Now that the dust is starting to settle after Dell’s blockbuster $67 billion deal to acquire storage behemoth EMC and, by extension, its controlling interest in server virtualization juggernaut VMware, it is time to start thinking about what this means for the datacenters of the world. This, more than any other factor, will determine the success or failure of such a combination.
It is easy to say that this is a merger of two legacy IT companies that are doomed, and many have said that in the weeks leading up to the rumored deal, and many will be saying it until the deal gets done sometime around the middle of next year. But the situation is more complex than that, and the outcome is far from certain.
The many hundreds of thousands of enterprise customers that Dell, EMC, and VMware collectively have will move more cautiously than a hyperscale datacenter operator or an HPC center. People sometimes forget that, and their caution is a bit like money in the bank. Hyperscalers and HPC centers can radically alter their infrastructure every couple of years because they control their own applications – which are relatively small in number even if they are themselves complex and large. Enterprises have much lower scale in terms of compute, storage, and networking in their environments, and they often have orders of magnitude more applications in their stack, and this complexity makes implementing radical change at speed simply impossible. Unless the whole company gets behind it and goes all-in to the cloud, as happens.
Buying Time To Build A Possible Future
What Dell has actually bought for that $67 billion is a vast potential base of customers for future hardware and software, and it has also bought some time to create and deliver the products that it hopes its expanded salesforce and partner channel can push and help it make it that money back over the next decade or so – and then some.
We can argue about this being too much money for such a gamble or investment (depending on your viewpoint), but there is no question that both Dell and EMC (and including VMware) had to act now.
For one thing, money is still essentially free until interest rates rise and for another, activist investor Paul Singer, who runs hedge fund Elliott Management, has been pressuring Joe Tucci, chairman at EMC, to do something to raise the value of the storage maker and its federated companies. The presumption that Dell was undervalued by Wall Street was exactly the same issue that caused Michael Dell, its founder, to go private two years ago in a $25 billion deal, backed by venture capitalist Silver Lake and a bunch of bankers, after a prolonged fight with activist investor Carl Icahn. And by the way, in the wake of the dot-com bust in the fall of 2001, a similar drive to find synergies and capture a larger share of the enterprise IT budget is what drove Hewlett-Packard to spend $25 billion on Compaq, itself the product of the acquisitions of Tandem and Digital Equipment a few years earlier. (HP closed its deal a year later, after epic fighting between CEO Carly Fiorina and the estates of Bill Hewlett and David Packard.)
Both Dell and EMC have been acquisitive as they tried to grow their enterprise IT businesses over the past several years, and one can argue that they have both been as successful as most companies when it comes to such deals. It is safe to say that acquisitions are rarely as smooth as anyone expects, and the HP-Compaq and HP-EDS acquisitions are but two examples of this. There are many others. The truth of the matter – and this is hard for founders and CEOs of companies that have been successful in the past to see – is that companies don’t get second chances very often in business, and this is Dell’s real problem. HP, Dell, IBM, and others that had substantial PC businesses missed the smartphone and tablet waves, which have made Apple fabulously wealthy. IBM, Sun Microsystems, and HP were talking about virtualization and utility computing back in the late 1990s, but it was online bookseller Amazon that arguably got the cloud computing ball rolling in 2006 with all of the kind of features customers wanted at a price they liked with ease of use. It was no surprise at all that AWS became the default platform for cheapskate startups just as the Great Recession hit two years later and it had fleshed out its storage and management services.
Recovering from these missed transitions is a difficult task. It may be impossible. IBM transformed itself into a services company in the wake of its near-death experience in the early 1990s, and has subsequently built and bought a software business that has enabled it to get through a decade and a half before it needs to do it again. This initial transition came after massive layoffs and gut-wrenching charges, but outside CEO and Louis Gerstner was able to buy time from Wall Street to engineer a turnaround. IBM has been selling off underperforming assets in recent years under CEO Ginni Rometty, and prior CEO Sam Palmisano did the same. But at some point, acquisitions and divestures do not add up to a strategy. They do not tell a coherent story. And IT shops of all kinds and sizes want a story, one that they fit into and thrive in.
What Is Dell’s Platform To Be?
If Dell needs to do anything, it desperately needs to create a platform and to stop being all things to all companies. (You expected us to say that.) Building a platform means a whole lot more than having servers, storage, switches, and software to manage it all. It means making decisions for customers and making it easier for them to create and run applications on that platform. This was the genius of the original System/360 mainframe from IBM back in 1964, and it was similarly the genius of the Digital VAX and the IBM System/36, System/38, and AS/400 in the midrange in the 1970s and 1980s. Distributed systems took off in the 1990s with Unix and then Windows and Linux operating systems, and it is safe to say that the diversity and complexity in the datacenter went up even as the cost of computing went down.
If Amazon Web Services has taught the IT market anything, it is that companies want a platform – and Google and Microsoft are building platforms. Google has credibility because of the ubiquity of its search engine and advertising and the next platforms behind them, and Microsoft is transforming itself from a vendor of PC and server software into a provider of software that runs on the cloud, on premises, or in a hybrid setup.
While change often comes from the bottom in the datacenter – minis usurped mainframes, Unix pushed out minis, Windows servers pushed out NetWare and then Unix systems, and Linux servers pushed out whole bunch of things including all of the above – in this case, the change is coming from all angles. SMBs will want to get out of having IT operations and will move to the cloud. Technologies developed by hyperscalers will seep down into enterprises, as they have been doing for the past decade, and many of their techniques will bleed over into HPC. And more than a few large enterprises are going to throw up their hands and say they have had enough with buying and managing infrastructure and just move to a cloud, while many others will be hybrid for many years – perhaps as long as a decade, perhaps forever.
So the big question is what does a Dell-EMC-VMware deal mean in this kind of world?
A few years back, Dell was aggressively building its own cloud, but then backed off. VMware has its own cloud, built out of frustration that service providers were deploying open source technologies on clouds more than its own wares. What will become of vCloud Air? Will Dell be content to get its share of Microsoft Azure server sales (we presume about half, split with HP with a little for Quanta), or will it build a cloud of its own in earnest? It looks like Dell will be trying to sell private clouds to large enterprises, but traditional IT gear is not a growth business even if it is a large one.
EMC does not have a switch business, although it has partnerships with Cisco Systems through the former VCE partnership that it totally controls. EMC had a fledgling server business for some of its appliances, and VMware obviously has partnerships with all of the server OEMs. The PowerEdge server will presumably be the foundation of Dell’s compute – which may not be good news for Cisco, which has EMC peddling lots of its UCS iron. Dell’s various switches will presumably be its network gear – also not necessarily good for Cisco. Dell will have several different flavors of hyperconverged server-storage hybrids – including EMC’s ScaleIO, VMware’s Virtual SAN, and Nutanix Xtreme Computing Platform running in its hyperscale-inspired PowerEdge XC servers, and substantial overlap between the Compellent and EqualLogic arrays it bought years back when it had EMC envy and the VMAX, VNX, and XtremeIO arrays from EMC. (Where do the unannounced DSSD products fit?) Dell has been a big proponent of OpenStack, but will try to balance that out with VMware’s portfolio, which also includes an OpenStack-vSphere hybrid.
This is just the tip of the iceberg. An exhaustive analysis of the product lines of the two companies needs to be done and many substantial overlaps need to be dealt with in some fashion. This will be tricky business indeed, and is certain to rankle plenty of customers, employees, and partners. (Remember how that HP-Compaq product line rationalization went?)
But even thinking this way misses the point, doesn’t it? What companies seem to want is a unified compute platform that can support any virtualization layer (including containers) and retrieve data in block, file, or object format with low and consistent latency for links between servers and out to storage and back. They will want automation for software deployment and for the running of these applications, and they will probably not want to pay a VMware premium for it. (Then again, enterprises have been saying that for almost a decade now about VMware, which is bringing in over $6 billion a year in sales.) But the future is Puppet and Chef and Mesos and Kubernetes and Docker and OpenStack and Ceph and things that look and feel like that – including next generation technologies that VMware itself is building. (We will be talking about these in a future story.) Interestingly, Pivotal, which mashes up Hadoop and parallel data warehouses and the Cloud Foundry platform cloud stack, is the closest thing to what we are talking about that EMC has – and it is a miniscule part of the EMC Federation and will be an even tinier piece of Dell.
The good thing for Dell is that it will have a much broader and deeper product line to use as a starting point for creating its platform, and at approximately $80 billion in annual sales, Dell will have a business that is that much more removed from the PC business that is its heritage and that is not going to be the key to its future. Dell will also have cash. The EMC business throws off $6 billion in cash a year, VMware throws off another $2 billion a year, which would imply the payback for the EMC acquisition and the controlling stake in VMware is about eight years. But such a simplistic analysis presumes conditions remain the same in the datacenter, and one thing we know for sure is this: That ain’t gonna happen.
It is hard to imagine a future where Dell builds cloudy infrastructure only for enterprises but does not have its own cloud, and it is also just as hard to imagine that Dell will build its own cloud to compete with and usurp AWS. With $67 billion, Dell might have been able to do the latter by itself, but it might have needed to spend $10 billion of that to build a time machine to get started a decade ago. Or, maybe it should have spent $67 billion buying a big stake in Amazon. (A controlling stake of Amazon would be twice that, and an acquisition would require at least four times that. AWS is probably a large part of that market cap, by the way, since Wall Street knows it is where the company’s profits are coming from.)
So what is the future Dell platform going to be? That is the question Dell needs to answer, and saying “future ready technologies” is not an answer.