Cool IT projects come and go.
Social media platforms used to be a cool IT project. VDI used to be a very cool IT project. Then it was your first private cloud.
As the unofficial customer coolhunting dude at EMC, I hereby annoint my candidate for the next cool IT project: creating a BI-as-a-Service capability and pointing it at your analytical power users.
We're doing it in our own business, and I've now met more than enough customers who are interested in doing the same for their business.
And, just to be clear, this isn't your father's business intelligence.
The Stage Is Set
I tend to see the world through a lens of opposing forces with conflicting goals often followed by smart leaders proposing win-win scenarios. That sort of drama is certainly playing out here.
On one side, we have business users who are increasingly motivated to consume analytical data. It used to be about traditional historical reporting, but it's now changed.
They want access to as many diverse resources as possible: some filtered, some unfiltered. They want to experiment with the data to find new and powerful insights. Maybe they even want to bring in a data science professional to work their magic.
On the other side, we have an IT group with predictable concerns.
They're concerned with resource consumption. They're worried about potentially bad things happening if data consumption proliferates without a watchful eye. They dislike tool proliferation. They dislike unsanitized data. And they think that -- somehow -- people should be coming to them for "the answer" rather than enabling users through a catalog of services that business people need.
Bumping up a level, I am a strong believer that motivated business people eventually find a way to get what they need. It's up to the IT group to decide if they want to help or not.
More often than not, I know how that movie eventually ends.
Enter A New Consumption Model
If you know anything at all about server virtualization, you're aware it can make infrastructure resources really easy to consume: easy for the people providing the resource, and easy for the people consuming the resource.
Once IT traditionalists get over the inherent horror of making IT really easy to consume, there's usually a general appreciation for the win-win all around.
OK, let's take that model and pop up a level.
Let's make analytical data (and its supporting resources) really easy to consume: easy for the people providing the resources, and easy for the people consuming the resources.
Once IT traditionalists get over the inherent horror of making analytical capabilities really easy to consume, there's usually a general appreciation for the win-win all around.
What Makes Analytical Data Hard To Consume
If you look at it from a business user's perspective, your first challenge is you don't know what's available for potential consumption. There's usually no place you can easily browse for potential feeds: sanitized and cleansed, raw and unfiltered, etc. etc.
Indeed, talk to any community of power users, and you'll quickly realize that there's usually a grey market in data sets: people freely trade and share what they've got, along with their insights and experiences.
Huh -- shouldn't we be enabling and supporting that behavior vs. discouraging it?
Good power users like to experiment with their data. They also like to experiment with new tools and methodologies as well. Juxtapose this desire against IT's typical practice of limiting access to computing resources (as well as tools!) and you can see the conflict.
If the goal of the analytics crowd is to empower the business to make better decisions, who's right and who's wrong in this debate?
Finally, the best power users are really communities: passionate practitioners who want to share, discuss and debate their insights with like-minded people across their organization.
It's one thing to send a few powerpoint graphs; it's another thing entirely to invite someone into your analytical sandbox to poke around for a while.
Getting To Easy
When I pitch this concept, I try to make it as simple as possible. It's all about getting to "easy": easy for your business users, and -- as easy as possible for the IT crew that has to support them.
The hard part for people is inevitably inverting the viewpoint. Rather than start the discussion with "what works best for IT", starting the discussion with "what works best for your business users" and working backwards. For example, users should get to use the analytics / data management / ETL / classification / ontology / workflow tools that they want to use for their individual use cases, as opposed to the Single Standard As Defined By The IT Experts.
Strange, I know.
The Components Are Familiar
In the EMC world, you can see the pieces we'd use to construct a BI-as-a-service platform. We'd start with Greenplum UAP -- the software environment that makes it easy for analytics professionals to discover, provision, experiment and share freely.
We'd probably put that on a good, virtualized platform like a modest Vblock, or a portion of one. We'd add in a few bits that might be unique to the customer's requirement: VDI, security, data protection, etc.
And we'd offer up implementation services, as well as perhaps a small consulting engagement to help program manage the introduction of the new capabilities.
Not to mention a bit of advanced training for your power users :)
Common Customer Scenarios
So far, I have come across three scenarios that capture 95% of the use cases around this topic.
The first scenario I call "controlling database sprawl". Everyone in the business has their own private decision support database, complete with feeds in and feeds out. The IT group doesn't get involved much, except to provision resources and the required data transfers.
Dozens to hundreds of these small databases aren't unusual -- and I can tick off customer examples where these creatures number in the thousands.
Is this a problem -- or an opportunity?
Obviously, there's a strong hunger for decision support data that's not being met by the current IT landscape -- why don't you think about embracing the demand vs. fighting it? The decision support databases will exist -- with or without IT's involvement. The only decision IT can make is how to best manage the demand.
And any form of shared service on demand will be vastly superior to the ad-hoc messes I routinely encounter: better for IT, and better for the business.
The second scenario I call the "progressive IT leader". I know you may be doubtful, but I routinely encounter very forward-looking IT leaders who are always on the lookout for the next big "win" for their business.
Sure, they have to deliver cost-effective services and run the factory, but they want to get ahead of the curve vs. always reacting. They understand what's starting to happen with big data analytics. They can easily see a world where their businesses are very analytically proficient.
And they get to thinking about what they can do to accelerate that proficiency from an IT perspective.
I would put EMC IT in that category.
The third scenario is, plainly put, "we're in serious trouble".
The analytical crowd has joined forces, and has started to storm the gates of the IT castle with pitchforks and torches. The conversation has escalated from "IT doesn't meet our needs" to "our IT guys don't have a clue".
There's more at stake now than anyone wants to admit. And the IT guys are looking for immediate help in de-escalating tensions between the two factions.
I would like to think that most IT organizations are in the second, more progressive bucket. Sadly, they appear to be vastly outnumbered by the other two groups.
It's An Easy Story To Tell
Your business is becoming analytically enabled: maybe with IT's help, maybe not.
Business users are demanding ease of consumption to as many data sources as possible with the resources they need to experiment and collaborate around their insights.
You have a decision to make -- do you invest in making analytics easy to consume, or do you resist?
And, although standing up a BIaaS platform is certainly becoming a "cool" project these days, there's usually far more at stake than a bit of bragging rights amongst your peers :)
Clearly EMC IT is more enlightened that the typical enterprise IT shop where the data warehousing guys have spent years, possibly a decade, getting the "right" data into the "golden" repository which fosters a sort of inertia, becoming a barrier in and of itself.
These traditional practices are at odds with the agile concepts and iterative design philosophies many are evangelizing today, and although BI-as-a-Service is cool versus the lame way most BI is done today, it's going to take awhile to realize a future where business analytics are consumed via an AppStore delivery model.
Enterprises are definitely looking to the likes of EMC Greenplum and its host of BI application partners to lead the charge.
Posted by: Michelle Agul | April 03, 2012 at 07:56 PM