Since the virtualization industry was largely created by VMware, is largely being defined by VMware, and is (currently) be lead by VMware it is worth taking a look back at 2011 from the perspective of VMware’s steps, key strategic directions, and missteps.
VMware’s 2011 Accomplishments
The most significant accomplishment for VMware is that VMware maintained and built upon its lead as the most important, and more significantly the most competent system software vendor on the planet. VMware is the most important system software vendor on the planet because VMware continues to define a new layer of system software (the virtualization platform) and VMware continues to redefine (to a reduced role) the role of traditional operating systems. On this front VMware continues to win the war with Microsoft and Red Hat as more and more traditional operating system functionality is subsumed into vSphere, and as VMware rolls out its own set of applications interfaces around vFabric and Cloud Foundry.
The question of competence is an even more interesting one. It is very instructive to compare how VMware rolls out new releases of vSphere with how Microsoft rolls out new releases of Windows Server. The key differences are that major releases of vSphere seem to come every year (conveniently right before VMworld) and releases for Windows Server are years apart, and that when VMware releases a new release of vSphere we do not seem to have to wait for three service packs before the software is stable enough to put in production.
With the above context in mind, let’s list what VMware accomplished this year:
- VMware announced and shipped vSphere 5.0. Again as mentioned above, there was remarkably little in the way of show stopping bugs, and the early reports of upgrades were all positive. It is, as always, a huge accomplishment for a major systems software company to ship an upgraded version like this, that adds so many features, and that does not have show-stopper bugs that require multiple service packs to fix.
- VMware announced Cloud Foundry. The SpringSource acquisition, vFabric and now CloudFoundry all speak to VMware’s belief that in order for a system software platform to be viable in the long term, it must have applications written to it. The importance of this was highlighted by VMware making Todd Neilson President of the Applications Platform Division, leveraging Todd’s previous success in building developer and ISV ecosystems at Microsoft and BEA.
- VMware announced and shipped vCenter Operations. VMware has long professed the belief that virtualization and cloud computing require a new management stack, and professed the intention of being the vendor of that management stack. While the first release of vC Ops was announced in the Spring, it was not until VMworld Fall that the management strategy really became clear. It became clear that VMware intended no only to build a monitoring and management stack for vSphere, but also to reinvent what management meant. VMware intends for automation, and in particular automated delivery of applications service levels to be its differentiation in the management market. The announcement of an upgrade to vCenter Operations at VMworld is a down payment on this vision.
- VMware entered the APM Market (again). VMware’s initial foray into the APM market with AppSpeed did not turn out so well. At VMworld Copenhagen VMworld announced a new APM product, vFabric APM that includes the AppSpeed technology, but adds significant functionality including detailed instrumentation of Java applications, application topology discovery, and a Performance Index (a score from 0 to 100 which indicates average response time and response time variability). VMware also showed the first features in the area of taking automated remediation steps when the Performance Index declines which is the end game for both the APM and Operations Management spaces.
- VMware announced View 5. Well, VMware finally got View 5 with Persona Management (the profile management technology from RTO Software) out of the door.
- VMware vShield 5. VMware has concentrated this year on providing tools that aid in assuring regulatory compliance within the virtual environment such as vShield App with Data Security and vCenter Configuration Manager (vCM), which can check the virtual environment against various compliance standards. VMware realizes that Compliance is not Security, but money is being spent on Compliance, and even though it is not security it is still an important part of any virtual environment.
- VMware intends to manage and commoditize storage, in the same manner as it deals with CPU, memory, and networking. The most tangible evidence of this so far is the vSphere Storage Appliance that turns local storage in servers into a virtual SAN for up to three hosts. However there were plenty of hints dropped in various sessions at VMworld that this is just the thin edge of a very large wedge. There are rumors that Windows Server 8 will also make the requirement for a SAN go away for features like Live Migration (vMotion). This will be very a very important development to watch, as VMware may end up being forced to embrace commodity storage to compete with Microsoft, which will not make VMware’s parent company (EMC) happy at all.
- VMware intends to reinvent the management software market. The depth and breadth of VMware’s ambitions here cannot be overstated. The goal is to create a new value proposition for virtualization and private clouds. The old value proposition was delivered by hard dollar savings from server consolidation (CAPEX). The new one is based upon hard dollar savings in Operational Expense (OPEX), delivered by automation of what is now a series of painful and laborious manual management processes. Imagine provisioning your application in AppDirector, having Infrastructure Navigator map it to it infrastructure topology, having vFabric APM automatically instantiate monitoring of application response time, and having all of the above integrate with vCenter Operations to ensure that application service levels are being met (monitor–>remediate–>notify).
- Project Horizon. The objective of Project Horizon is to “hasten the evolution towards a post Windows-centric end user device world”. Now that is an ambitious undertaking. But VMware demonstrated a series of concrete steps that it is taking in this direction. Managing identity centrally for SaaS delivered applications for a variety of devices is a good step. Allowing an end user’s environment to be delivered to an Android device via mobile device virtualization made for an interesting demonstration at VMworld.
- Cloud Foundry as a Second Virtualization Platform. We discussed Cloud Foundry as an application platform above. However, it occupies a far more strategic role in VMware’s future than as just another JVM or version of Tomcat. At the end of the day, everything always ends up being about applications – because applications are what users and businesses use to get work done. vSphere is “just” a device driver to the data center. The requirement for vSphere to be backwards compatible with legacy workloads limits what can be done in vSphere. But with a new applications platform like CloudFoundry, and new applications written to that platform, those limits are cast aside. CloudFoundry is therefore not just about running new Java applications on vSphere (although this certainly will be supported). CloudFoundry is a “meta PaaS Platform” that is about supporting today’s and tomorrow’s innovations in application frameworks (potentially, ultimately just about everything except .NET) and about running those next generation applications in internal and external PaaS Clouds – that may or may not have a vSphere infrastructure. CloudFoundry therefore gives VMware an entree into the unbridled future of where applications run, and how they are managed – one not limited in any way by what is possible or not possible in vSphere.
- Developing a Public Cloud Partner Ecosystem. One of the drivers of the success of Windows was that Microsoft developed the PC and Server vendors as strategic channels to the market. When you bought a PC or a Server, it just came with Windows and this became an accepted fact. VMware is well down the road towards doing the same thing with public cloud vendors. The idea is when you go to any public cloud vendor to buy cloud services from that vendor that those services are in fact powered by VMware technology, and that VMware shares in the revenue stream when you buy those services. This means that the public cloud vendors will likely be VMware’s long term strategic channel to the market, just as the hardware vendors were that channel for Windows.
- vShield APIs. VMware has also made vShield the integration point of the future for third party security modules, while VMsafe modules still exist, use of the vShield APIs is the future.
- New vRam Based Pricing. This has been cussed and discussed a thousand different ways already and there is no sense in going over old ground again. The bottom line is simple. VMware believes that pricing upon just CPU sockets puts it at business risk due to the constant increase in cores per socket that are being driven by the continued viability of Moore’s Law. Therefore VMware decided to also license on the basis of configured vRAM which amounts to a tax on memory heavy applications and on density within an environment. The most important ramification of this new pricing is that it forced a lot of VMware administrators to take a look at (but not yet buy) competing products – who then found that the competing products have gotten pretty good since the last time that they looked. VMware therefore opened a door to its competitors that had largely been shut due to its technical customers’ satisfaction with vSphere. This may have set off a resurgence in the prospects for Microsoft Hyper-V and Red Hat KVM in enterprise accounts.
- Per VM Pricing for Management Software. Whether this truly is a mistake or not remains to be seen, but this was the year that VMware rolled out enough management products around this pricing model so that it became apparent that this was going to be the model for pretty much everything except vSphere and vCloud. The biggest impact of this model is with vCenter Operations. The reason for this is that if you have 10 VM’s per host, and you install vC Ops Enterprise you will pay $200 per VM or $2,000 per host. Then if as a result of running vC Ops you learn (as you should) that you have room to put more VM’s on each host and you go to 15 VM’s per host, you will have to pay another $1,000 per server to do that. Most customers view paying the initial $2,000 for vC Ops as the price of the software. They then view that increase in density as “their” benefit of buying that software and resist the idea of “paying twice” to get that benefit. Per VM pricing is therefore a particular issue for any management solution that has increased density as its benefit, as it creates a second “density tax”.
- Poor Execution on the View Front. Yes VMware finally shipped View 5 with Persona Management, but it took way too long. However shipping View 5 does not address the fundamental issue with View, which is that server hosted VDI is just a niche strategy compared to the wealth of options that Citrix offers to deliver applications to end users. As long as VMware is just in the VDI niche, Citrix will keep dominating VMware in the market – as Citrix owns that circus and View is just a one trick pony. Horizon is important because it gives VMware a chance to break out of this niche – but of course Horizon needs to ship in order for that to happen.