Observations on VMworld 2014

Last week I attended my 14th VMworld and the first one where I was not representing VMware. I’m now just your average startup CTO attending a premier IT conference as a partner. One of my former colleagues put it very succinctly – “attending VMworld as a former employee is like moving back to your parent’s house!”

And storage was extremely well represented at VMworld. A look around the exhibit  floor showed numerous storage firms in all categories showing off their wares; the latest in Hybrid Arrays, All-Flash Arrays, Hyper-Converged Infrastructure. The leading I/O Optimization technologies were all there in force.

Infinio 2.0

And it was a great show for Infinio! We announced our V2.0 product that includes:

  • Unified Storage support, adding Fibre Channel SANs, iSCSI, FCoE and hybrid topologies to our existing NFS support. Infinio 2.0 will continue to be completely transparent to your operations, supporting all native storage array capabilities.
  • Application-level reporting with end-to-end drilldown from datastore to per-application views. This capability allows visibility into workload composition to better analyze performance and perform root cause analysis on I/O anomalies all while Infinio also mitigates their effect.
  • Cache Advisor capability which helps admins empirically determine how much memory to dedicate to Infinio’s cache by calculating the actual results that would have been achieved for selected period of time with different cache sizes.
  • VDI cache tuning with heuristics and algorithms for even better support of VDI workloads.

Our booth was quite popular, bringing in over 3000 visitors during the show eager to see Infinio Accelerator 2.0 in action. I also gave a vBrownBag talk on storage architecture and Peter Smith and I gave a well-received Infinio update at Tech Field Day. All in all a very successful show.

VMware Announcements

As far as the VMware announcements of personal interest that I’d like to comment on, I was pleased to see VMware launch a preview of some technologies I was instrumental in cultivating. VMfork, now called Project Fargo, is a very innovative vSphere technology that essentially does for runtime execution of VMs and their memory what linked clones did for storage. This capability enables live/run-time cloning of a VM and its memory space with page level Copy-on-Write semantics for modified pages. My colleague, Daniel Beveridge (@danielbeveridge) was the first person to recognize this technology’s potential impact for rapid, efficient provisioning of virtual desktops and application publishing. Also, inspired by Daniel’s years of work on this capability in the EUC CTO Office, Aaron Sweemer (@asweemer) and I made the connection to utilize this same capability for rapid, efficient cloning of Docker engine instances within a VM. These are very exciting, ground-breaking technologies and it was personally satisfying to see both the technology and use cases showcased at VMworld.

Project MARVIN also debuted, front and center at VMworld with the product names VMware EVO:RAIL™ and EVO:RACK™. This is VMware’s packaged software intended for server OEMs to enter the Hyper-Converged Infrastructure space with all VMware software by offering an “SDDC in a Box”. In my opinion, VMware strongly validated the pioneers of this market such as Nutanix and Simplivity, the former of which turned around and announced a new funding round at a staggering $2 billion valuation on the heels of a $1 billion dollar valuation only 6 months ago, while Simplivity announced a new Cisco partnership.  With the EVO:RAIL launch, Hyper-Converged Infrastructure is further validated and should heat up with a lot more competition!

VMworld also debuted two important storage technologies, VMware VVOLs for exposing VMs as first class policy driven objects to storage arrays and VAIO I/O Filters, a step forward from the existing PSA architecture for a more general purpose storage stack filter architecture. Infinio has been tracking and intends to leverage these capabilities. Note that the Unified Storage for Infinio 2.0 will initially use the PSA framework in a carefully designed hybrid architecture in order to optimally support current vSphere versions.

In addition to Project Fargo, the VMware EUC business unit announced the acquisition of CloudVolumes, which I expect to be a cool new feature of Horizon View. It enables rapid, seamless surfacing of applications by cross-“mounting” a shared virtual disk that contains the application layer into a guest VM through the hypervisor. This is accomplished without any application installation in the guest VM image itself nor any file copying into the VM. Both CloudVolumes and Project Fargo serve to enable a much more dynamic and rapidly provisioned environment, particularly well suited to non-persistent desktops. Initially CloudVolumes is an independent and separately managed capability and it will be interesting to see how VMware EUC goes about integrating all these new capabilities together into a unified management  experience and automated workflows. Btw – After a CloudVolume is mounted into one or more VMs, you still have to load the blocks comprising the application image into each VM’s memory in order to execute it and Infinio Accelerator should be very effective and complementary for caching and offloading this traffic, decreasing latency and improving user experience.

VMworld was exciting as ever and I very much enjoyed reconnecting with friends and colleagues at the show. And hopefully my readers appreciate my independent comments on these announcements!


This entry was posted in Software-Defined Data Center. Bookmark the permalink.

2 Responses to Observations on VMworld 2014

  1. Pingback: Observations on VMworld 2014

  2. Pingback: Completing my 2015 Predictions with Observations about Docker | Talking Tech with SHD

Leave a Reply

Your email address will not be published. Required fields are marked *