Thoughts from VMworld 2011
It was quite a busy week for me, as it was for most people:
- Recorded The Cloudcast (.NET) - Live - Eps.18 - vCloud and vCloud Security
- Part of a Silicon Angle "Cloud Realities" panel with Jay Fry (@jayfry3) and Matthew Lodge (@mathewlodge)
- Recorded the Daily Blogger Techminute for Cisco
- Recorded an Intel "Conversations in the Cloud" podcast with (@techallyson) - airing date TBD
- Presented "Cisco UCS and Cloud Vision" whiteboard in Intel booth.
I didn't get to explore all of the areas in my original list. But I did get to walk the exhibitor floor, attend the keynotes and have many hallway conversations with experts in most of those areas. So here's what I gleaned from the week:
VMworld ("the conference"):
Of all the events I attend each year, VMworld seems to be the one that does the best job at building interactive communities. Whether it was the outstanding "Hangout Area", huge social media wall, vExpert events or the daily vTweet-up events (often hosted by non-VMware groups), there were always casual opportunities to meet new people and interact around new ideas. Just an awesome collection of brain-power and really great people. Kudos to everyone at VMware for making community a top priority.
The only two suggestions I'd make to the organizers would be:
- Put people's name on both sides of the badge, and in larger fonts. Whether you're meeting someone in a booth, in the hallway or an event, it's good to know who you're talking to (especially in crowds) and all too often the badge is flipped over.
- I heard about many sessions, especially in the new areas (Cloud Application Platforms) where the rooms were under-reserved. While I understand the policy was that all sessions must be pre-booked to attend, it would be great to know when attendance is low and allow first-come-first-seated attendance to those events (full-pass or partial-pass). Just send a tweet 5 minutes before if seats aren't filling up.
As I watched Steve Herrod's keynote, which covered an incredible range of technologies and business use-cases, it became very clear that VMware has reached a point (that many companies with huge market-share in a certain category reach) where they are going off in many different directions. And as has been the case with many companies before, it's not exactly clear what the identify of the company will be going forward. Back in 2010, I joked that the company should change it's name from VMware to CLOUDware. Maybe that wasn't too far off. Or maybe APPSware or SERVICESware would be a better fit. At a macro-level, they are trying to play in all the game-changing markets (virtualization, IaaS, PaaS, SaaS, VDI, mobile) and their portfolio is very impressive. But managing the revenue transitions, since not all those technologies need VMware vSphere, and the broad mix of skills to make it all work are the challenges. It will be interesting to see how this evolves over the next 2-3 years.
vSphere is now a 5th generation technology, which is often (historically in technology cycles) the point where innovation slows and companies either move into broad market-segmentation or they begin to mostly add functionality that previously existed elsewhere (partner or competitive products). vSphere 5 cleared the final hurdle of being able to handle any size, any performance-level VM ("Monster VMs"), so what's left to conquer in the hypervisor space?
The partner ecosystem has always been a huge part of VMworld, but this year seemed different for a few reasons:
- As was noted back when the session catalog was announced, the bulk of the sessions were either given by VMware employees or were focused on VMware-only technologies. In the past, it was fairly balanced between VMware and complimentary ecosystem discussions.
- Neither Paul or Steve's keynote made any significant mention of partners, unlike in the past when partners often came on-stage for demonstrations.
- Many partners expressed concern about the way that the "open API or open plugin" architecture that was so robust and successful with vCenter has not been extended with vCloud Director. This may just be that it wasn't until vCenter 3.5 that plugins started and we're only on vCloud Directory 1.5, but the limited guidance or consistency from VMware about how this will evolve did come up in conversations many times.
Next-Generation End User Computing:
This was probably the area that had the most people excited, at least from a VMware technology perspective. Between Project Octopus (Enterprise DropBox), AppBlast (HTML5 App Client), Horizon Mobile (Mobile Hypervisor) and "Universal Service Broker", it appears that VMware has laid out an architecture for the next-generation end-user experience. As someone who doesn't want IT involved in the desktop/mobile space, I was excited to see the possibilities. But it created some questions in my mind:
- Will this be something that is run by IT, with their history of locking down functionality? Or will it be offered by Cloud Providers directly to Lines of Business or small groups, where they may value flexibility and pace more than the typical control of IT?
- It wasn't really clear how AppBlast works (where are those OS-bound, or Windows applications actually running?), but it did feel like a way to leap-frog past traditional VDI. So are we at the beginning of an era of "legacy-VDI" and "Cloud-VDI" (or "Cloud-Desktop"), where "Cloud" could refer to either the delivery model (in-house or hosted) or the viewing model (in a browser, independent of OS)?
- Enterprise-specific tablets have had a difficult enough time finding success in the market, so did Horizon Mobile just put the final nail in that coffin? It doesn't work with iOS yet (huge gap), but if you're an Enterprise (or Gov't), why not just have a BYOD policy (subsidized or not) and use the mobile hypervisor? The downside is lock-in to the VMware Horizon Manager for AppStore. It also seems like a huge opportunity for some company to fill in the lack-of-iOS support with a product/service/solution.
This is a very new space for VMware (Spring, vFabric, Cloud Foundry), but it's definitely near and dear to Paul Maritz's heart, as was stated throughout his keynote. There were some announcements around vFabric Data Director as well as partnerships (and micro) for Cloud Foundry (Cloudcast interview with Dave McCrory about Cloud Foundry), but the show is still dominated by infrastructure-centric attendees. While topics like vFabric or Cloud Foundry might be more appropriate topics at conferences like OSCON, GlueCon, LinuxCon or Cloud Camps, VMware might want to take a page from EMC's book (Data Scientist Summit at EMC World) and hold an 1/2 day event specifically for PaaS developers.
From purely a technology standpoint, Storage got the most buzz at VMworld. Specifically, any storage that involved the words "SSD" and/or "Flash". This has been building for most of 2011, so this wasn't new, but it's starting to reach a point where big things are going to happen. Those things will be M&A activities and huge customer deployment announcements. Fusion-io still gets the most buzz, but there was quite a bit of anticipation about where all-flash arrays like Pure Storage would stick in the market, and where heavyweights like EMC or NetApp would go next.
The other two concepts that seem to get occasional mention were:
- Combined compute + storage solutions that deliver HA/DRS/vMotion capabilities without the use of external SAN arrays.
- The need to move past traditional File/Block methods of describing how data is stored for VMs, and moving to native-VM-aware technologies that aren't bound by the limitations of File/Block overhead.
VMworld isn't a networking-centric show, like InterOp, so you don't expect a lot of networking to be discussed. But there were a few interesting areas:
- The IETF Draft of VXLAN was announced - jointly written by VMware, Cisco, Arista, Broadcom, Citrix and RedHat. VXLAN provides a L2 overlay for L3 networks, putting control of VM mobility at the edges of the network (hypervisor, vSwitch, vFirewall edge).
- With the VXLAN announcement, many people asked what this meant for existing "VM Mobility in the Network", such as OTV, LISP or the variety of ways that network vendors are attempting to build next-generation Layer 2 (or "flat") networks. I suspect that customers will ultimately use a combination of technologies and architectures. It's not like this is the first time "all the smarts at the edge" approaches have come along. Microsoft tried to put IPsec and SSL in every Windows end-points to make the network dumb ("and simpler") but that never quite caught on. Maybe this time will be different, or maybe it won't. Only time will tell.
- Cisco and Juniper announced enhancements to their virtual firewall products, but (as mentioned above), it's not 100% clear how the VMware vShield roadmap and/or APIs will allow these to integrate and innovate.
Cloud Management, Automation and Orchestration:
Nothing huge here in terms of product announcements, but it was definitely top-of-mind for almost every infrastructure-centric professional I met. There is an urgency to "get smart about Cloud Management, Automation and Orchestration". They seem to realize that they can either watch their roles get automated, or be the expert on how the automation gets designed and deployed. The trick is figuring out which Management/Automation/Orchestration tools work and how to differentiate them. This still seemed to be an open-ended question for many people.
Huge opportunity for people to create trainings, organize local discussion events, recruit talent, etc.
This demo got me thinking quite a bit about the possibilities for bringing data closer to compute, or compute closer to data.
(Note: Opinions expressed in this article and its replies are the opinions of their respective authors and not those of DZone, Inc.)