skip to main |
skip to sidebar
Both Sun Microsystems Chairman and co-founder Scott McNealy and Oracle CEO and founder Larry Ellison took the stage to deliver keynotes that kicked off Oracle OpenWorld last night.
In keeping with the industry's fascination with stacks, McNealy made the case for why Oracle is keen on Sun's technology. According to McNealy in his keynote address,
"If you think about the Sun technology that we're bringing to the party here, it's the data center. It's the servers, the storage, the networking, the infrastructure software, all the pieces, all of the executable environment within the cloud, the data center, the distributed computing environment, whatever else you want to say, and then you bring in the database, and the applications and ERP and middleware capabilities and developer tool capabilities of Oracle, and you have a very nice data center. A very robust, very scalable... enterprise data center."
Like the other big players in the enterprise IT market, IBM, HP, and Cisco (via its VCE alliance), Sun-Oracle recognizes the power of owning all the pieces of the end-to-end stack. There are clear advantages in a company being able to offer its customers fully integrated cross-domain bundles and be a one-stop shop to the enterprise data center. One of the biggest advantages is having a sticky reason for customers to want and value you that's not based on price.
Clearly, the major vendors are seeing a shift among enterprise customers toward a focus on higher level application value and services rather than a focus on specific technologies and products at the bits and bytes level. This shift is partly due to the economic downturn and the pressure on IT to cut costs and add tangible value to the business.
When it comes to the single integrated stack concept, analysts say they've seen this movie before. Back in the dotcom era, a lot of industry pundits talked about the "God box" in the networking space. It was basically a big, intelligent network switch that functioned as a router, switch, PBX, VPN, and firewall integrated within a single device. In the end, no customers wanted to go there. For the networking vendors, it was sufficient to articulate the vision as part of their corporate strategy, and have the majority of the individual components within their product portfolio.
Time will tell whether the integrated stack concept is a remake of this movie. For now, we're watching industry consolidation in action and having a fascination with stacks.
Shares of Brocade Communications jumped nearly 19% today due to a report in the Wall Street Journal the company has put itself up for sale. Based on today's closing price, Brocade has a market capitalization of $3.8 Billion, which is a relatively low multiple for a company that holds a strong market position in the data center LAN and SAN switching segments to ride the next wave in the transition to a converged network.
Not surprisingly, the WSJ report included speculation on the potential buyers, siting HP and Oracle as the most likely suitors. IMHO, Oracle is a very unlikely suitor given it's preoccupation with the Sun Microsystems acquisition, and the unlikelihood Larry Ellison wants to own a LAN and SAN hardware business.
HP, however, could gain a lot from the acquisition of Brocade's business, particularly in light of its heightened competition and bitter rivalry with Cisco Systems since the networking behemoth entered the market for blade servers this Spring. Let's take a closer look at the various dynamics at work that make an HP-Brocade marriage a good bet:
- With the trend toward IT industry consolidation heating up, many of the large, enterprise IT vendors are finding it cheaper and less risky to buy and control their own destiny than build R&D and/or forge partnerships with companies that might one day be their fiercest competitor... Precisely what happened to HP when it once enjoyed being a 10% customer for Cisco as a reseller of its switches and routers, only to later become its biggest target in the server market.
- There is a large push towards a converged IT infrastructure to deliver more agility and less complexity to IT operations. The converged network (aka Unified Fabric) is a big piece of this vision, bringing together IP, Ethernet, Fibre Channel, CNAs and switch ports. When it comes to networking, Cisco is in a position of strength, both in market share and driving the standards bodies. The only other networking and storage interconnect company with great technology and the FC installed base to compete with Cisco is Brocade.
- Although HP has recently beefed up its networking business with its own ProCurve line of Ethernet switches, it's rarely deployed in the core of the network, and to date, HP has yet to articulate a comprehensive roadmap or a compelling story around FCoE or DCB to its enterprise customers who are keenly interested in hearing about its converged network vision.
- HP ProCurve could be repositioned for the access and aggregation layers of the network while Brocade's Foundry Ethernet/IP switches could be positioned in the core along with its FC Directors and FCoE convergence products so there would be very little product overlap and the merged technologies would be a genuine challenger to Cisco.
One of the potential areas of concern would be around the impact to Brocade's OEM business should HP acquire Brocade. Reportedly, IBM is a 15%-20% revenue customer for Brocade and EMC is a 10% customer. Worst case scenario, it could mean a 25 -30% hit to revenue if the IBM and EMC OEM business went away entirely. Perhaps EMC could funnel all its FC switching business to Cisco in retaliation, but that would be unlikely given Brocade's FC installed base. As for IBM, they have multiple reseller agreements, including Brocade, Cisco, and Juniper, and they are currently supporting Juniper's development of an FCoE switch, so a change in control with respect to Brocade probably wouldn't make much of a difference.
In summary, I think HP and Brocade would be a terrific match. It would be great for the two companies as well as the industry, and it would certainly give Cisco a force to be reconned with in its quest for domination of the data center.
For two decades, horizontal business models in the high tech industry reigned supreme, as companies like Intel, Microsoft, EMC, HP, Cisco, and Oracle came to dominate within their horizontal layer at the expense of DEC, Wang, Apollo, Data General, and others that clung to vertical business models despite fundamental shifts in the industry.
Fast forward to today, coming off a near-death experience in the financial markets, where we live in an environment of increasing consolidation, reduced competition, and greater uncertainty. An environment ideal for vertical reintegration by select companies who can utilize the changing dynamics to potentially build winning vertically integrated strategies to own the end-to-end stack. Here are a few examples of vertical reintegration in action:
- Oracle buying Sun to deliver hardware & software combinations to become the backbone of most enterprises worldwide
- Cisco getting into the server market with UCS - and with a $33B war chest, they can buy their way into storage and IT services if they choose
- EMC getting into server virtualization (aka VMware), systems provisioning and management in the data center
- Dell buying Perot Systems to get into the higher margin, more strategic enterprise IT services space
If you're skeptical these examples equate to a 'back to the future' shift towards vertical business models, consider the following dynamics:
- Co-opetition is out; Bare-knuckle competition is in. Larry Ellison has declared war on IBM to become the dominant player in the software and systems business from application to disk. Cisco served notice to HP and IBM concerning the server business, and is willing to sacrifice their networking OEM business for world domination in the converged infrastructure business.
- Vertical is relatively cheap. Despite a much improved stock market, company valuations are still relatively cheap. Combine cheap valuations with open source, fabless semiconductor businesses, commodity hardware, and SaaS, and it's not as expensive to buy your way up the stack as it used to be.
- Businesses buy solutions, not products. Products don't solve business problems, solutions do. A lot of companies make good point products, but the companies that can package them up so customers don't have to allocate their own IT resources to put them together, will have a seat at the table, be strategically positioned, and win the sale.
- Eroding margins. This is the nature of a maturing industry. But if you own the end-to-end stack, are the backbone of most enterprise infrastructures, and are sticky and strategic to IT operations, you're in a better position to control margins.
Whether this vertical reintegration trend is good or bad for the industry and customers is a topic for another blog. It's way too early to judge, anyway. But it's a good bet there will be more M&A, further shifting battle lines, and several bumps in the road that will prevent a quick return to the future in this unfolding plot.
To make up for a summer of silence, Oracle and Sun are now taking the gloves off as Larry Ellison gave tangible proof on September 15th of his intention to be a formidable player in the hardware infrastructure business, launching his first joint product with Sun Microsystems called Exadata V2.
Effectively dumping HP, after launching Exadata V1 with Mark Hurd at Oracle OpenWorld 2008 nearly a year ago, Ellison has quelled any lingering speculation he might spin off Sun's hardware business to HP. Although HP was not directly mentioned during the Exadata V2 launch, it was quintessential Larry Ellison on center stage, calling out a few of his favorite rivals, IBM, Teradata, Netezza, HDS, and the unnamed in-memory OLTP database start-ups that venture capitalists, according to Ellison, "may be sorry they invested in."
Exadata V2, benefiting from the inherent performance gains of Intel Nehalem 5500 series processors and DDR3 memory, also incorporates Sun IP designed by the brilliant Andy Bechtolsheim such as "FlashFire" flash memory cards and 40Gb/sec Infiniband switches, that all told, can reach up to 1 Million random read/write IOPS. These performance numbers make Exadata V2 scream in running OLTP workloads, but can also run data warehousing workloads twice as fast as Exadata V1.
Subsequent to the joint Sun Oracle product launch, several analysts were quick to point out that Exadata V2 incorporates Intel x86 processors and Oracle Enterprise Linux as opposed to Sun's latest generation "Victoria Falls" dual socket, eight-core CMT processors and Solaris 10, appearing to contradict Oracle's recent WSJ ad proclaiming its increased investment in the SPARC/Solaris platform.
But rumor has it that Larry is saving his joint Oracle-Sun SPARC/Solaris product announcement, that will likely be even bigger on superlatives, for October 12th at this year's Oracle OpenWorld.
Although 2009 marked the summer of silence for the technology sector, it's gearing up to be anything but that as we close out the remaining months of a challenging, yet interesting year.
Thursday, Oracle ran a follow-on to its August 27th Wall Street Journal ad. The Thursday ad was much more pragmatic, and directed at Sun customers who continue to wrestle with the uncertainty surrounding the still-pending merger, officially delayed by the European Commission on September 3rd due to its "serious concerns" about the database market impact. Unless Oracle makes a dramatic concession concerning MySQL, the Europeans will keep the merger in limbo-land for another quarter.
On its face, the ad gives every indication Oracle will keep Sun's SPARC/Solaris hardware business. The ad also is an acknowledgement that "Ellison and company" realize full well that Sun is bleeding server market share, and therefore, needs to do what it legally can to prevent further customer defection of the SPARC/Solaris platforms.
Call me a nitpicky parser of words, but the first three promises in the ad all end with the operative words "than Sun does now;" Since Sun's R&D budget for SPARC and Solaris is likely at a historical low, and the company has made drastic cuts to its employee headcount over the last year, Oracle's assertions are not as bold as they appear.
In addition to Oracle's claim it will increase its investment in Sun SPARC and Solaris, the ad takes a strong competitive posture towards IBM. It's not surprising that Oracle would target IBM since it would be the only other company that could truly offer a comparatively integrated enterprise hardware and software stack. But it's a bit curious why HP has been spared from attack. After all, both HP and IBM have been relentlessly pursuing Sun's customer base with enticing trade-in and migration services programs over the last few quarters.
The ultimate irony is that my parsing of Larry Ellison's words took me back to 1993 when I was a sales rep at Sun Microsystems. At the time, IBM was on the watch list for extinction - victimized by missteps of its own making - and so the board brought in Lou Gerstner as IBM's Chairman and CEO. There was a lot of uncertainly surrounding IBM's future viability and whether Gerstner would break up the company. Gerstner made the bold decision to keep the company together, defiantly declaring, "The last thing IBM needs right now is a vision."
Most people took this statement to mean that Gerstner didn't believe IBM needed to be a visionary company. Years later, Gerstner clarified his operative words "right now". Perhaps Larry Ellison is taking a page from the old IBM-Gerstner playbook in carefully choosing his words in his latest Oracle WSJ advert.
VMware's CEO, Paul Maritz, has been touting the vSphere "software mainframe" analogy since his VMworld Europe keynote earlier this year, and undoubtedly, it's been resonating with the 45 and older IT crowd. But is it all good?
Numerous industry analysts have given the mainframe analogy a big thumbs up, and I acknowledge, it's clever marketing. With vSphere 4.0, VMware is taking defining principles of mainframe computing like Reliability, Availability, Serviceability (RAS), performance, and centralized control, and bringing them to commodity x86 hardware with its new application services like VMware Fault Tolerance, Hot Add, Host Profiles, and vShield Zones. With these mainframe-class features, VMware intends to drive deeper and broader adoption of it virtualized infrastruture software into the data center so enterprises will begin to virtualize Tier 1 applications.
Clearly, VMware's platform has grown from a hypervisor into a more comprehensive operating system that can handle much of the virtual resources in a data center. But does the mainframe vision come with a monolithic, expensive development and operating expense curve that lacks flexibility?
As VMware' competitive battle with Microsoft and Xen variants intensify, it will need to be agile and continue to deliver lower operating expense curves while adding more mainframe-like functionality up the software stack. This will be the ultimate challenge -- mixing agility and flexibility with mainframe features and functions at a lower TCO than the competition. My hunch is that time will drive adjustments to the vision and competitive focus. This isn't necessarily bad for VMware, but it's not all good.
The big build up by the press that Citrix was going to majorly diss VMware by unveiling its Xen Cloud Platform (XCP) initiative at Monday's VMworld kick-off was a dud. In fact, there turned out to be far more written about the technical glitches experienced in the lab sessions on Monday.
The meat of the XCP announcement was around the possibilities of community alignment, and the value proposition of open source technologies that can truly deliver on innovation, openness, and interoperability to free customers from proprietary, vendor lock-in.
Clearly, if the Xen.org community can fulfill its new charter to create stable, well-defined APIs around a single, compatible code base, it will drive rapid adoption of the Xen hypervisor as service providers and enterprises expand their virtual data center deployments in 2010.
