Saying Yes!! (or at least Maybe) to NO SQL

Since the beginning of this year, the amount of chatter around the “NO SQL” (Not Only SQL) topic has increased. EDW conference seems to have made more people aware and whether on LinkedIn or other forums, there is a health amount of information exchange going on.

  • Some are excited about the possibility of a new way to analyze data
  • Others think this would make data quality management even more difficult

I think both groups are right. I also think it wouldn’t necessarily make data quality management more difficult, but make the challenge more obvious.

Where we all seem to agree on is the need to understand what the different tools are and the core strengths of different approaches. I think as a data profession, we need to be personally accountable to understand what can offer value to our colleagues and customers, and even if we don’t have a lot of time for research. I fundamentally believe that even if I don’t make the time to be aware of how a new technology or process may be of use, if the value (or marketing 🙂 is good enough, others will experiment and then when things stick, we may be playing catchup. I think it would be an err for the data profession is to repeat the same mistakes that were made when XML was first becoming popular and leaving these data structure definitions to the developers.

So saying Yes, No, or at least Maybe to NO SQL or any other innovation it up to us. In many instances, our experience with past technologies (relational, IMS…) can carry forward to how new technologies can make older approaches more scalable and viable. For the truely innovative thinking, it should at least be interesting enough to do some reading on it.

We live in “interesting times”, with the excitement and challenges of it all.

Another Dance and Leadership moment, still with fun

When I wrote “On A Dance Lesson, Leadership, and Letting Go“, it was meant to be a stand alone entry. That said, I think there will be many more parallels to discover and share, so here comes the next one in the series.

There is a club that hosts live Swing dancing one day a week and I try to go whenever I can. It is fun, low key, and good exercising on top of it. Since it is a social environment, there aren’t many pressures and people with different backgrounds, ages, dance levels all come together to have a good time. This means sometimes

  • new followers can do much more than they have learnt to date, when there is a strong lead
  • followers don’t do as much as they can if the lead doesn’t know the same moves
  • as long as partners have a good attitude every dance ends up being fun

Effective Leading
In swing dancing, you often have one or both hands of each dance partner connecting some way. So it is quite possible and often easy for a lead to lead their partner into multiple or elaborate spins and turns, making a nice show and an engaging dance. In general, there are three main ways a lead can achieve this.

  • partner already knows the queues for the various moves
  • lead can use their lever and hand-hold to move their partner when/where/how they need to go
  • lead and follower agrees on basic leading approaches, vs the specific moves that may have a longer learning curve

When you and your partner (team, customer, …) are in sync, doing what you’ve frequently done is effortless, and can look elegant as there are few mistakes. More on this later.

If you know the moves and can move your partner the way you need, one can still do relatively fancy moves. Sure a step could be missed here or there but often, you can get back into the beat and most of those around you wouldn’t remember it. If the follower is new, they are probably thrilled that they did so much. Perhaps this was all they were ready for even.

The challenge is not every step can be done through physical strength of just the lead or look well done. Also, as people gain more experience and confidence in dancing, they want to understand how things work better. So we have the third option: agreeing on how to lead and follow. This could be as simple as “if I pull back, mirror and pull back. If I lean forward, lean in…” and then start using this new found balance to still do the same moves.

There is often a single common barrier to get leads from option 2 to option 3: someone having mentioned it to them first. This way, they would know how else l to lead, more effectively and softly, without the harder, energy consuming efforts. Once the lead understand the difference, he or she can pick the right approach for the right moment.

Two weeks ago, I met a self proclaimed “first timer” who was there with her friends, looked unsure and almost guilty she didn’t know as much when people would dance with her. We did a few dances and the third time I asked her if she can close eyes and focus just in the basic step, and let her body follow the lead. (this is not that different than the workplace, when as leads or managers we ask people to believe in us and follow). It worked out, in fact she was on beat more than before and that was that.

This week, the same group was there again. I talked to the two people I had exchanged dance step ideas with, mentioned to the third person I was glad she had come back. On the dance floor, she looked more comfortable and later even mentioned the previous week’s exercise of just doing it had really helped her. As we got ready to dance, I decided we would try to focus on how I would signal what I was going to so just based on one hand’s resistance. Few minutes after that, we were moving more smoothly and she was more comfortable, so that when her friends wanted see the move, I suggested she show it. In her response, there was confidence and eagerness. I feel there is a new dancer now who is comfortable not only to try and but also teach new things. Isn’t this what we try to do at work with our teams and staff? Continuously get better and confident, and even better have fun doing it?

Before closing, remember I said “more on that later” on people that knew each others moves extremely well? What if they did that without the simple queues or the same lead techniques that would help them spin faster or do more complex steps, simply because they were used to what they always have done. I am not saying this is a problem, as I do believe people have a choice in what boundaries they set. I do however believe there is always an opportunity to learn and improve, and many ways even make things easier vs more difficult.

That’s all on this topic. Next post, I may explore followers than know more than the lead, and what can be good about it.

Cheers

Data Models, Date-A-Models and the missing course content

Fellow data professional, Chris Bradley, had a great story from a conference. A passer by had asked what the event was about and the answer “it is about how to data model” came across as “how to date a model”. Chris’ story telling is priceless as the dialogue had gone on for sometime until the disconnect was discovered, but if got me thinking, could we get more out of this story?

When people learn how to data model, there is a lot of focus on understanding the concepts. What are the Entities? How may they Relate to each other? Could this be a one-to-one relationship?

Not that I am speaking from personal experience, but presumably if you want to date a model, you’ld also want to be aware of what kind of relationship could be possible, what other entities are there to see things may fit and how lines are drawn.

Whether the course is on how to data model or date a model, there would have to be emphasis on how to manage what you create? Once you get the date, how do you make it worthwhile? Will there be a second chance, a third? Will there be long term value? What ongoing complexities may need to be managed? When it is about a collaborative relationship, how will you deal with unexpected questions, semantic misunderstandings, or gaps?

In Data Management, we often do a great job in defining a data model that is flexible with content and constructs we are proud of. Too often however, the modeler focuses just on the creation and not how the information is shared or stored through the model. What are the quality expectations? How would we know if, not just the model but the system it enables is meeting the need?

This to me is the key difference between a Data Modeler and a Data Architect. A great modeler understands the concepts to represent them effectively, and perhaps through different mediums (relational, dimensional, XML…). A data architect also understands who needs the information for what purposes, what are the available data sources, what are integration or quality complexities and how they could be managed over time.

It is not about having all the answers but understanding what is it that would truly matter and continuously working to achieve it.

Whether you are working on data modeling or dating a model, good luck in your endeavors.

Cheers