20080917 VMworld Dr Stephen Herrod Keynote
I'm not sure why conference directors think that people enjoy walking into a darkened room, facing disorienting flashing lights coming from the front of the room, watching company ads and customer profiles on overhead videos combined with extremely loud, unpleasant, pulsing music. I don't remember the show being as intrusive and unpleasant before. This reminds me of a Microsoft event. Hmmm. I wonder why.
Dr. Stephen Herrod, VMware's CTO and Sr. VP of R&D, presented a keynote address this morning. The following notes present what I heard and a quick analysis of what was said.
- VMware has 2,500 people in R&D, 6000 staff members in all
- Dr. Herrod touched on the key catch phrase of the event several times - "virtually anything is possible"
- Virtual Datacenter Operating System Roadmap
- Two different functions - coordinate hardware functions and coordinate application functions
- Claims that this is for all applications even though it doesn't support mainframe or midrange system applications at all
- Three infrastructure services
- vCompute - touched on partnerships with Intel but, once again avoided any mention of Mainframe or Midrange applications as if they don't exist. Up to 8 virtual processors and 8 GBytes of physical memory are on the roadmap. He also touched on another form of virtual processing, clustering. vCompute is going to support "resource pools" that will support up to 64 nodes, or 4096 cores in a single resource pool. Up to 64 Tbytes of memory will be supported. (Sound like things Virtual Iron, Scalent Systems, Surgient, VMlogix have been saying?) Distributed Power Management (DPM) will allow some physical machines to be powered down when no longer needed (sound like Cassatt's messages?)
- vStorage - touched on partners focused on site recovery manager - replication, snapshots and recovery. He also spoke about a new Clustered file system - VMFS. Storage vMotion mentioned (sounds like storage replication offered by quite a number of suppliers) He discussed "thin provisioning" as a way to leverage shared storage more effectively. VMware is going to offer a new API allowing others to use the facilities of VMware's software.
- vNetwork - work to work with networking in an abstracted way. This sounds like technology many others have been working on for quite some time. He pointed out that each ESX server today is working with a virtual network switch - vSwitch - today. In the future, this will expand to datacenter wide capabilities. Once again, I heard no mention of the fact that industry standard systems are only a part of organizational datacenters and that organizations will still need to do something about what their mainframes and midrange systems are doing. Dr. Herrod mentioned Cisco's Nexus 1000V will work with VMware's tools. I guess that this is how VMware thinks that large organizations will integrate their entire datacenter.
- Application services
- vApp is a new construct that will be based upon standard frameworks. This wil wrap entire workloads into a single package. This will allow metadata that defines how applications should run, what resources they require and the like to be attached to the applications themselves.
- Dr. Herrod pointed out that their strategy includes ways to manage planned and unplanned downtime for interconnects, storage, servers (all of this is in Site Recovery Manager).
- He went on to talk about VMware Fault Tollerance. This goes beyond the capability of the current VMware HA's ability to reboot a failed VM on another machine. This forces a shadow copy of a VM to run in lock-step with another VM to create a fault tollerant environment. vLockStep sounds very much like Marathon Technologies' everRun to me. The demo looks a great deal like Marathon's. It appears that Marathon's failover is faster if the demonstrations are any guide to performance in the real world.
- Security vServices - its really important to be able to specify security policy outside of the virtual machine. This means that intrusion attempts and other securty problems can be dealt with in ways that protect virtual macines. VMSafe is VMware's entry. Citrix has had this capbility for quite some time with their software. Dr. Herrod pointed out that a number of partners are integrating their software into VMware's framework.
- Management - VMware vCenter
- Here's where Dr. Herrod brought in VMWare's Lab Manager, Stage Manager, Lifecycle Manager and Site Rcovery manager.
- There are a some new ones including vCenter Config Control, vCenter Orchestrator vCenter CapactityIQ, vCenter ChargeBack. As with earlier comments, all of these things have been available from other players for quite some time.
- vCenter AppSpeed - discovers the topology of an application, monitors the behavior of the application and then remediate the issue by allocation more physical resources as needed. I'm, once again, reminded of offerings from server other suppliers. That being said, the demonstration was very impressive.
It appears to me that the folks at VMware have been tracking what others have been doing, cherry picking the best concepts offered by others and then starting projects to add those features to their environment. While in the end, end users ended up with a more complete, more functional environment, it doesn't demonstrate industry leadership that one has come to expect out of VMware.
The overall capabilities appear to be valuable for the applications assigned to industry standard systems in an organization's datacenter. That, of course, is only part of the organization's overall IT assets.