It is quite possible that we are experiencing a “back to the future” moment in how IT solutions are delivered and purchased. For a bit of historical context, let’s go back to the IT industry of 1980. There was IBM and the “BUNCH”. The BUNCH were Burroughs, Unisys, NCR, Control Data and Honeywell. And there were the upstart minicomputer vendors like DEC, Prime, Wang, HP and Data General. All of these vendors used a similar business model. All of these vendors sold “systems” that at the minimum were comprised of storage, storage connectivity (anyone remember IBM “Bus and Tag”), the computer, and the systems software. The systems software was almost always unique to the vendor – each one had their own proprietary operating system(s). These systems were so closed and so proprietary that in many cases the applications software was unique to the system as well. In those days when you bought “an accounting system” you got everything from the applications software to the spindle on the hard disk from one vendor.
Now we all know what happened. With the introduction of the IBM PC, IBM set the stage for the disaggregation of this delivery and business model. The CPU came from Intel, the memory from any one of a number of vendors, disk drives could be purchased from anyone and plugged in, peripherals could be added into expansion slots, and most importantly the systems software (at the time DOS) came from Microsoft which in turned spurred the creation of a new applications software industry.
This ushered in the golden era of the PC, with vendors like Compaq (now part of HP), Gateway, and Dell riding the innovation waves of their component suppliers and themselves innovating to provide rapid improvements in and design to their customers. Apple was seemingly the lone holdout in this era, insisting upon the design of their own chips and their own OS – and thereby retaining control of the entire “system”.
This “horizontal” model has served our industry very well. It has allowed for innovation at each layer of the “stack” in a manner more or less independent of the other layers of the stack. This disaggregated model of innovation has let storage, storage networking, switches, local area networking, CPU power, memory density, and systems software functionality all evolve in a loosely coupled manner – with each making their own leaps in price/performance independently of each other. It has also created some remarkably successful companies who each focus upon their layer of the stack. Examples include EMC and NetApp in storage, Cisco and Blade Networks in networking, Microsoft and Red Hat in operating systems, Intel in CPU’s, IBM, HP and Dell in servers, Brocade in SAN’s and finally VMware as the leading vendor for a new layer in the stack focusing upon virtualization and cloud computing.
Now for the back to the future part. There are two very recent vendor initiatives that hearken back to the days of single vendor tightly integrated systems. The first is Oracle’s pending acquisition of Sun. While Sun certainly has some very significant software assets (like Java), Sun is principally a hardware company with a substantial revenue stream from servers and storage. Once the Sun acquisition completes (it may face some anti-trust challenges in the European Community), Oracle will be in a position to deliver (by way of example) a financial system based upon Oracle Financials as the application, BEA Weblogic as the applications server, Oracle JVM (based upon Xen) as the virtualization layer, Linux or Solaris as the OS, with Sun server and storage hardware. This will be a modern example of a fully integrated systems with “one throat to choke” in the eyes of the customer.
The second example which has potentially far more impact upon VMware is the recently announced EMC/Cisco Vblocks. The first level of integration in Vblocks is really at the hardware level with FCoE serving as the glue to integrate storage networking and local area networking. However, the next level of integration is far more profound and potentially impact upon VMware. This next level focuses upon the management software used to provision and manage a Vblock, which it seems is to come from an EMC subsidiary Ionix.
The question of how virtualized systems are to be managed reveals a rather stark difference in opinion and strategy between VMware and its parent company EMC. This will be the subject of its own series of posts, but the short version of the story is this. VMware understands that the hypervisor is going to get commoditized by Microsoft, Xen and Red Hat. VMware intends to provide its value at the management layer by providing a new horizontal layer of management tools for its hypervisor. VMware Lab Manager, Lifecyle Manager, and the forthcoming ConfigControl product are the foundations of this new VMware management stack – one that is intended to reach all of the way up to include a service catalog. EMC Ionix on the other hand views virtualization management as something that has to be done across both physical and virtual systems (a horizontal strategy) and something that has to be done in a manner that is tightly integrated with the EMC/Cisco Vblock hardware (a vertically integrated strategy).
How this plays out will have a huge impact upon whether or not VMware succeeds in creating a new layer in the stack for itself that is both significant and durable in the enterprise IT realm. If the pendulum swings all of the way back to tightly integrated vertically delivered systems, then VMware could be left with nothing but an OEM hypervisor business to the major systems vendors. EMC’s role in moving the industry in this direction calls into question whether or not it is in VMware’s interest all all to be owned by EMC. On the other hand, if the Oracle and Block models are exceptions for the a small segment of customers who wish to purchase things in this way, and the larger IT industry remains horizontally integrated, then this increases VMware’s chances of establishing itself as a significant and durable new layer in the stack. In short, VMware probably views Vblock as “just another piece of compatible hardware”, whereas EMC and possibly Cisco probably view Vblock as the start of a pendulum shift back towards more integrated and monolithic systems. Which view prevails will have a great deal of influence upon the ultimate role that VMware will play in data center and cloud architectures for the next 10 to 20 years.