The rumors of the demise of data analysts have been great exaggerated. It seems that the latest trend in BI is talking about how now because tools are so powerful that there’s no longer a need for data analysts. And certainly while vendors have done a good job of driving down total cost of ownership and implementation time if anything there’s more of a need for good data analysts now than ever.

The prior five years was the age of the Developer. Stuffed into their corners they dreamed up and built all kinds of systems to take in data and get data out. But now technology has gotten to a point where a developer isn’t the lynchpin anymore. Some have taken this to mean that there simply isn’t a lynchpin but that is not the case.

Companies have never had so much data so easily accessible to them. But even though our airplanes have very advanced autopilot functions no one would fly without a pilot. So too are companies in need good data analysts to help them traverse the swathes of data that are available, ensure that the right questions are being asked and that the answers given by the data are pertinent and helpful.

Through technology we have pulled data out of its dragon guarded castle and have begun to socialize it. Data analysts are there to make sure the data socialized is data worth having. We are now in the age of the data analyst. Viva the data revolution!

Hi all, in our business intelligence blog I made a post about 4 things that you’re typically not told when undertaking an OBIEE 11g upgrade.

Feel free to read it here: https://blogs.perficient.com/businessintelligence/2013/01/16/4-things-they-forgot-to-tell-you-about-your-obiee-11g-upgrade/

Yeah, yeah, another post about Big Data. And, yes I know, for someone who complains about the number of posts on Big Data I’m only adding to the noise. But as I heard more and more about big data  I decided to build myself a little program that I could point at companies and track what people were saying about them. After a few months of data gathering I’ve found for a majority of companies most of the data I gathered was just noise. In fact in a large majority of cases (>85%) over half of what the social data out there was put out about a company was put out by the companies themselves or their employees, in some cases that percentage is as high as 98%.

If you filter that out you’re left a tiny bit of data and of that tiny bit only about half is of any use.  So now you’re essentially looking for needles in a haystack. It’s enough to make a man go home and be a family man. However, their rarity doesn’t necessarily make them valuable. Sometimes a needle is just a needle.  Which leads to an obvious question which Michael Wu made a really great post on:, “If 99.99% of Big Data is irrelevant, Why Do we need it?”

The answer? Maybe you do, maybe you don’t. Like I said in my little non-scientific study it was certainly true that most companies put out most of the noise about themselves. However, for certain companies (companies with large customer facing arms such as Hotels,  food, entertainment, etc) it certainly makes sense to make use the swell of information out there.  And certainly there’s big data that lies outside of the social sphere. Perhaps your company gathers data from the product base. That makes a lot of sense. However, for a lot of companies Big data might just not make sense yet.  In the post mentioned above Michael Wu postulates 3 scenarios to help ascertain if you’re ready for big data they are:

1. If you have access to the talent and can do it cheaply. That includes the talents to extract and analyze the relevant data in order to derive insights and value from it
2. If you are a DaaS provider and need the data to serve your customers
3. If you have specific questions, then all you really need is just the “right” data, which is usually not big at all!

I could not agree more and my initial findings back this right up.

If the hype is to be believed, the Cloud is the most amazing thing ever. Something that you’ll want to wad up into your IT life and roll  up to be a singular star in you IT sky.  I’ve seen taciturn CIOs turn into giggling fanboys at the very mention of the Cloud. I’ve even seen respected publications say that ANY risk you could possibly face in moving to the Cloud are far outweighed by the benefits.  If IT had a savior its name would be Cloud.

And why not? The promise of the Cloud is expansive, the possible savings huge and the ease of certain aspects is shocking.  But let’s be honest,  this is the real world and despite the giddyness we tech geeks feel there is no mythical panacea to IT woes.  Pipers must be paid. Certain risks get traded for others.  So in trying to help readers make sense of it all I’ll be writing a series of posts to try and explain the major areas of the cloud and what it means for BI.  This post will cover the Software as a Service (or SaaS) model.

SaaS overview

Software as a Service is essentially what it sounds like. Instead of buying the servers and the licenses those are “rented” (or “subscribed” as they like to put it) in a recurring payment. Many people that have been in the industry a while will think “Oh this is like that ASP thing in the 90s. Yeah…that didn’t work so well.” And they’d be partially right. Yes, it’s similar but with the leaps in technology and support  we have now makes it vastly different and far more viable.

There are many, many players in the SaaS model and there are as many variances in the model as there are players. Many start you off as “co-tenant” with another customer on the server. But you can pay to have your own instance in their data center. You pay more for this but then you have more control of your environment.  Some providers charge you per “seat” others by transactions and others still by volume of data.

The obvious appeal is that you essentially move your licensing and servers costs from CapEx to OpEx and you could save substantially on server, maintenance and licensing costs. Getting an implementation up and running takes a shorter amount of time and as such costs less.

At this point most are chomping at the bit: “I don’t have to maintain servers in my data center? Faster roll out? More control over my licensing? Where do I sign up?!”

Yet with all the benefits there can be draw backs. Yes, you don’t have to maintain the servers in your data center, but you also get less control of said servers. SaaS providers do their best to try and alleviate this, especially with their dedicated instance offerings. But ultimately they wont be able to replicate that feeling of being able to go into your own data center and hot-swapping out the disks yourself.   Then there’s the data security issue.  Some vendors can already meet HIPAA rules and other important standards. Yet again since it is not your data center you do not have the ultimate control so be certain to verify your vendors data security measures. Implementation time might be a lot quicker, but many times SaaS offerings are not as infinitely customizable as their on premise cousins.  You do gain more control over licensing, however unlike the “pay once and you’re done.” model you’re renting, so the cost might be less but it never goes away and can be raised. Lastly, obviously since it’s not your data center in your building you might experience some performance lag. Obviously Vendors work feverishly on this.

As with anything the pros and cons need to be weighed and explored before making any decisions:

Pros:

• More control over licensing
• Potentially lower licensing costs
• No need for a data center or server admins
• Generally faster implementation
• Easy to “spin up” an additional server
• CapEx vs. OpEx (depends on your accounting model)

Cons:

• Less total control
• Offerings tend to be less configurable than On-Premise offerings
• “Subscription” model instead of “Ownership” model as such costs could rise
• Data security concerns
• Not as fast as having your own data center on a different floor of the same building
• If you want to take all the data out sometimes under certain SaaS implementations that can be a problem

What does this mean for BI?

Well, there are two things in essence:  What does this mean if you want to do BI in an SaaS model? And what does this mean if you want to use your on premise BI tool to SaaS tools?

If you’re looking to buy a BI SaaS offering you’ve probably come across it as an addition to a transactional SaaS offering. These offerings usually mitigate the I/O and bandwidth concerns as the servers tend to be in the same data center. However as noted above many SaaS offerings are not as extensible as On premise offerings. As such care should be taken to verify that any customizations or configurations you’d wish to do are possible within the offering. Also, many of the big players offer SaaS offerings that are based on the mature and thoroughly tested On-Premise engines they’ve built. However, given the nature of the beast there are plenty of new players in the space. Some of these can hold their own against the mature offerings from bigger companies, while others are just not there yet.  Very careful consideration should be taken (as always).

If you’re looking to take data from a SaaS offering and put it into your On-Premise BI tool, you’re sort of a pioneer. Some people do this but many don’t. Most end up relying on a SaaS BI offering on top of their other SaaS offering.  Some vendors allow it some don’t. By no means does this mean it isn’t done or is impossible. Make sure you talk to your vendor before planning anything. If you can, the main consideration is obviously I/O and bandwidth, especially if your subscription agreement is per data usage.

Read Part Two: Making Sense of the Cloud Hype: Infrastructure as a Service