Mark's Blog

Episode 103: Flat IT Revisited

I was right.  Well, mostly right.

It has been 5 years now since I proposed the notion of "Flat IT." I thought it would be interesting to go back and see how things turned out. Below is the original blog entry with my comments below each premise. 

=====

In this installment, I would like to outline some of the key changes that will/would occur in a “Flat IT” environment. Here are the some of the key principles.

Infrastructures become Virtual – Virtualization has been a hot trend for some time now and I believe the technology (virtualization) will exist wherever there is Hardware. Virtualization is important for utilization but also ultimately critical for building a truly dynamic IT environment. Virtualization will simply free IT from any specific coupling to HW.

I think I got that one right: Grade A+

 

All Functions exist as Services (SOA) – You may think there is nothing new here but this is where I see a major new change coming. We have always considered applications services as a part of this construct but I also believe that all interaction with data/information will occur at the SOA layer. Applications and users will receive and store their information by interacting with Information Services. These services will provide the protection, archiving, compliance, security, and other capabilities as a service. A single application will no longer “own” data. Information will exist as an independent element that can be managed independently and used by any authorized application. Combined with delivering resources, this creates what I call a Services Oriented Infrastructure.

While not directly SOA, the concept of Web Services is now ubiquitous in the industry. We see more and more data services becoming available every day and the concept of a blob stor is now considered mainstream: Grade A-

 

Composite Applications are built without code – Within the services framework complete applications are simply connected with workflow (BPM) tools just like working with Visio. Composite applications are built by coupling information, security, application, and other services together in a prescribed way.

There clearly has been progress in the direction. Developers are using new high-level languages and a base of open source and web services to rapidly build feature-rich applications. We are getting there but are still a good ways from "drag and drop:" Grade B

 

IT becomes Information-Centric - In the existing IT environments, I believe we moved from being Server-centric to OS-centric to Application-centric. In the next generation, we become more network-centric but fundamentally (for the first time I might note) start building Information Technology actually around the Information. This is powerful. It means that Information is no longer captive to a single application but can be leveraged across any number of applications.

I would argue that we are well on the way to realization of information-centric IT. The need to leverage "big data" and improve our insight via BI and Analytics is at the forefront of today's IT efforts. More and more our applications exist to serve up information in unique ways: Grade B+

 

Virtual Appliances become the preferred delivery model for Application Services - As all interaction and communications between application services and information services will operate at the SOA layer, many of the complex, driver-centric functions that exist within today’s operating systems will simply no longer be utilized. Base operating environments will exist principally to provide a compute environment for applications. Hence, we will start to see more Applications embed base OS and other base capabilities directly with their offerings. This will simplify integration, test, security, delivery and support. We are seeing major examples of this today – for example with Oracle’s recent embedded Linux announcement.

Yup. The VM (virtual machine) approach is totally mainstream. Developers just pick their foundation elements that a best suited to their app and go from there. Consumers can not longer expect or require that a server-based application run on their chosen OS: Grade A

I would also note that the device and client side has gone through a much more rapid transformation than I expected. Client-side app development is moving almost in the opposite direction where developers must have applications that can run across the variety of platforms (android, iOS, etc).

 

Model-Based Management Provides Orchestration or Resources and Services – To pull all of these capabilities together we need management. Traditional framework-centric management is just not going to cut it, however. Today’s management technologies simply can’t handle the virtual, dynamic, and complex environments that will be constructed. This is where model-based management comes in; it will transform how we think about management. Simply, model-based management will provide the orchestration necessary to deliver highly reliable and scaleable systems across these complex environments.

I would say I missed that one. What I now believe needs to evolve in IT management is less focused on model-based capabilities (although they will still be used) and more focused on policy-driven automation:  Grade C

 

While Flat IT never caught on as a term (I wish I could say I was thinking of calling it Cloud Computing) it seems that things have evolved amazingly close to what I expected. You can be the judge, however, as it's all there in the archives....

Episode 102: Ken Olsen

The passing of Ken Olsen caused me to think about how much pioneers like Ken did for the industry, and how much our industry has changed over time.  As we enter yet another generation of computing (Cloud) I find it interesting that we really can’t associate it with any one individual.  There is no doubt, however, that Ken Olsen was clearly the father of the 2^nd generation of computing (minicomputers). What is even more interesting is how much Cloud Computing employs many technologies that evolved from the mainframe and minicomputer eras.

I joined DEC in 1984 and, for the next three years, had almost a surreal experience working for such a wildly successful company. First and foremost, DEC was all about having the best architecture, technology, and products. People may remember that DEC initially didn’t even pay commission to sales people so they wouldn’t be tempted to exaggerate capabilities.  It was a time of simply amazing resources applied to technology and a time when, even as a junior engineer, Ken considered our time so valuable that we had helicopter service between the DEC facilities and the airport to avoid the traffic (I crack up now thinking about it, as my last trip to Hopkinton involved driving a rented Toyota.)  As you can see, I have really progressed in my career!

I was fortunate (for being at the very bottom of the org chart at that time) to actually spend a good amount of time with Ken.  At the time we had made the decision to build some of the first enterprise RAID products and then some of the first array products designed for non-DEC platforms.  These StorageWorks products ultimately came to completely breakaway and lead the mid-range storage market. Even the brand still exists with HP today (although not the market share!).

As I interacted with Ken, I was amazed both with his passion for technology “goodness” and his passion for what we were building with modular storage.  We had built some of the first hot-pluggable storage array systems out of snap-together components and Ken loved the simplicity and modularity so much that it became a driving force across the company. He later founded a company called Modular Solutions to further these efforts.

 DEC, to me, had an almost internal VC culture under Ken where teams would be allowed, and often encouraged to develop competing products. I often felt like some of the toughest competitors were not IBM or HP but other DEC teams. This internal competition was sometimes fun, always challenging, often brutally competitive and sometimes downright dysfunctional. In the end, it’s hard for me to judge whether it actually helped or hurt.

The fact that DEC no longer exists should not diminish the value that DEC and Ken brought to the industry. In fact, the list of companies that have led the market across generational computing and technology shifts is essentially zero. Many have survived but there are always new winners across each successive generation. As I think back to my (few) programming days I still am amazed by what DEC accomplished.  In my opinion, VMS is still the best OS ever created!  

Hat’s off to Ken – the industry we enjoy today exists, in part, to his efforts.

 

Mark...

Episode 101: Big Data Unplugged

Next to “Cloud” one of the hottest new techie areas is Big Data. Of course, being a data storage company at our roots, the term Big Data to a company like EMC is akin to telling me that I get a Big Rib-eye steak for dinner; yum.

The simple fact is that Big Data could transform business and society in very positive and meaningful ways, but the point of this post is to just provide some basic context as to why Big Data should be on virtually everyone’s short list for investigation.

I start with two simple axioms for Big Data:

                More is better

                Extracting value is the trick

While I think we always can get overloaded with information, the opposite I believe is true with data. More is always better.  Before you think this a naïve statement, consider two caveats. First, I assume we can separate valid and invalid data. Second, I make the assumption that there is a way to analyze the data. These two needs, eliminating invalid data and analysis, will become the key ingredient in delivering new value.

So, what can Big Data do for us? The potential is limitless. For example, it has the ultimate potential to help find cures for diseases and revolutionize personalized medicine. It has the potential to change how we deal with poverty, and related Big Things. Big Data can also change our day-to-day lives. Here is a simple example: imagine that when you setup an appointment for lunch 20 miles away, a program automatically calculates the drive time looking at both the predicted weather and the traffic delays for the time of day and day of the week, and then automatically allocates the time block on your calendar.

Much of Big Data will be about combining and correlating data across multiple datasets. Where business today uses BI software to track customer buying trends by some facets (like store location, time of day, etc), the increasing trend will be to look at more and more datasets as potential indicators and sales predictors.

Another new trend in the "data" world is mining and using unstructured data as a data source. Examples of unstructured data are video, files, and even Facebook and Twitter feeds. The challenge here has been how to extract semantic data out and then index it in a useful way. For example, one use has been to capture "sentiment analysis" datasets around a company or product based on looking for the number of positive and negative words found adjacent to the particular term.

This is just the beginning. I expect that video recording in stores to go far beyond being used just for security and businesses will start to build datasets for everything from how long we have to stand in line to how we walk through the store. I am not even going to engage in the "big brother" implications here. This is just the evolution of cultural norms as we now share our daily lives with hundreds of "facebook friends" and carry a GPS transmitter in our pocket or purse. The clear hope is that this data will be used to improve our experience, efficiency, health and quality of life.

So, the base technology is here today. We have the capacity to store, the technology to extract semantic data, the database capability to structure the data, the indexing and map reduce capability to analyze and create UI's to present the results. This opens the door to a whole new set of innovators and innovations that can leverage all of this technology and solve problems that we could not imagine solving a decade ago. I think of it as not just finding a needle in a haystack but finding matching needles across 10,000 stacks of needles.

Very exciting stuff.

Episode 100: Security and privacy - a battle of extremes

I watch in horror as I look at what people post in open forums on Facebook, and I just want to ask them -"do really understand what you just said will exist on the web forever?"

At the other end of spectrum, my financial advisor was just required to install a new "secure email" system to insure that no one (including me) can read anything she sends. After spending 10 extra minutes to trying to decrypt, decode and download every email, I am pulling my hair out with the inefficiency. Thanks to secure EMAIL - I now think the post office can make a comeback. 

So why is it after all these great years of "good enough" email, we are now relegated to a choice of systems that virtually guarantee what you say will become public record with the alternative being new "secure" systems that are so cumbersome and time consuming that I need to FEDEX important stuff? It's not right!

Technology has been a great boon for the world economy but I feel like inability to provide "transparent security" risks is wiping away much of those gains. The risk I see is that, every time we make it harder to do something to accommodate a security need, I believe we actually increase the risk that individuals will bypass the security to regain the efficiency. At some point security additions could potentially increase security risks.

A great example here has been passwords. We all now have so many passwords that folks are forced to either write everything down, or use one password for everything. While companies create rules to have exotic and ever-changing passwords, do they also supply the users with password management software so they don't just write everything down on a scrap of paper?

On the eve of the RSA conference I thought it would be great to present this challenge. We clearly need reasonable security and privacy but we also want to use our iPads for business without having to lose all of the productivity gains. When I get a bank statement, I want it to come up on the screen, not be a link to the bank's home page where I get to spend 10 minutes logging in and navigating to the right spot. When I buy gas at the same station every week, I don't want to spend extra time typing in my zip code.

The true revolution in security, to me, comes when users can have it with little if any visible impact to the ways we would like to work. I would encourage everyone building or evaluating security capabilities to really make user ease-of-use a top priority. The best security is security that users really adopt. 

 

Mark…