried to convince customers to retire their customizations (Shai Agassi makes the plea about every third blog posting), there is a backlash from the customers. In addition, it’s completely incongruous for Oracle to ask this after spending the previous hour touting and showing demos of their new “open model-based” Fusions toolset. You have to ask why a customer would want to use open models. So they can change things, silly! Even Shai admits customizations are a mess, but are a necessary evil in packaged applications (and beneficial part of their platform). For an extended exposition on the benefits and challenges of customizations, check out another blarticle I wrote called Plato’s Software Children.
Five’s a Charm
As if introducing a new toolset into the mix wasn’t enough, Oracle decided to lob another grenade into their customer’s fox hole concerning Fusion’s next generation applications.
John Wookey: “We looked at the all the applications we had [PeopleSoft, EBS, JDE, and Siebel] and took the best requirements from each. We’re rewriting applications based on these requirements.”
Basically they realized they had four of everything and decided to rebuild a fifth from scratch. At first blush, this seems to make a ton of sense. Many have claimed that Oracle bought Siebel because of their recent prowess in industry specific process and features. Why shouldn’t they take that proprietary IP and work it into Fusion? But hang on, take a deeper look. Think through the reality of what this means to a customer trying to move to Fusion. Most likely, a lot but not all of your current application’s features will be present. Well, what do you do if your business process depends heavily on something that Oracle decided didn’t make the cut? You have two choices – change your business process (and retrain your staff on how to use it) or, guess what, customize the Fusion application! We’re totally back to square one again.
Here’s a simple analogy that I put together to explain what Oracle is asking of their customer base. Imagine if Microsoft came to you and said the following:
“We know you like MS-Office and you use it a lot. It definitely has some great features. But we really think next generation is OpenOffice. That’s the way we’re headed. Hey, you should love it, it's standards based. And you can still extend it with your own functionality. And look, we have this totally cool OpenDoc format that you can share stuff in. So what are we going to do? We are going to take the best word processing features of MSWord, WordPerfect, Framemaker, Lotus Notes, PDF, and anything else you have lying around and sort of merge it all together. Well, some of the features won’t be there. Like sub-bullet points. They didn’t make it. And Track Changes, that will work a bit differently because we based it on Notes. But don’t worry; we’ll give you some tools to convert all your Word, PDF, and WordPerfect documents over. But, just so you know, you can’t bring over anything based on a template (like Word letterhead) or any custom formatting you have done, and those Tables of Contents in PDFs, they aren’t going to be supported anymore. But don’t worry, all Word Processors go through a 7 year evolution – remember VI?"
Now take this recommendation and consider how many Word, PDF, etc.. docs you have lying around. Then consider rolling this new policy out to about 45,000 people in your organization. And then consider what happens if OpenOffice’s word processor is there in 2008, but not the spreadsheet application. In the words of Bill the Cat, “Ack!” It’s a total mess. Want more proof? Go to work for the Commonwealth of Massachusetts. This exact conversation (although in reverse of the way I said it) is happening there right now.
Second Mover Advantage
As another side note, a different big packaged application customer tried this about 5 years ago. I’ll give you a hint, they are German and you should be receiving a call from them very soon if you own PeopleSoft. You see, SAP tried to move their customer base from R/3 to mySAP starting in the end of the 90s. To do this, they basically decided to rewrite all of their applications on their new web based mySAP platform. Not surprisingly, half a decade later, they still don’t have all their customers on the new platform. What’s the reason – it took SAP a long time to rewrite ALL the applications in any given industry suite (HRMS, FINS, SCM, Manufacturing). Go figure that many of their customers refused to switch until all the applications they used were ported. SAP still struggles with this issue as they are now trying to sell their own application operating system, NetWeaver. Keep in mind this is the situation they are in without totally abandoning their existing toolset (through WebDynPro is a big upgrade) or their customers’ proprietary code bases. To put the problem simply – it’s just a ton of stuff to get done no matter who you are and how many developers you have working on the problem.
Beware Executives Bearing Alliterations
One of the catch phrases of the event was, “This is not a rip and replace”. I think almost every executive said it at least once, so I can’t attribute it to any one person. Each presenter gave many examples of how they were going to stage their customers onto Fusion as an upgrade and not a total replacement. Here’s their game plan:
First you get onto the latest release of your toolset and then start integrating with Fusion Middleware. Second you start converting all your reports (e.g. SQRs) to their XML middleware format. Then you use their data model conversion tool to set up the new Fusion apps. And viola, you are done. Um, I think they might be missing a few steps. What about the 100,000 lines of custom code you have and the 15 applications you depend on that have not been ported to Fusion yet?
Sorry folks, I just don’t buy the upgrade pitch.
While I do buy that upgrading to your application’s latest toolset gets you certified and integrated with Fusion Middleware, by 2008 though you’ll be starting down the barrel of a 2 year conversion project to get the applications on board. This seems slightly ingenuous in a market that only 7-8 years ago won all of their customers by replacing their legacy systems with packaged applications around the Y2K scare. Does anyone seem to remember this and how much of a total disaster this was? Accenture and Deloitte must be licking their Fusion chops like hyenas spying a wounded gazelle.
I Hate Admitting I was Right
The waiting is over. Oracle finally ‘fessed up to what a lot of the nay-sayers have always said their game plan was. Essentially Oracle wanted to become a key player in the next generation of IT platforms but didn’t have enough customers on EBS alone to get there. So they went out and bought a bunch of them in the hopes they could do an on-side kick or make a two point conversion. I actually called this one about three months ago in my blarticle The Emerging Application Operating System. Here is the excerpt:
“The problem though is that they didn’t have a big enough installed customer base of EBS users to convert to their own Application Operating System. Well with PeopleSoft and JDE’s combined 11,000 customers plus their own 10,000 or so EBS customers (the numbers are hard to pin down), it gave them a pretty good answer to SAP’s 25,000 strong customer base. It clearly wasn’t about the apps (what company really wants to own 3 different financials packaged applications if you count EBS, PeopleSoft and JDE). This move was about owning the future of IT. It was about owning the operating system that all applications were going to be built on in the next 10 years.”
Doing some rough math, and assuming that every customer they bought converts, that’s about 1M per customer spent (16 Billion paid for approximately 6000 (PSFT) + 5000 (JDE) + 5000 (Siebel) customers). To give Oracle credit, even if Oracle only converts 50% of their new found customers and they have to spend another five billion more building Fusion and its applications, that’s still a pretty good price per customer. I really hope that some analyst works out the exact conversion number at which this ended up being a bad deal. Right now my gut tells me they are well under this ceiling. So in the end analysis, all technical grumbling aside, I have to admit Oracle is making a smart move with Fusion (which I also admitted in that blarticle).
The New Adult Alternative
So what choices does that leave an Oracle customer? As a start, Oracle did stress that the path to Fusion was “eligible” for their customers but not required. They did recommit to their lifetime support policy for customers who don’t want to make the change (more on the ins and outs of this policy in an upcoming blarticle). And in a gesture of good faith, they are slapping some Web services front ends on existing application to allow them to integrate into their SOA based Fusion Middleware Platform.
For most customers, my guess is the “Wait and See” strategy is the one they will take for the next few years. For the customer, and thanks to Oracle's help, there is no harm in doing so. But there is huge risk to Oracle in having their customers sit around like fish in a barrel while they figure out how to build the world’s best expense reporting application. SAP and the likes of TomorrowNow have already reloaded, and the Oracle customer base is squarely in their sites. If you believe most of what I have said, there is a very simple argument that can be made by SAP, “If you’re going to have to move to a completely new development platform and do the gap analysis on completely new application functionality, why not take a look at SAP in the process. And oh yeah, you can get it right now, like today for example, with the completely same technology stack, still use your Sybase database, and we already have the applications written”.
I fully expect there to be 0% unemployment in Waldorf by the end of the week, as I am sure SAP is hiring every Thomas, Dieter, und Heinrich off the streets to hit the phones and start calling Oracle customers.
Agnostic or Unitarian
One glaring omission from Wednesday night’s episode was any comment regarding Oracle’s religious database affiliation. Remember, one of the big differences between EBS and all the other applications they bought is that EBS is an Oracle-only product. The others are database agnostic. In fact, a lot of the technology and toolset of PeopleSoft et al is designed to accommodate this choose-your-own database requirement. It would not be a stretch to think Oracle will head back to a Unitarian stance due to the fact that EBS is built on Oracle and Fusion is built on EBS. While it appears to early to tell, Mr. Phillips did show some of his cards in the Unitarian direction in an interview with eWeek. If I was a PeopleSoft, J.D. Edwards or Siebel customer not running Oracle, this is absolutely the first question I’d be asking Monday morning.
By all measures, Oracle’s latest A-Team episode did not disappoint. Troubled companies (PeopleSoft and Siebel) have been rescued, a Frankenstein of applications has been assembled to take on the bad guys, and it appears that despite all odds, Oracle may still actually win this battle. But as with any good A-Team episode it would not be complete without the appropriate catch phrases. As the lights were about to dim on City Hall, Charles Phillips walked on stage for his wrap up remarks. I can’t exactly remember the quote, but I’m pretty sure he said, “I love it when a plan comes together”. Pure A-Team magic.
For those die hard fans of the A-Team, you’ll be quick to point out that I forgot to mention one of the key cast members. And indeed I have, because much like his character on the show, when the fight was on, he was nowhere to be found.
Larry Elison, Chairman and CEO, a.k.a.
Lieutenant Templeton "Face" Peck - The A-Team's pretty-boy was good with the women, but when it came to fights Face was usually nowhere to be found.
Setting aside all my other comments, I have to say that I was pretty disappointed that Larry did not show up. Yes, I know, he had the flu, but I was looking forward to seeing him live for the first time. I’m not too worried though. I have a feeling there are still a few application damsels in distress and Oracle’s A-Team can’t be far behind.
Note: If you want to Subscribe to this feed, click here.