A New Chapter Begins In The IT Stack Wars

I was at a customer dinner event last night, and predictably I got asked what I thought about yesterday's NetApp announcement by several EMC people.

My answer?  I didn't see it as big news, simply because Oracle's formal acquisition of Sun is far more intriguing and challenging in so many ways.   That's the real news, from my perspective.

Now, working for EMC, I'd always like to have fewer competitors than more competitors -- but many of us recognize we've got the potential for an entirely new competitor for us to contemplate and consider.

Much will be written and said about this combination going forward -- so I guess it's time for me to go "on the record" as to how I see this playing out.

The IT Industry Is Changing

Our IT business is rapidly maturing. 

There are fewer independent players needed, and -- as a result -- all focus is on the new end-to-end stacks that are forming: hardware, software, services.  Scary stuff for vendors and customers like.

IBM has always been the poster child of this model.  HP's acquisition of EDS put them into the same league.  Last November's announcement of the Virtual Computing Environment (VCE) is driven by some of the same forces.  And now we have the Oracle/Sun combination.

Four stacks.  Four strategies.  Four different combinations of strengths and weaknesses.  Two using traditional models: IBM and HP.  Two using relatively newer models: VCE and Oracle/Sun.

In one of the news clips I scanned, Larry Ellison nailed it, saying -- in effect "the cost isn't in the components themselves, it's costs associated with gluing the components together, and running them effectively".  I couldn't have said it better myself.

Indeed that theme is one of the major drivers behind the VCE Coalition -- vBlocks as pre-integrated virtualized infrastructure at scale.  Aligned professional services w/partners to migrate and deploy effectively.  New joint ventures like Acadia to provide new operational models and consumption models behind our partners.  All driven by VMware's ability to virtualize physical computing.  It's obvious that many of us in the industry have the same things on our mind.

So, viewed through this lens, how do the various stacks line up in -- say -- a few years down the road?  Because large-scale IT buyers are wondering the exact same thing, or will be soon.

Hardware

Simply put, Intel and IBM have a renewed competitor for CPU architectures.  It's an expensive game to play, though, almost like yacht racing :-) 

The SPARC family will need a massive infusion of R+D as its roadmap has been generally neglected for a while; not to mention the need for next-gen fabs to build the parts.  IBM shows no sign of backing down from Power (despite limited adoption), so -- in some sense -- this may turn out to be a SPARC vs. Power battle, with the lion's share continuing to move to Intel architectures.  If you haven't noticed, Itanium is generally regarded as not being viable for much longer.

However, the SPARC business doesn't have to be profitable on its own, when you contemplate all of the new Oracle potential business model.  Plenty of room for "margin expansion" in the software side of Oracle's business to subsidize this effort.

Server architectures?  Sure, Sun has that DNA in spades.  However, one could argue that when the server and network world converge, Cisco is in a very strong position by comparison.  Yes, we all remember Scott McNeally's famous saying "the network is the computer", but only Cisco's UCS has made good on that vision so far.

Storage?  More of the same.  The storage part of EMC's business has always competed favorably with server/storage vendors (including Sun!) and historically done quite well.  Compute and storage are fundamentally different disciplines, as many server vendors have discovered.  Sun's current storage portfolio has some interesting stuff (ZFS, for example), but -- without a massive infusion of investment (organic or acquisition), storage will continue to be nothing more that a side business for Oracle/Sun.

Networking?  Not much to look at that I can see -- although there are plenty of smaller and potentially disruptive network vendors who'd give their eye teeth to get inside the Oracle / Sun combination.  Also, many of these are attractive M+A targets as well.

Bottom line -- no problems in the Oracle / Sun portfolio that several billion dollars -- spent over several years -- couldn't fix.  Of all the hardware disciplines, I'd argue that storage will end up being the most difficult for them to get right.  Certainly, it's been difficult for every server vendor so far ...

The Magical Hypervisor

VMware's success in separating logical from physical seriously concerned Microsoft when they first realized the implications.  Although they've come to terms with it; right now, it's apparently seriously concerning to Oracle as well.  The reason is obviously clear: lose control of the hypervisor, lose control of the stack, lose many of your arguments (and associated profit margins) for your integrated approach being superior.

Personally, I think this particular game is mostly over -- VMware has largely won, and they're busily cementing their lead and widening their strategic gap.  And just about every large enterprise tell me that they're deploying VMware just as aggressively as they can.  If you want to watch one essential strategic conflict play out over the next year or so, focus on this one. 

After considering all the options, Microsoft has wisely made the decision not to go head-to-head with VMware ; what will Oracle decide to do longer term?

Middleware and Databases

This is ostensibly Oracle's greatest strength -- and potentially their greatest weakness. 

On one hand, they've got gaziliions of customers who are heavily invested in Oracle's database and related technologies.  It also is perhaps the easiest part of their business to potentially commoditize due to a number of strong industry trends: aggressive competitors, open source, SaaS models, new Web-oriented data models that aren't build on relational databases, etc. etc. 

If Oracle raises prices in this business to invest in other parts (e.g. hardware and related infrastructure), there's going to be even more incentive to start looking at alternatives.

Indeed, over the past year or so, the number of customer conversations that were around potential Oracle alternatives has picked up dramatically.  Watch this space carefully, because if enough people start going elsewhere, the end-to-end model that Oracle is so intent on pursuing is in serious trouble.

Management and Orchestration Software

Neither Oracle nor Sun were especially renowned for their ability to discover and orchestrate IT resources, especially in enterprise IT environments -- never mind the newer flavor of dynamic and virtualized IT. 

You may be dismissive of this as not a big deal, but I'd argue that much of IT optimization has to do with newer operational models, and -- without a robust toolset and frameworks to implement those new operational models -- much of the big savings from Oracle's proposed integrated approach will remain largely theoretical.

It took EMC many years and over $1B of R+D and M&A to create the Ionix stack, by comparison.   Some of that integrated goodness is clearly visible, for example, with EMC Ionix UIM (unified infrastructure manager) which is basically an integrated element manager for a vBlock.  More investments that Oracle will sorely be needed to become a legit player in this strategic IT infrastructure category.

Security, Security, Security

The same thing can be said about Oracle and Sun's combined footprint in security -- so much of IT today demands a comprehensive approach to this important topic. 

Sure, there a lots of bits and pieces in their combined portfolios, but what's missing are many of the newer security technologies as well as the integrative frameworks that are becoming oh-so-important.   One of the attractive parts of the VCE approach is RSA's increasing contributions in this area.

Again, more serious investments that Oracle will need to make -- over time -- to have a seat at the table here.

Packaged Application Software

Oracle's applications are an exceptionally strong asset -- since so much of IT infrastructure purchase is largely driven by application requirements.  However, Oracle's historical platform-neutral stance has conditioned buyers to expect and demand infrastructure choices in this category: HP, IBM, Dell -- and, more recently -- Cisco. 

Not many customers respond favorably to having choices taken away.  Trying to strong-arm customers into having application decision drive infrastructure decision is a notoriously tough game -- only Teradata and Netezza seem to be making a decent-sized go of it, and I believe it's increasingly tough to defend their appliance models going forward.

Application Developers

This is where Scott McNealy and Steve Jobs are different.  Scott focused on having insanely great tools for application developers.  Steve focused on having insanely great business models for application developers -- the iTunes App Store.  Sharing the wealth with others is not exactly a core part of Oracle's DNA. 

I'm guessing that -- despite arguably great developer assets -- the opportunity will go to stack vendors who can create great business models for application developers -- and that's not likely to be Oracle anytime soon.

Partner Ecosystem

Here, I'm not really talking about distribution (although that's important) -- I'm talking about value add from integrators, solution providers, system integrators, consultants, outsourcers, service providers -- all the different ways that value-add is created between manufacturer of technology and the eventual consumer of technology. 

Way back when, Sun did a great job here -- but their ecosystem has been seriously neglected for many, many years, and people have moved on to other opportunities.  Much of the same could be said about Oracle.

To fix this, you need two things -- a sustained investment in the partner ecosystem, as well as a willingness to share the wealth with your partners.  All the first requires is a big checkbook ; the second requires a cultural shift, and that's somewhat harder :-)

Putting It All Together

Taking a dismissive view of Oracle and Sun together is understandable, but dangerous.  Sure, there are strengths and weaknesses in any combination of vendors and technologies, but we now have a new player at the big table who's got more than a few interesting cards to put in play.

There's obvious multiple needs for multi-billion-dollar sustained investments.  Even given Oracle's prodigious cash flow, we're talking some serious numbers -- which begs the question of which investments will get made when, and which ones won't be made at all. 

As the famous quote goes "a few billion here, a few billion there -- pretty soon, you're talking real money!".

Even more interesting is the prospect that Oracle may use their current positioning with customers to raise prices to fund this agenda, putting all sorts of interesting reverse dynamics into play.

One thing is for sure: we'll all be watching how Oracle plays the game going forward, and how we should potentially adjust our collective game as well.

Who said IT was boring?  Game on ...

Comments

[Dan Baskette]

Chuck, I think the "network is the computer" vision was realized by everyone. It had much more to do the SaaS than it did with physical hardware design. You need email....you don't need an email server or client...you just need a connection to the network. The actual plumbing hiding behind the network was the truth behind the slogan. Fast forward to current time and we see that model being applied inside the data center. Allow connections to services, but hide the plumbing. The real advantage to that model inside the enterprise is the ability to consolidate "peak-sized" silos into a share virtual infrastructure and Oracle's new strategy would seem to be in direct contrast to that. If I buy their stack and I am using 40% of it's capacity....can I use the other 60% to deploy my new SQL Server App or email? No. They are seemingly contributing to the overall cost problem and hiding behind the integration cost savings.

[Phobos11]

A few comments.

The comment about DB and middleware being Oracle's biggest strength and weakness applies equally to VMware. More so, since virtualization is now commodity already, and both Sun/Oracle and Microsoft offer their own virtualization solutions in their stacks.

Both Sun/Oracle/VirtualIron and Microsoft can offer a complete Virtualization/OS/DB/Middleware/Applications stack right from under the same roof, something VMware can not yet offer, and while Zimbra was a good first step, there's much road to be walked before they can compare to the big players. They "should" be in talks with SAP people by now if they want to play on the same league as the rest. They also need DBs, much more middleware, an OS... Not that it can not be done, but they need to do it first.

Meanwhile, both Oracle and Microsoft are quickly getting to a virtualization solution feature comparable with vSphere, as recognized by IDC.

While many offer DB and Middleware, those products don't really compare to Oracle's offerings. Oracle is for medium and high end, high performance computing. MS SQL and DB2 are for small to medium businesses. That's why most backends run Oracle, independently of who was the hardware provider (be it HP, IBM, Dell or other).

Sun/Oracle does lack on networking in comparison, but as you state, that's not a show stopper and can be fixed quickly.

Also, as announced not long ago, HP and MS will continue to integrate their products, so the actual stacks to be followed now seem to be:
-EMC/VMware/Cisco
-Oracle/Sun
-IBM
-HP/Microsoft

Overall I find your analysis interesting, but understandably biased.

Truly, the future will be interesting. Competition will be great. Each stack has it's strengths from it's origins and has been grown over time to become the monsters they are now. I'll repeat your words, Game on!

[Massimo Re Ferre']

Chuck,

interesting times ahead indeed.

Your analysis seems to be pretty accurate. I was wondering though.... why is everybody today so interested in monitoring this trend of end-to-end stacks? While I am sure there is value in a fully integrated, single vendor (or coalition) stack ..... are we sure that's what customers are looking for? I am wondering that because, when I hear these discussions, I picture an AS/400 in my mind (and by the way do we know anything that is more end-to-end and fully integrated than that?)... yet it doesn't seem to me that that platform is in a good shape at the moment (from a business perspective I mean).

While the end-to-end theory is sound it appears to me customers are still geared towards selecting the best piece/vendor for every layer (based on their own needs).

Definitely interesting times ahead.

Massimo.

[Chuck Hollis]

Your example of the AS/400 (now known as the iSeries) is very insightful.

On one hand, it's hard to find a more integrated stack in mainstream IT. On the other hand, it's pretty obvious that major components of modern IT capabilities aren't available in that world.

You're right, though. Interesting times ahead ...

-- Chuck