The 10 Big Ideas That Are Shaping IT Infrastructure Today

Hi everyone.  I had the pleasure of jumping over to Paris to deliver a short keynote this week. 

When I do this, sometimes I use material I've used before.  Other times, I try and mix it up a bit and try something different. 

Such is the case here.

I wanted to make this presentation all about the big ideas that are going on in our industry today, especially as it applies to IT infrastructure, since it directly and indirectly impacts all other aspects of IT. 

I wanted to make a point to try and stay away from individual technologies (and their associated product incarnations), and focus instead on how technology is being built, deployed and consumed.

See if you agree with my choices ...
 

#1 -- IT Infrastructure Is Maturing

If you're a regular reader of this blog, you'll recognize this thought. 

In many senses, IT infrastructure isn't really all that different than other forms of infrastructure that deliver shared services: communication, power, distribution, manufacturing, etc.

Each of these forms of infrastructure has undergone major structural changes on a periodic basis.  In each case, we think of the infrastructure as a service, rather than a set of technologies.

And, as each form of infrastructure matures through phases, it's often build differently, operated differently and consumed differently.  It's built as a shared service.  It's operated in terms of service delivery.  And it's often convenient to consume for the people who use the service.

So much of IT is infrastructure.  And one could persuasively argue that it's started to go through a similar maturation process.  How we did things a few years ago isn't going to be the same way we do things in a few years.

#2 -- The IT Industry Is Consolidating

The first trend leads to the second trend.  As different forms of infrastructure mature, the inevitable outcome is fewer, bigger and more consolidated players.

And, once again, our world of IT isn't all that different.

You'll hear the term "stack wars" being used to describe the new structure of our industry.  The real action has moved to larger (and more integrated) IT players.  Smaller players will find the game progressively more difficult going forward.

The first and best at this integrated model at this integrated model has always been IBM.  In many respects, IBM's model is a reference point for the industry.  For example, their progressive shift to a services-led model (and somewhat away from enabling technology) could be seen as a reflection of their relative maturity.

Apparently working hard to chase IBM is HP.  Their acquisition of EDS puts them on a relatively equal footing with IBM in the traditional services businesses (integration, consulting, outsourcing, etc.), among other things.  Certainly, they're another stack player to be reckoned with.

In many senses, the VCE Coalition (VMware, Cisco and EMC) aspires to be a third stack, but based on a different set of assumptions than other players: the use of very differentiated technology, for example, or the emphasis on an open ecosystem of services partners, a pronounced focus on cloud delivery models, and so on. 

And, unlike the others, we are distinctly separate and independent companies.

Of course, Oracle acquiring Sun leads us to the inevitable conclusion that there will likely be a fourth viable stack in the marketplace over time.

Just to put your fears to rest, even though these stacks are forming, and consolidation is inevitable, all the technology from these various vendors will likely plug and play together reasonably well.  All of these vendors make products that lots of people use. 

No one's going to overtly lock anyone else out (well, sometimes I wonder about Oracle's stance on VMware, but that's an exception).

But, that being said, if the value moves to tight integration up and down the stack (which is most certainly the case today), the stack players will have a strong advantage in many regards.  There will be strong incentives for larger IT customers to stay in the stack.

#3 -- IT Delivered As A Service

The first two trends lead us to a third one -- enterprise IT groups are starting to think of themselves as internal shared service providers to the organization -- again.

Back when everything was done on mainframes, the only viable model was a shared service model -- I mean, how many applications could justify their own mainframes? 

The secular shift to distributed systems -- minicomputers, UNIX and other servers -- meant that each application could have its own dedicated infrastructure.

And that is what has been largely happening for the last 20 years or so.

Old school IT infrastructure was all about buying dedicated physical equipment to support specific applications that the business wanted. 

New school IT infrastructure is all about creating a catalog of services that the organization can consume -- using a pooled and dynamic set of resources, if possible.  The services are relatively standardized, don't require an enormous upfront investment in time and money to use, can be dialed up and down easily, and so on. 

Just like a phone system, for example.

For those of us who cut our teeth in large mainframe data centers, all of this seems eerily familiar.  Maybe we'll get back to calling IT "information services", or IS.

This "as a service" mindset is happening outside of enterprise IT as well. 

There's currently a massive influx of investment in newer enterprise-class IT services that can be consumed by IT organizations to help them deliver their own services more flexibly and efficiently.  The list of viable external players and services grows in leaps and bounds with every passing month, and shows no sign of slowing down.

Finally, IT organizations that *don't* think of themselves as internal service providers will run the risk of being bypassed by their user communities.  Why?  These external service providers are more than capable of selling directly to end users, and completely cutting IT out of the loop.  Not ideal, but it happens all the time.

The stark message is pretty clear -- start thinking like a service provider, or you'll find yourself competing with them.

#4 -- Fixed vs. Variable Infrastructure Models

So much of IT infrastructure thinking is focused on relatively fixed application workloads: maybe it's SAP, or email, or something similar.  The infrastructure requirements are well-known, the choices are clear, and it's all about delivering ever-higher SLAs at ever-lower prices.  Incremental optimization is the name of the game.

The good news?  The resulting infrastructures for fixed requirements are extremely reliable and efficient.  The bad news?  They're relatively resistant to adapting to new requirements, and the processes that build these environments aren't too nimble, either.  It's like pouring concrete -- very sturdy, but a pain to change.

The new idea?  Think of variable infrastructure to complement a fixed infrastructure approach. 

Infrastructure that's designed to handle a wide range of variable consumption requirements make sense when the future requirements are hard to predict or forecast with any reasonable precision. 

In many cases, getting something up and running quickly matters more than the utmost in precision and efficiency.  And variable infrastructure allows IT to dial up (or dial down!) service levels around performance, availability, protection, etc. as needs change without having to plan and pay for it up-front.

Here's the pitch: most business innovation requires variable infrastructure, rather than fixed infrastructure.  Most new ideas start out as a pilot or prototype, and either succeed or fail.  If they succeed, more resources are dialed in until the requirements are more well known, and then a fixed infrastructure approach might make more sense.

In most of my conversations with enterprise IT leaders, I usually can uncover some large, unmet needs that could be well served by a variable infrastructure approach.  Application development and testing.  Trying out new customer-facing applications.  Technical computing.  Business analysis and decision support.

None of these do well on fixed infrastructure models.  All do well on variable infrastructure models.  And all are about empowering knowledge workers that are supposed to be creating new value for the organization.

The bottom line?  If your business model is depending on significant new innovation to drive the business, you'll most likely need at least some variable IT infrastructure to accelerate that part of your business.  And you'll most likely find the strongest resistance to this idea within the IT organization -- and not the people who use it.

But if you don't want to do this for your business users, that's OK.  There are plenty of service providers who want to talk to them.

#5 -- Pre-integrated IT Infrastructure

Put all of these ideas together, and you end up with the notion of pre-integrated infrastructure that can deliver IT as a service.

Here's the example from the VCE Coalition the V-block. 

A V-block uses a pooled resource model as a default -- everything is shared, everything is virtualized, everything is dynamic.  When you think of delivering IT as a service, it compares favorably in "cost to serve" -- the all-in price to deliver a usable virtual machine to the business.

Part of doing IT better is changing the operational models as well, and here you'll find support for both next-generation management models as well as next-generation security models.  Regarding management and security, I would argue it's not just "as good as" what you're doing today; it has the potential of being a much better way of delivering IT services.

It's consumed differently, too -- by IT organizations.  To speed time to value (and lower support costs), it's packaged, delivered and supported as an integrated unit.  I think of it as a "vending machine" for IT resources.  And you don't get to pick the components in this model -- we do.

And, of course, it supports a variety of multi-tenancy and self-service models, if you'd like.

In one sense, offerings like a v-Block are a logical outcome of everything we've talked about up to this point: IT infrastructure is maturing, the industry is consolidating, integration matters, IT wants to be delivered as a service, there's a need to provide variable consumption models at scale -- and now the notion of a V-Blcok might make a bit more sense.

But not everyone sees change as a good thing.

For one thing, it's very different than what most people are doing today in their IT environments. Depending on your perspective, that can be a very good thing indeed.

Obviously, an integrated platform built on dynamic and fully virtualized resources does well at supporting variable infrastructure models.  But a v-Block also compares favorably when evaluating its capabilities against traditional fixed infrastructure models -- better utilization, better operational model, better security model, and so on. 

And since one platform can do both -- well, IT comes out even farther ahead by using one platform to support both needs: fixed and variable.  That might explain why they've been unusually popular so far.

#6 -- Compatible Service Providers

So many external services (outsourcing, cloud, etc.) have been built around what I would call a "hands off" model.  The service provider delivers the service, signs up for some sort of service level agreement, and that's that.  Don't look behind the curtain, please.

The reality is that most IT organizations need more than just a nice service level agreement and a number to call. 

These groups are held accountable for delivering the service -- regardless of who's providing it -- so they need various degrees of control over the service provider environment if needed.  They usually need the ability to drill down, investigate and change things -- just like they do in their traditional data center environments.

I think the newer generation of service providers have started to figure this out, and are now creating compatible enterprise IT services that use the "tenant in control" model.

In this approach, the service provider's capabilities are exposed in such a way that the enterprise IT user can be in relatively full control if needed.  The tenant can view and inspect their service provider environment in just the same way (or better) than they can do with their traditional assets.

And, when thought of in this way, there's ample opportunity for these "tenant in control" services going forward -- moving information back and forth, composing and pushing application images, monitoring the service delivery by the service provider, resource utilization and costs, monitoring backup and recovery protection, security and compliance monitoring, even automated support for auditing.

Control over important IT functions is needed by most enterprise IT organizations in their data center.  Why wouldn't these same people want the same sorts of things from their service providers?

And that's exactly what we're starting to see.

#7 -- Private Clouds

So, put all the pieces together, what do you have?  Something we call a private cloud model.

What makes it a "cloud" is three things: it's built differently (dynamic pools of virtualized resources), it's operated differently (low-touch models that span traditional disciplines) and it's consumed differently (support for variable consumption models).

What makes it "private" is that it's completely under the control of the enterprise IT organization -- regardless of whether the services are provided internally, externally or any dynamic combination.

It presumes that just about everything has been virtualized: servers, storage, networks, desktops, etc. -- or is in the process of being virtualized.  It provides far more efficient services than traditional physical approaches.  It reacts quickly to new requirements.  It is better managed, and better secured. 

It is -- in a nutshell -- next generation IT.

We believe that a private cloud model is nothing more than the full evolution of virtualization concepts at scale.  And a number of progressive IT organizations are setting out in this very direction.

#8 -- IT Governance

It's pretty clear that there are some powerful forces changing how we think about how IT is built and delivered, but how do we effectively manage change itself?  Change entails risk, change is often resisted -- change is not always seen as a good thing.

This breaks into two distinct agendas -- what happens inside IT, and what happens outside IT.

Within IT, we'll get around to talking about some of the new roles and responsibilities in a minute -- that's a very big topic in itself.

But sometimes the bigger challenge is moving the larger organization forward along with IT's progression.  There are investments to be made.  New ways of doing things.  Concerns about identifying and mitigating risks.  It's not just an IT discussion -- the entire leadership team has a stake in the outcome.

Especially if the goal is to move fast ...

The business world uses the term "governance" to refer to the leadership forum where all of this is hammered out and debated across the various business functions.  Many organizations already have multiple governance functions already in place for things like finance and HR.

And, if you listen carefully, the need for good IT governance is becoming more widely recognized, especially if your company things IT should be a core competency.

Done right, it can dramatically speed up progress: risks are identified and discussed, business leaders are engaged and understand the payoffs, investments are made, and it's not just an IT thing anymore.

At EMC, our IT organization is now moving full-speed ahead to a private cloud model.  You can see some of our progress here.  And, not surprisingly, we found clear advantages in creating a good IT governance model to accelerate this transition.

I thought it'd be useful to share some of the practical outcomes that can come from good IT governance.

Example -- Draft Cloud Governance Model

Here's an early example of our draft governance model for use of cloud technologies within EMC IT, whether they be internal or external services. 

We put the various IT application use cases into buckets, graded them in terms of infrastructure requirements, and decided which ones were suitable for different cloud models.

There are several pages of detail behind this high-level summary, but I think you get the idea -- there's a managed framework for having the discussion that's relatively stable over time.

As internal and external capabilities mature, the specific recommendations will change. 

But the governance framework likely won't change ...

Example -- A "Bring Your Own PC" Model

If you're like most enterprise IT organizations, you're enamored with the idea of provisioning user experiences that follow people around independently of what device they're using.  You don't want to provision specific devices anymore if you can avoid it.

Take a look at this university class.  Look at all the Apple computers -- except for that one student who insists on being different by using a Windows PC.

So many of our businesses are built on the contributions of knowledge workers.  As a knowledge worker, I want the freedom to choose the device (or devices) I want to use to get my job done.  I don't expect my corporate IT people to support the device -- I expect them to provide the experience on my device.

Even if it's an iPad ...

Example -- ROI Model For Virtual Desktops

So many organizations struggle to create a viable ROI for desktop virtualization, and EMC IT was no exception. 

Our first attempt at a pragmatic ROI model didn't work.  But our second attempt created 8 distinct ROI models that could be blended to justify a virtual desktop approach for each of our different user populations.

Taken this way, the case for virtualizing desktops moves far beyond "let's save money', and towards better support of our knowledge worker employee base.

I could give you dozens of examples of the benefits of good IT governance from within our own IT organization.  And I could add many dozens more from other customers I've spoken to about the subject.

#9 -- From IT Governance to Information Governance

In one sense, creating good IT governance is a necessary step for a much more important function -- information governance.

At some point, organizations start thinking about information more like money, and less like a bunch of 1s and 0s.  Information can be valuable stuff.  It can be expensive stuff.  Or dangerous stuff.

If we think of information as an asset -- independently of applications and infrastructure, who manages and governs this important organizational asset? 

Who's responsible for coming up with policies that maximize its value?  Minimize its cost?  Identify and mitigate any risks involved?  And who's in charge of making sure that any new policies are working effectively?

If you said "the business" or "IT", you're only partially correct -- it's all of the above, and more.

Indeed, I can point to at least two dozen progressive companies who have instituted an information governance function.  Some have had it in place for a while, and can attest to the importance -- and the benefits -- for all involved.

Including IT.

#10 -- The Next-Gen IT Career?

We speak so much about next-generation data centers, and next-generation IT -- what about the career angle on all of this?  It's during times of great change that new opportunities and career paths present themselves.

There's a personal angle around all of this that should be important to everyone in the industry.

One way of thinking about IT organizations is in quadrants.

There's a part of IT that architects, builds and runs the application service delivery platforms.

There's a part of IT that deals with individual business users and associated projects.

There's a part of IT that deals with external suppliers, vendors, consultants, providers, etc.

And -- ideally -- there's a part of IT that ties off with overall business strategy and business leaders.

What are the new career opportunities in each of these quadrants?

First, if you're in the architect, build and run group, there's a whole new world of enabling technologies to go learn and understand.  Rather than be deep in any single one, the goal is to understand enough about all of them -- and how they relate.  You need to be able to communicate where it's going, and -- more importantly -- how the technology will get us there.

I'd suggest you think in terms of some of the concepts I discussed here: IT infrastructure that's built differently, operated differently and consumed differently than what you're doing today.  Thinking in terms of a service catalog to your users, and adding variable consumption models to the mix.

Second, if you're dealing with business users and their projects, you need to arm yourself with a new weapon: speed.  Being able to bring variable infrastructure models to the table means that you can enable the business to move faster, and with less risk and cost -- than before.

And you might have to do some serious re-education of your internal clients around what is now becoming possible.

Third, if you're dealing with vendors and suppliers, you need to be asking for more.  Sure, you always want a better price (don't we all?) but you should also be demanding more value from your suppliers.  Better integration, more complete solutions, aligned services, flexible consumption models, and more. 

You're not just buying products any more, you're buying capabilities.

And if you're fortunate enough to be responsible for tying off with the overall business strategy and senior leadership, your opportunity is clear and compelling: you can re-engage with an entirely new set of scenarios about how you organization consumes IT services -- and the need for associated governance functions to speed the transition.

And these are just the highlights -- everywhere I look, I find new career and leadership opportunities for IT professionals everywhere.  Sometimes I wonder, though, just how many people will see the transition and navigate in this new direction.

Final Thoughts

For many of us who have spent our careers in the IT industry, these are exciting times indeed. 

I think our collective challenge will be to not fall back to our traditional strengths in debating the pros and cons of individual technologies, but putting forth new strategic options for the organizations we serve.

Thank you for listening, and I hope you found at least part of this interesting.

Comments

[Omer Ansari]

Great post, as always Chuck. I would add to #7 (Compatible SPs) that we at Cisco are starting to see changes in the way SPs interact with vendors and even other SPs..

To start with, SPs, mostly in the emerging markets, are starting to drive vendors to a risk/reward based model.

I recall the very public Bharti Airtel offshoring model earlier in the decade where they started exploring such models with Ericsson/IBM for their mobile business.
Since then, other SPs in closeby geographies have started to do the same, and the framework has bled itself into Data Centers and cloud computing models.

Apart from payment deferment, which helps SPs acquire customers and pay back vendors when the customer revenues starts flowing, other areas where smaller tier SPs are starting to show interest is re-branding, and white-labeling a subset of bigger SPs offerings.

In fact I believe this "white labeling" use case (instead of the multi-SP disaster recovery use-case) will be the catalyst to driving development and real implementations of inter-cloud workloads. This is because it brings different SPs directly together as business entities, as opposed to being forced together by enterprises (as in the case of the hosting at different SPs for DR purposes).

While US-based cloud providers have the "white labeling" revenue stream squarely part of their business case, the risk/reward sharing model appears to be a non-US phenomena at this time. But much like how the US has learnt and embraced innovation from outside (Toyota's lean manufacturing comes to mind), I'm certain its only a matter of time that the risk/reward based models will become a mainstay here as well.

** these are solely my thoughts; I do not represent Cisco **