Marketing Memos

Where is your cloud and how do you manage it?

I toss that question at an occasional CIO and get borderline lucid answers. Most don’t yet use clouds but lust after them. Others leverage public clouds for non-privileged and mission-uncritical work. A scant few have cobbled together their own private clouds (p-clouds).

P-clouds and rentable clouds are as similar to Sherman tanks and kangaroos.

The promise of agile clouds is that you would be able to establish your own internal p-cloud and extended it ad hoc to external, rented cloud resources. Currently this requires either a significant amount of home grown engineering or adherence to one or another public clouds tools and management protocols.

Either approach is anathema to IT.

All radical growth spurts in IT technologies have occurred when open standards were popularized. UNIX killed MPE, VMS and other also-rans. Likewise Linux is slowly killing proprietary UNIX and blocking Windows Server growth. TCP/IP and Berkeley Sockets killed Netware, Vines and a slate of sluggish competitors. x32 and x64 chips have all but eliminated SPARC, Itanium and other red-headed step children.

Standards make stuff happen because it invites commoditization and interoperability, the top two CIO wet dreams.

It is unsurprising then that a number of the smarter industry players are pushing standards for cloud computing. When a document titled the “Open Could Manifesto” is singed by IBM, Sun, VMware, Cisco, EMC, SAP, Advanced Micro Devices, Elastra, Akamai, Novell, Rackspace, RightScale and GoGrid, you see that standards are as important to vendors as well as their customers.

Oddly, HP and Microsoft are not on the list of signatories.

Microsoft’s absence is understandable, as is Amazon. Ballmer’s bezerkers have launched Azure. In their effort to own everything, Microsoft wants no part of plans that commoditize clouds. Amazon is as an understandable absentee too. Having popularized clouds they do not wish to diminish their lead in the industry.

HP however is puzzlement. The manifesto asks vendors to “ensure that the challenges to cloud adoption (security, integration, portability, interoperability, governance/management, metering/monitoring) are addressed through open standards.” HP has made good profit from standards be they PCs, servers or the network management protocols that feed their still wildly popular Openview suite. Perhaps HP was excluded from the manifesto group as IBM/Sun merger talks were in progress. Nothing marginalizes a competitor quite like excluding them from a standards group.

In their trashing of the manifesto, Microsoft marginalized itself.

The bugga in the boo is that the manifesto is nothing more than a declaration of desire – a love letter to the market. It sets forth high-level principles for cloud vendors to adopt and little else. It is a more threat than action – a way to get competing vendors to either commit to open interoperability or appear to the public as old-school lock-in tech companies (hence Microsoft’s absence). Like other religions, it articulates noble goals through pretty words.

The beheadings come later.

The Open Cloud Manifesto web site is sparse – more of a staging ground for discussion than a hotbed of action. Without action, without translation of high minded desires into concrete cloud interoperability, it stands as nothing more than a wiki of wonder. Let’s hope that IBM buys Sun and repurposes it as the leader in open cloud technology, using the Open Cloud Manifesto organization as engineering epicenter to the next wave of IT infrastructure.

Finally, a good use for Sun.

Bookmark to:

I openly speculated last January about who might buy Sun Microsystems since investors were trading shares for Sun’s cash value and little more. This morning Internet cables were buzzing over IBM’s alleged discussions to buy Sun for about double the current share price.

If the rumors are true this will go down in the annals of the technology market as one of the great milestones, and perhaps not a good one.

IBM is if nothing a shrewd entity. Outside of Congress, making a $6B investment is not trivial for anyone and IBM will certainly invest more wisely than the set of currently elected simpletons. Yet doubling the going price for all shares of Sun is a seemingly extravagant gambit and indicates that IBM sees something of value that the market does not, and they want to keep competitors away.

What does IBM want to acquire? When walking down Sun’s product list we can make some reasonably safe assumptions:

SPARC: IBM may see some intellectual properties in Sun’s chip design, but it is doubtful the patent portfolio is worth Big Blue’s attention. IBM’s Power processors might achieve some added oomph, but the market is moving ever onward to commodity computing in clouds. It is doubtful that IBM thinks SPARC innards would even slow that trend.

Storage: Sun overpaid for Storagetek and never realized the value of the product portfolio. This says a lot about Sun’s market clout, which is none. Storage is a growth market as we upright apes continue to horde bytes as if the Internet itself was scarce. IBM has storage solutions covering all the Sun corners. It is unlikely there is some game changing component therein.

Solaris: UNIX is dying thanks to Linux, and IBM has its own mutant variant called AIX. If the deal goes through Solaris will live as an obscure Open Source project after being jettisoned from the IBM Death Star.

Networking, desktops, and other hardware: All uninteresting markets or where IBM already has superior market position.

This leaves software, a market in which Sun has made repeated mistakes but still retains gravitas. Glassfish, Netbeans, Sun Studio, and Identity Management are all throw-aways. MySQL is of minor interest to IBM who still touts DB2 as the single biggest threat to Oracle (which it is not … market trends are) but gives IBM a possible block-and-grow strategy against Larry. StarOffice is intriguing to continue IBM’s perpetual push against Microsoft. Sun’s virtualization products have potential, especially the Virtual Data Center (VDC) management console and to a lesser degree Sun’s desktop virtualization backend (which would put IBM in competition with Citrix and dovetail nicely into p-clouds, and area IBM is not ignoring). There are gains to be made by buying Sun’s installed base for telecommunications and shunting competitors.

Then there is Java.

Despite being (more or less) Open Source, Java is a linchpin in daily computing. Your browser runs Java. Your cell phone runs Java. Embedded devices run Java. Enterprise server applications are written in Java. This set of technologies has an infrastructure reach broader than anything aside from Linux, and with a little work Java wouldn’t even need Torvald’s virus.

IBM is no stranger to Java, having adopted it as one of its own. In Java’s early days of popularity, IBM employed more Java programmers than Sun, using Java as the backbone of their own products and thus detaching themselves from competitors (i.e., IBM products would work on competitor platforms without IBM having to get in bed with the competitors). IBM can gain from taking leadership of the world’s second most successful software technology after Windows (please note I said successful, not usable). Since server-side Java is popular in enterprises and since IBM supplies entire data centers, adding Java management is a crown jewel that will solidify IBMs enterprise dominance while HP is splitting itself between business and consumer markets (though they are doing so quite well).

IBM’s purchase of Sun is a multi-part strategy. First, it consolidates the IT hardware market which needs it. IT hardware is rapidly commoditizing, and commodity markets tend to reduce the number of competitors. Say bu-bye to SPARC, Solaris and other also-rans.

Second, it is a blocking maneuver. HP, Dell, Fujitsu and some others would like to own parts of Sun and could make good use of them. IBM’s purchase would prevent this and keep those competitors in a weaker position.

Third, a few desirable piece of Sun will augment existing IBM product lines or move IBM into a greater market leadership position. MySQL, Java and VDC are examples of pieces of p-clouds that combined with other IBM infrastructure components help to create a whole product for new millennia datacenters.

Now if those damn mainframes would die already …

Bookmark to:

When I read rumors that Oracle is buying one of our former clients I wonder what we did right.

Long ago and in the earliest phase of their interesting existence, we consulted to Virtual Iron. We helped with event management, presentations and removing kinks from their messaging when Virtual Iron was vending a lost cause. In those long ago days, Virtual Iron was selling sophisticated virtualization systems that not only let you divide one big server into many smaller servers (what most buyers perceive as virtualization) but they could glue several servers together via Infiniband switches and thus make many small servers look like one huge box.

They were a few tiny steps away from private cloud computing, though they did not know it nor was there an industry buzz word to propel their value proposition.

Virtual Iron was in what I call the ‘bottleneck bypass business’, which is always a lost cause. Products in the bottleneck bypass business are designed to work around some current computing bottleneck. This business model only works for a short time because hardware geeks are perpetually inventing better hardware specifically designed to eliminate such bottlenecks. New technology makes bottleneck breakers obsolete in short order. That is why I have gigabit networking in my house.

Have you evaluated a dial-up accelerator lately? There used to be plenty of them available … before cheap residential broadband came into being. They died.

Virtual Iron was if nothing else lucky and agile. In order to make their original virtualization system appealing, they spent considerable time engineering the virtualization of the individual components in servers. CPUs, memory, disks were all virtual objects that could be reallocated on the fly. Virtual Iron also wrapped this core technology with a world class user interface making a miniature virtual datacenter a reality.

Now Oracle wants them.

Oracle can rightfully be blamed for instigating virtual data centers. They were a bit premature and thus centered design decision on grids, which are limited forms of virtualization. Grids were good enough for what Oracle customers needed at that time –the abstraction of the database server. To Oracle’s credit, they invented and gave away some key technologies in order to popularize and standardize grid computing. Yet grids were chunky forms of virtualization. As with their database itself, Oracle’s early incarnations of grids were as difficult to manage as a child-filled bus on the way back from a field trip to a candy factory.

Then came VMWare and their virtual data center gestalt. VMWare is quite successfully articulating and engineering the p-clouds (private clouds). They are also doing what we at Silicon Strategies Marketing did for SuSE Linux to make them competitive against Red Hat – focusing not on the current and well understood solution advantages, but on the next step in product mindshare. When Red Hat was talking to system admins about cost saving we got SuSE talking to executives about integrated operations. When Red Hat started talking about integrated operations, we had SuSE speak about strategic partnerships.

VMWare may well own the future data center. Total abstraction of commodity hardware to fluidly manage data center workloads – as well leverage public clouds to extend data center horsepower on-demand – is nigh. This includes the virtualization of even the most sacred part of every datacenter, the massive centralized databases … including Oracle databases. VMWare has the lead and the marketing smarts to grab almost all this open field.

Nothing ticks off Larry Ellison like someone else dominating a software market.

When it comes to the virtualization of the datacenter, the second slickest product on the market is Virtual Iron. With VMWare’s vDC (I decline to use their poorly crafted VDC-OS name) threatening to one-up the Oracle grid and removing differentiation from Oracle and DB2 and MySQL implementations, Oracle had to do something. Building more virtualization (Oracle already has a lack-luster offering) would take too much time, produce uncertain results and detract from their core competencies (management lesson: never build products that are not part of your core competencies … in fact, never in-source non-core competencies including marketing strategy development). Buying their way into a competitive position and putting F.U.D. into the path of VMWare’s new vDC offering would be a classic block-and-run maneuver.

Hence Virtual Iron. Virtual Iron leverages the popular Xen hypervisor, which fits nicely in with Ellison’s Open Source infrastructures initiatives. VMWare products are infrastructure but not Open Source. Thus anything Oracle can do to promote open alternatives further blocks VMWare’s momentum.

However, this time Larry may lose.

Virtual Iron has been artful in vitalizing hardware and the management thereof. However, VMWare has already taken a substantial lead in abstracting the entire data center, including commanding issues such as storage, security, on- and off-premise hardware and application-level service assurance. Virtual Iron looks at the hardware while VMWare looks at the mission and the people managing applications. Virtual Iron and Oracle are too close to the bits while VMWare is empathizing with the administrators and helping them achieve their expected outcomes.

There are many technology marketing lessons in this rumor. First, if you have a way to leak rumors, you can create consternation in the market and for an opponent. Be warned though – this tactic carries a cry wolf effect, so use it sparingly. Second, solving problems from the expected/desired outcomes of your key genotypes is essential and gives you nearly unbeatable advantages that even wealthy opponents cannot buy their way out of. Third, stay away from bottleneck bypassing business models unless you want to make a quick buck and exit – they never work in the long run.

Forth, figure out where Oracle is not building toward trends in the market and supply them with a second place alternative. Larry has money and he is more than happy to spend it to maintain or expands the Oracle Empire.

Bookmark to: