Microsoft PPS: Dead.

Nigel Pendse reports the death of Microsoft Performance Point Server:

Despite all the positive auguries, it seems that PerformancePoint Planning has flopped.
Microsoft announced on January 23 that it was to be discontinued. There will be a final SP3 Planning release in mid 2009 to deal with immediate customer issues, but no further enhancements. It is possible that some planning functionality will be added into other Microsoft products, such as Analysis Services and Excel, but this will not be based on the abandoned PerformancePoint Planning product.
The Biz# project will have run for almost six years from commencement to abandonment, with little to show for it apart from some disappointed customers and partners.

Disappointment indeed. It's now official. I can say with finality that I wasted a year in my career. In fact, a lot of good people that I know personally wasted a year and even longer chasing the rainbow that was Microsoft Performance Point Server. It's a damned shame because some very smart people worked very hard and some hard earned money turned out to be a bad investment. I personally know people who went halfway around the world trying to make Microsoft PPS do what it was supposed to do. It's not just the clients of Madoff who are pissed these days. 

I was fortunate to get hit over the head with a brick almost exactly a year ago when my boss made the judicious decision to trim his sails and begin cutting his losses in an ambitious project to corner the consulting market on PPS. We were the consulting vendor of choice. We proved that we could do things with that product that almost nobody else could and we survived on hope, skill and the kind of gallows humor dedication of the dogfaced GI. 

There were several occasions when I had to stop and scratch my head wondering if it was just me or were we pushing ropes. But I'm no quitter. If there was a way to make that stuff work, we were going to do it. Now what I remember of the story can be told. 

First of all, PPS was a decent product. It had particular strengths and weaknesses that you might expect from a version one launch, but in my revised opinion - the kind of opinion I could afford to develop after I left the team, there were certain things that simply doomed it. Only one of them had to do with the technology.

The first and foremost thing that doomed PPS was that it was a Microsoft product and what I learned was that Microsoft, certainly as far as PPS was concerned and probably in general is a technology company, not a product company. That is to say that Microsoft is the kind of company that is well suited to marshal its resources on a global scale with a large existing installed base of interlocking technologies but that ability doesn't scale down. Microsoft is incapable of profiting from small products with small markets, and from a Microsoft perspective PPS was tiny. 

Secondly PPS itself belonged to nobody. Not that there were 'nobodies' in charge of it. In fact there were multiple big somebodies in charge of various parts of it. There lies the problem; there simply was not one integrated product organization that could make decisions in support of the product and then stand behind those decisions internally to Microsoft or externally to consultants and customers. PPS primarily belonged to 'OBA', the guys responsible for Microsoft Office. Now I know BPM people are saying what!?. I did too, but that's the way it went. PPS also depended on Excel, which was another group. PPS also depended on SharePoint which was another group. And of course it depended on SQL Server & Analysis Services, yet another group. So who coordinated all of those development efforts? Effectively nobody. 

Thirdly PPS had no product roadmap. The product roadmap for PPS was basically a function of sales. We throw it out there to 50,000 customers, we expect an adaptation rate of 3% at a price of X per pop, there's your revenue stream. Completely absent from that was a feedback loop. Microsoft didn't care a hoot about what customers and developers might have to say about the product's features or shortcomings. It was as if a mighty focus group or marketing exec had spoken and that was the end of it. So when things went wrong, and believe me a lot of things went wrong, there was no recourse. So nobody inside or outside of Microsoft, situated to support the product knew if, for example, there would be a published API for the product. Nobody knew if there was actually going to be a next release or what the priofities might be in that release. For all the things we knew that worked or didn't work internally to the product, we would be stuck telling our customers - well we don't really know when Microsoft is going to address that. In fact, we didn't even know IF Microsoft was going to address that.

Fourth. PPS just didn't perform. I had guys working the guts of this product and one of the things that was entirely too clever about it was that it had a kind of code generator that would determine whether or not it would use MDX or T-SQL to execute its functions. I don't recall what people have said about how well optimized Microsoft compilers are, but obviously none of those geniuses worked on optimizing PPS. PPS would generate a meta-language that was basically full of crap. Our guys were hand-coding all of the internals and forcing the product to use SQL instead of MDX, period. It was the only way to get it performing worth beans.

I recall a famous meeting I had up in Redmond with some really sharp guys that Microsoft had redeployed from FRX to get on the PPS bandwagon. I didn't like it because it meant that Microsoft was hedging on us as a consulting vendor, but we all figured that the market would eventually take care of us all so we played nice. One of the top three PPS architects in the world was there explaining and demonstrating one of the few successful internal implementations. As we went through the demo the architect did a drilldown in Excel. You know me, I counted seconds for response time. ..6..7..8..9..10. After about 12 seconds we got about 50 rows of data. Let me make that clear, 50 rows with 12 columns, 600 freaking cells. This, I know because I asked, was coming off of a 64bit server. Anybody with any experience in OLAP knows that kind of performance was unacceptable back in 1992. At the time of this meeting, there was only one person who actually had any experience doing any back-end tuning behind PPS. All I could say was, well, FRX has done well in the mid-market, so maybe this isn't so bad. But I knew from that moment on that it would be a cold day in hell before we could beat Oracle, Cognos or even Outlooksoft for Enterprise customers. 

And yet Microsoft people had the nerve to believe that their entry into the market - their purchase of ProClarity was this fearsome move that *prompted* Oracle to buy Hyperion and IBM to buy Cognos. Unbelievable. 

I have to say in retrospect, that PPS was the crappiest product I've ever had to work with for its market. ProClarity was 3 times better by itself. ProClarity as a company may have been destroyed here, but there's an object lesson that the brains behind ProClarity and the organization was much better suited to manage that product than Microsoft. I've come to terms with my frustration with Microsoft. I don't hate them, but I understand why people hate them. I have never, in my entire career, had customers so furious at me for a failure to deliver on promises implicit in a product. I have never seen smart consultants so frustrated by a product's complexity and poor performance. I have never seen businessmen have to tapdance so fiercely to keep everybody on the same page.

I've learned a great deal about Microsoft, people and myself through my experience with PPS. I can say this. When it comes to delivering products and services to the enterprise, Microsoft is better to have as an enemy than as a partner.

To all my friends and associates in Redmond and Century City, to all of my customers and partners in Washginton, Arizona, Oregon and Hong Kong, you know I have great respect and admiration for your courage under fire. I'm sorry to have failed you. I did all that I could, but now you know that Microsoft just didn't back us up. 

Now there's somebody who I have to buy a cow. 

Comments

[anon]

Microsoft's new BI motto: "Uncommited BI"

[Cobb]

I try not to laugh out loud when they say that they enable the 'People Ready' enterprise.

[anon]

Some good points but I have no idea what your opinion actually is?

"First of all, PPS was a decent product" or "PPS was the crappiest product I've ever had to work with"?

And just clarify PPS performed just fine if you knew what you were doing

[Cobb]

I thought I made a convincing case that PPS performance compared disfavorably with BI products from a generation ago. 12 second response time for 600 cells is the kind of drilldown performance I could get in the days before Windows 95. Furthermore, nobody knew how to do capacity planning for PPS. If I were to tell you that I wanted PPS to handle 50 concurrent users against my GL, how much hardware should I buy, you'd only get a fuzzy answer. That's because nobody had seen PPS work on that scale.

If you compare PPS to other products it competes against in the enterprise space, it's crappy. If you have simple requirements or are in the SMB space, it's decent. The problem was that Microsoft expected it to compete in the enterprise space. That means that you'd have implementation teams with only one or two years' experience on PPS side with a version one product competing with implementation teams with 8 or 9 years' experience on a version 9 product.

So I had teams of people with 10 or more years experience in Data Warehousing on projects tell me how much they disliked working with the product. I could look at what it took two people a couple weeks to do in PPS that I could do single-handedly in Essbase in a day or two. If I remember correctly, PPS could not handle ragged hierarchies or measures dimensions with more than a couple thousand members. I think to be fair I should do some of those close up comparisons, but since the product has no future there's not much point in it. Smart customers will move away.

One of the biggest problems with PPS is that it had no facility for migration between environments. We had a customer where this took several days to accomplish and it didn't work the first time. So we had to call in a second expert consultant, whose time was split between multiple other customers, and it took that person a week or so to fix the mess and do it right. As these things happened we learned the hard way that now we should schedule two weeks for cutovers with 1.5 bodies. That's an example of PPS being more expensive to develop than the competition which has migration facilities built in. You'd ask Microsoft if they would add migration support into the product and you'd get a fuzzy answer.

Now this is not a problem for a small business that is only thinking about building one or two applications. But for an enterprise platform, it could easily be a deal killer. But you wouldn't even have this kind of discussion and insight with a Microsoft sales guy because they don't get that deep into implementation questions - there were no channels to get that information back to sales. With mature products in the BI space, total cost of ownership questions are part of the sales pitch and have been for years.

Probably the most frustrating thing about PPS was the way that it 'handled' business rules. Without going into much detail, there were no shortcuts. This meant that for certain types of data models you would have to split the application into multiple pieces with different sets of business rules working on different parts of the PPS schemas independently. When there were dependencies between them, it made design and implementation a real nightmare.

A simple example would be if you had a metric like profit per employee on a quarter by quarter basis. Everybody in BI knows that headcount aggregates differently than profit. You would have to handle the business rules for those two sorts of aggregations in separate runs instead of one. So if you wanted that metric on the same report as other metrics you would have to worry about precedence and manage that manually and also look at all of the edge conditions etc... a real pain.

PPS performed just fine if you knew what you were doing. But what everyone in the PPS world knows is small compared to what everyone in the Cognos, Business Objects, Oracle and Outlooksoft world knows. And it will remain that way. If your business requirements are simple, and you actually like Sharepoint, save your money and go with PPS. If you want to compete with the big boys in EPM, forget Microsoft.

Oh yeah. UNIX.

[my2cents]

I think PPS planning was built to me more generic flexible and powerful and initial release has some perf issues but they have significantly improved performance and migration issues in SP1 and SP2 releases and performance is not a issue now.
I think V1 was released in hurry because of long delays.

Microsoft hoped that partners will develop the integration and migrations modules but none implemented. Some good graphical integration and migration modules might have made the difference.

Many company’s has issues implementing with the initial release because there was no good knowledgebase and trained consultants and best practice implementation guidelines, but SP1 and SP2 had solved lot of issues and are surprised now.

I think the product rightly belongs to dynamics team but Forecaster is a toy compared to PPS planning which is a powerful enterprise tool, I think Forecaster which is a legacy tool should be replaced by PPS planning.

I think V2 of the product with minor updates would have been on par with competitors(Cognos/Hyperion) and more powerful.

[Cameron Lackpour]

Wow, I thought I had worked with some turkeys -- you unfortunately won the prize for Most Irritating Product implementation.

It really is a question of vendor commitment and perseverance. Remember the disaster that was Planning 1.x? I remember taking the 1.5 training class and thinking “How could they release this?” It wasn’t a really decent product until 3.3x. It’s not half bad now. :)

Regards,

Cameron Lackpour

[Cobb]

Yes. I remember getting seriously burned by scalability issues in Planning 2.x. It was, unfortunately, one of the biggest planned Planning implementations in the Hyperion world at the time. But I will say this - we had the ability to get engineers working on patches for customers and reprioritization of development and release schedules. We were able to get Hyperion management involved in escalation and modify delivery frameworks, even though it wasn't Hyperion consulting that was the prime consulting contractor. That's the difference a commitment to a product makes for an enterprise customer.

[Joe Horton]

I convinced our state agency to go with PP and Proclarity. We paid their choice vendor quite a bit of money to come from Seattle and train about 12 people on site.

We did manage to get out some fairly decent dashboards and reports but it is so lacking in basic things. No decent control for legends, we documented some 30 serious defects that was supposedly "passed on" to Microsoft and got no feedback.

I feel really burned by this. It wasn't even two months ago they had big BI presentation still pushing this product.

All I do is Data Warehouse BI implementations - believe me I will be looking always to the other vendors and recommending to my clients and bosses, despite being neighbors to Microsoft.

[Doug Moran]

Pentaho open source BI won over PPS Guru and author Ron Barrett http://www.networkworld.com/community/node/39219 He found it easy to switch "It was simple to go from Performance Point to Pentaho and I was comfortable with it in minutes"

[Frank Mustari]

"disfavorably"?....ain't no such animal.

[buy viagra]

The X Games is a commercial annual sports event, controlled and arranged by US sports broadcaster ESPN, which focuses on extreme action sports. The inaugural X Games was held in the summer of 1995 in Rhode Island.

The Winter X Games are held in January or February (usually in January) and the Summer X Games are usually held in August, both in the United States. Participants compete to win bronze, silver, and gold medals, and prize dollars. The competition often features new tricks such as Tony Hawk's 900 in skateboarding, Anthony Napolitan with the first ever double front flip on a bicycle and Travis Pastrana's double backflip in Freestyle Motocross. Concurrent with competition is the "X Fest" sports and music festival, which offers live music, athlete autograph sessions and interactive elements. The location of the Winter X Games is in Aspen, Colorado through 2012, while the location for the Summer X Games is in Los Angeles through 2009. The X Games also has international competitions and demos around the world that are held at varying times throughout the year.

[reverse lookup]

I guess I am unfamiliar with this. Never heard of it.

[online viagra]

I like microsoft as they always introduce
variety of software. Microsoft Performance Point Server is now definitely dead as its place
is taken by another software from Microsoft itself.
Thanks a lot for providing this nice information.

keep posting such a nice post.

.......Alex

[RamonGustav]

I think you have a thorough understanding in this matter. You describe in detail all here.

[buy vicodin online]

If you need more traffic to your website check out the website in my name. It really helped me and i think it can help your website.