Are Open Source Reporting Tools Ready for Primetime?

Z0mb1eman asks: "My company is considering replacing our aging CrystalReports with an open source solution. We are currently doing our research, and the choices seem promising -- JasperReports, Actuate-backed BIRT, and Pentaho, which seems to combine other open-source reporting tools. All have some level of commercial support, but are they ready to replace established solutions like Crystal Reports or even Actuate? Is your company using an Open Source reporting tool, and what have been your experiences with such tools? Are there any other choices we should consider? What should we expect if we make the decision to switch?"

Comments

This seems like the first comment

[superwiz]

... on Slashdot of all places. That alone should indicate the amount of apathy the open source community has towards this.

Pentaho?

[Anonymous Coward]

Sounds like a word in the educated pimp's dictionary.

pentaho -noun: A group of 5 hos.

Ideally, you need to have a pentaho on each corner you control.

JasperReports

[speedy1161]

At my former place (a Fortune 50 retailer), we used JasperReports for a lot of our internal reporting needs. Generating PDF's for the web-applications, XLS docs for the business folks, and using all OS/Free software to do it.

We used JasperAssistant to create the reports, training even non-technical folks on how to create the reports they needed. The reports they made weren't going to win awards, but they got what they wanted and we saved time.

Just be aware that the JasperReports libs do their own thread management (at least it did as of the pre-1.2 release) so be aware of running inside of a J2EE container. We chose to write an asynchronous app that utilized JMS and a java daemon that read the queue and processed the messages, storing the output as a BLOB in a db.

Re:JasperReports

[chellbelle1000]

I was asked to find a web based solution to replace Crystal Reports. I decided upon Jasper Reports, which as a programmer I found to be great. There is a lot of choice in report designers - the ones I tried were iReport, Jasper Assistant (eclipse plugin). As a programmer, I found these great to work with. I also found it easy to integrate into our java web application, with the use of both database and xml datasources. It was also easy to print the report to pdf or html.

The graphing support is not as advanced as that in Crystal, I had to hand code a lot and therefore it would have been impossible for non-programmers/customers to include graphs in reports.
I also found it difficult to replicate the "drill down" feature in Crystal Reports.
Both of these points were extremely important in the replacement solution, and thus Jasper Reports was scrapped, deemed not to be as user friendly, and rich in advanced features as crystal :(.

So, I guess - moral to the story - make sure you have a set of clear features you require and that what you decide meets them.

I use BIRT

[Anonymous Coward]

I was the first easy to use system that had powerful report creation. You can drag and drop from palette to create a layout which is pretty nice. It has smart charts (coloring based on marker values), PDF and HTML output. It supports POJOs (scripted), JDO, JDBC, Hibernate and is fully integrated with Eclipse if you are into that side of things.

Another good thing is that Actuate has really responsive support and there are lots of examples, tutorials, and online help, etc.

The only real downside so far - installation is a bear (they have promised to fix this soon) and there are still some bugs here and there (though they are responsive via newsgroup, bugzilla, and email.)

Why are you switching?

[Timothy Chu]

You haven't talked about why you'd want to switch. If something works, why change it?

Re:Why are you switching?

[Timothy Chu]

First off, I'll state that I do work for Business Objects.

Version 8...that's almost six years old (quite old in terms of reporting technology). Are you developing with Report Designer Component (RDC)? We currently support report modification through something called a RAS API, which might suit your needs. It can be accessed through .NET (which is how most of our RDC customers upgraded). Unfortunately, this is a completely different API from RDC. The good news is that the reports format itself is backwards compatible.

I'm not in sales, so I can't tell you how much it would cost compared to open source alternatives. All I can do is tell you that it can be done.

Re:Why are you switching?

[petard]

Huh. Business objects buying the crystal reports line slipped completely beneath my radar. Having used Jasper Reports, Crystal Reports, and what is now (I think) Business Objects Enterprise, I'd rank costs something like this:

1. If you have developers handling your reporting anyway, the open source solutions are hands down cheaper. You'll get reports better tailored to your needs for comparable levels of effort with lower up-front costs.
2. If you have non-developers handling your reporting, Crystal Reports is a false economy. Take either option 1 or option 3. You'll get more useful reports at lower effort.
3. If you have non-developers handling your reporting with good DBA support, Business Objects enterprise is fantastic. Though the up-front costs seem insane compared to the alternatives, BO enterprise actually allows a moderately tech-savvy business analyst to create reports herself, at least on a one-off basis, without an immense effort. She'll get exactly the data she needs for her analysis without custom development. But you have to just get over the up front costs.

So to summarize my memory of the costs:

1. (OSS) $0 up-front costs, 3x developer effort, 1x business analyst effort, business analysts might get exactly what they need.
2. (Crystal) $700/seat up-front, 2x developer effort, 2x business analyst effort, business analysts might get exactly what they need.
3. (BO Enterprise) $1500 up-front, 0x develoepr effort, 1x DBA effort, 1x business analyst effort, business analysts definitely get exactly what they need.

My knowledge of these products is admittedly a bit dated, but that's my recollection. If your management is clueful enough to buy Business Objects Enterprise, the savings in the end more than cover the up-front costs if you have good Business Analysts and good DBAs. Of course, for any of this to matter the reports have to actually be important and useful. If the measurements you're reporting on are BS anyway you should always go with option 1 so that maximum savings can be realized by scuttling the reports.

Depends what you need!

[mikeburke]

Have you identified what your requirements are from a reporting tool, be it open source or commercial? Your definition of 'prime time' is totally dependent on this

- who will be writing the reports (techies, business folks, both)?
- what layout and formatting capabilities are absolutely needed?
- any really big reports? performance may be an issue
- what are the security needs - authentication, visibility, auditing, etc?
- do you need overnight automated report runs?
- what about bursting (automatically splitting a report into sub-reports based on department, product type, whatever)?
- do you need to integrate with custom developed software? what language+platform, etc?

The first point is particularly important - if business staff want to dolly up simple adhoc reports, then this will seriously narrow down the open source field pretty quickly.

Cognos should be on your short list

[camt]

I'm a Reporting Engineer, so business intelligence is my speciality. I do not work for a company that sells BI software or services, I'm just an administrator/power user. I've used BIRT, Crystal Reports, Crystal for Eclipse, MS SQL Reporting Services, Excel PivotTables with SQL Analysis Server cubes, Proclarity, Cognos ReportNet, Cognos Impromptu and Cognos PowerPlay.

Business Objects seems to have a pretty solid platform these days, but the company tends to use underhanded advertising techniques and make dubious claims in their marketing material. I'm not sure how they are once you are a customer, but I have been unimpressed with their sales pitches to me. Microsoft SQL 2005 has some pretty neat tools bundled. They have really come a long way, and they are fantastic on the back-end or for use internally to the IT department, but they really have nothing in the way of user-facing front-ends that I would consider ready for "prime time". BIRT looks promising, and within a decade I would say it will be fantastic software.

I highly recommend looking at Cognos Series 8. The back-end is a web services based framework and the zero client front ends blows everything else I have seen out of the water. It is simply the best platform I have *ever* used, though it can be a bit pricey. They used to work with you quite a bit on the price; not sure if they still do that with their new named-user-based licensing model. Their support is excellent as well. I was an early adopter of ReportNet, and Cognos flew an engineer down to our site on their dime to troubleshoot a critical bug; we were not a big site, at only 60ish end-users.

Call up a Cognos rep, they are usually willing to come on-site to do a demo.

Re:Cognos should be on your short list

[mooingyak]

I currently use (aka am stuck with) Cognos. While there's pretty much no chance of us replacing Cognos with something else, we've recently also looked BIRT, Actuate (actuate is currently supporting the BIRT project), Oracle, Crystal Reports, and Jasper Reports among other things. They don't really occupy the same market space. There is some conceptual overlap, but Cognos (or at least, what I've used from Cognos) tends to be a snapshot image of data while the various other solutions are more along the lines of on-demand reporting against a potentially volatile database.

Re:Cognos should be on your short list

[GuyverDH]

Business Objects, once you are a customer, spends more effort in telling you why your problem isn't their problem, even when you use their *default* installation and setup.

They are constantly changing the layouts and deployment locations of their sub-components, and make it nearly impossible to throw a decent proxy front-end on their applet, so that it can fit into a standard (for our company) multi-tiered/multi-firewalled layout.

It used to be that the docroot for all applets was /businessobjects, now they are moving them in a seemingly random manner. That's not the biggest problem either. Their current incarnation, will not allow the web server front end to be configured for SSL, while their Tomcat listener is non-ssl. For some reason, they hard-code the url components to be non-ssl, and the proxies we've tried cannot override this (at least the time we've spent attempting this has been unsuccessful). We finally enabled SSl at the application layer, and changed the proxy to simply use a passthru mode. It's ugly and it's not what we'd prefer to be doing.

We've been waiting close to 6 months for a fix to this problem. One they were able to reproduce in their office, 6 months ago. Recently, when asked by us about the status, they stated that "We were finally able to reproduce the problem, and have sent it off to engineering." When we mentioned the fact that that's where we were at 6 months ago - they were like "Oh!"...

Anyway - that's a customer's experience with Business Objects - and to put it mildly.... In my opinion, it sucketh greatly.

More details...

[Z0mb1eman]

Wow, my first accepted submission :) At last, an excuse to read Slashdot at work tomorrow!

I kept the submission fairly general, to avoid steering the discussion in any specific direction. Here are a few more details of our situation:

We're a small-to-medium business, and our reporting needs are relatively modest (at the moment - we are steadily growing). I would say no more than 10-20 people need to generate reports. Our software department is also quite small - we have a fair amount of in-house Java expertise, but no one who actually knows Crystal Reports. We're also using an old version, and the reports themselves are out of date. Our choices are to invest in a new version of Crystal, the time for one of us to learn it, and rewriting the reports anyway - or choosing an open-source Java solution and hope for the best. The constraint, as always, is very limited manpower...

Any feedback is welcome - I've seen some very good posts already, I'll have to read all the comments tomorrow in detail.

Thanks!