Febdback for OMERO staff

[Alan H]

hi, I had a skype call with Jason today, and he mentioned that you folks hear from people when they've got problems, and then when the problems are solved, people just disappear on you, and you never really get any feedback. OK, so I told him that I'd come back to the forum and give you feedback!

!.) everybody who responded to my sometimes very frustrated posts was really helpful, not judgemental, and the responses were quick, almost always within 24 hours. From those standpoints, you guys are doing a really excellent job, so pat yourself on the back.

2.) whoever is the person who writes the code to "decipher" all the different proprietary company image formats and gets them into the proprietary OMERO format so that everything is viewable, is a programming goddess (Jason said so, too) and her work is, to my mind, the real draw of OMERO. So, whoever you are, you are THE BOMB.

3.) OK, problem.....OMERO is open-source. You guys live and breathe Open Source. The problem is that I don't. You know what ICE is. I don't. You know what RHEL 4 and RHEL 6 is/are. I don't. Well, NOW I do, but when I got started in this, I didn't know. You guys think and talk this stuff all day long, I don't, and I bet that a whole, whole lot of your users don't, either. So when someone explained something to me on the forum, half the time the writer made natural assumptions, that I would know enough about LINUX or Open Source projects to be able to decipher what they said. The problem was that I, in fact, do NOT have that sort of expertise.

This means that you have to write/talk to me like a dummy, for me to "get it". On your end, to do more effective support, you have to either A.) spend time figuring out the relative level of expertise of the user you're helping, or B.) talk to them really simply. Make NO assumptions. NONE.

LIke this, I copied and pasted this from the forum...

Re: setting PYTHONPATH variable doesn't "stick"

Postby Alan H » Tue Jul 16, 2013 4:01 pm
Took me a while, like a full day, to figure this out....omero_dist .. the "dist" is the distribution code of that build or beta or whatever. Like

OMERO.server-4.4.8-ice33-b256 ---- the "dist" = "server-4.4.8-ice33-b256"

See, you have to talk to me like a 4th grader or I don't get it!

It, whatever "it" may be, might seem obvious to you, but to me, NOTHING was obvious.

OK, a major frustration here....

When the faculty member (Lucy O'Brien) and I got interested in this, I did an installation on am OSX Mac in her lab. I got it to work with the "homebrew" instructions on the OMERO sys admin web page. Understand this...I have NO idea what "homebrew" is, whatsoever, and I bet a LOT of your users don't, either. When it came time to try to move data from Pilot #1 (the mac) to Pilot #2 (a windows install), come to find out that "homebrew" puts things in all sorts of nifty, special subdirectories. I couldn't find /bin/OMERO for example, and I spent a good hour and a half just looking for the directory. Upshot was, I couldn't migrate data from the first pilot machine to the second.

This is a perfect example of how YOU GUYS know what "homebrew" does, but a fair number of us slobs out here do not.

[Alan H]

Keep in mind that your target "audience" is NOT Open Source Program Developers. Your target audience is scientists who take pictures with microscopes, or the poor grad student or postdoc or local computer support wonk who gets assigned the project of installing this for the lab. These are people for whom there is NO JOY in wandering around the open source online world, looking through massive lists of builds of ICE or whatever, where every single permutation of the program you have to download has nearly an identical name, but God forbid that you get the wrong one, out of a choice of 60 possibilities, because it won't work.

OK, crunch time.

Did I recommend to my boss that we ramp up a big OMERO installation and offer it as a service to Stanford faculty?

Answer: No.

Why? Two reasons: 1-Because it was just too "twiddly". There are too many little bitty parts that have to be searched out and found, and installed "just right". There were going to be updates, was I confident that I could install them? Not really. However, given the time, I'm sure I could get a pretty decent installation up and running. I mean, I did so, the second time, on WIndows Server 2008 after the horrible nightmare of trying to get it up and running on CentOS. The REAL reason was that I kept running up against glitches...little headaches here, confusions there (see"homebrew" and data migration, above), and the spectre of the dreaded "upgrade". The core questions was this....

"If something bad goes wrong, am I 95% confident that I can get the whole system functional so that all my users can see their images and metadata within 24 hours of a failure?"

The answer was "no", and from a services sys admin point of view, that's a deal-breaker. There is NOTHING worse than offering a service and then not being able to support it within an extremely short time frame when something goes wrong.

How to fix this, from your end? That's a tough one if you've committed to the Open Source pathway, and I respect that decision. I don't know the answer.

OK, most of these posts is whining and complaining. Blecch. I hate being a whiner! Go read the first two paragraphs about how supportive and helpful and non-judgemental you all were here on the forum. Also, my initial enthusiasm for the concept of OMERO remains. I think that OMERO's main point, making a bucketload of microscope image formats all viewable and annotate-able under one slick user interface is a GREAT idea, and I hope you guys keep at it.

Also, it looks like Lucy O'Briens lab is going to crank up a small one-lab OMERO server, they've hired a LINUX consultant, who knows this stuff better than I do, to help get it online. I'll be watching the progress of the project with interest.

Alan

[jrswedlow]

Hi Alan

Thanks for all your comments and feedback. Sorry for being slow responding-- a few things going on.

First, it's great you are up and running. We hope the trial in Lucy's lab goes well, and as always we'll support that as well as possible on these Forums. Please do pass on the links to these Forums to the person she has running the new Linux system.

Next, thanks for your endorsement of Bio-Formats. It's a great project, and arguably our biggest success (in terms of installs). So on behalf of Melissa, Curtis, Roger, Sebastien, Chris, Josh, and many others (although really the person whose praises you are singing praises is Melissa), thanks for the positive feedback and comments.

Finally, installing and supporting OMERO. You're right, we assume a lot of background knowledge and expertise. Installing OMERO takes reasonable familiarity with database systems, enterprise Java frameworks, and certainly in your case, multiple operating systems. When you mentioned you were moving from OS X to Win, I think the responses assumed reasonable familiarity with what that meant-- conventions and architecture in Win and OS X are different. If you don't know them, you'll spend a lot of time searching on-line for answers, and it is not easy. In any case, our assumptions were not correct and that was a mistake. Your feedback is important and we've started discussing how to change our approach, docs, etc.

Soooooo... can you and your department use, depend on, and run OMERO? Like many enterprise tools you probably already run, OMERO is complex and takes expertise. It does run at several hundred sites. In our experience, it has run production, multi-TB systems for years, but it is not perfect, and yes, as you stated, we will continue releasing updates, bug fixes, etc., and anyone who wants to take advantage of those will need to have access to the expertise to do so. We've tried very hard to reduce the visible complexity and the difficulty of install, but running enterprise software and managing large data systems does require a lot of knowledge and that won't change. In my experience, many biological research depts assume that the person who handles their computational resources can handle ALL their computational needs and will be expert at every informatics technology that their scientists might use. That's a big ask, and probably unrealistic. Research programmes that want to pursue large, systems-level questions will, out of necessity, become enterprise data generators. If they don't access real enterprise data management expertise, then they'll miss their research goals. Either that means bringing in your own help, hiring a consultant, purchasing a commercial version of OMERO from Glencoe Software, or some other committed investment. That's how imaging, proteomics, and sequencing platforms are usually run; enterprise-level data management requires the same strategy.

In this case, you've been heroic, successfully installed OMERO twice (!!), demonstrated a possible solution for your scientists, and extra help has been found to move to the next step. That's probably all anyone could have ever asked.

Thanks again for your comments-- the whole team appreciates your feedback. We look forward to hearing how it goes from here.

Cheers,

Jason

[Alan H]

Jason, you wrote:

In my experience, many biological research depts assume that the person who handles their computational resources can handle ALL their computational needs and will be expert at every informatics technology that their scientists might use. That's a big ask, and probably unrealistic. Research programmes that want to pursue large, systems-level questions will, out of necessity, become enterprise data generators. If they don't access real enterprise data management expertise, then they'll miss their research goals. Either that means bringing in your own help, hiring a consultant, purchasing a commercial version of OMERO from Glencoe Software, or some other committed investment. That's how imaging, proteomics, and sequencing platforms are usually run; enterprise-level data management requires the same strategy.

SPOT ON. *Exactly*...on all points, IMHO.

Personally, I learned a lot through the process of installing OMERO twice and I'm glad I did it, though I might not have said that at 2:00 AM while wrestling with a PYTHONPATH that wouldn't "stick" on the CentOS installation. If I'd started with Debian, things probably would have gone differently, and I am by no means a Linux whiz.

I'm guessing that you guys are in this "in between" state right now, in regards to your client base. You have both single-lab or even single-person installations going on, on whatever machine might be available to some grad student, who-knows-where, who thinks that OMERO might help him/her out. At the same time you have multi TB, multi-node installations, administered by highly-trained sys admins, handling data streams from multiple large, automated systems. How to adequately support both sides of your client base is a significant challenge. I get that! How do you coach the wizards and the dorks at the same time? .....is not so easy! I suppose part of the answer could be focusing on one end of the spectrum, or the other. The obvious place to focus is the enterprise end, as that's a whole lot more "glamorous" and high-profile, but on the other hand, having that one little lab in some small institution that you are helping out is consistent with the whole open source philosophy.

As a complete aside...and I bet I am preaching to the choir, here... the OMERO/sql interaction, with the really cool omero.insight and omero.web interfaces are an extremely interesting product in and of themselves, totally aside from the BIO-formats part of the picture. Lots of photographers accumulate very large quantities of images which need to be annotated and organized. OMERO absolutely shines at this function. To make a silly point....think about wedding photographers. Trust me, telling the mother of the bride, even two years later, that you don't have their pictures any more is NOT fun. Not that I'm a wedding photographer, but I have friends who are! Also, not all biological image acquisition is microscope stuff! I would consider possible paths that omero.insight and omero.web might go, outside of strictly microcopic image storage and annotation.

Anyway, I think OMERO and Bio-formats is *cool*, and I'll be watching!