Well – if you’re a nitpicker – technically, VCE and Vblock were launched on Nov 4th, 2009. But, hey, if you want to be a REAL nitpicker, VCE existed internally to the 3 parents about 1 year prior to public launch in a less visible way.
I remember fondly being introduced to Cisco’s project California before the world knew about UCS (see Ed Bugnion and I presenting at VMworld 2008), and I fondly remember meeting with customer with esteemed colleagues from Cisco and VMware early in 2009 with the idea (and those customers applying pressure to us on what they wanted to see).
Looking back – lots of lessons learnt. A lot of experience under our belts. A lot of feedback – heard and incorporated (both into state of today and state of tomorrow).
I was with a series of folks in the press and with customers here at the vForum in Australia, and it made me think of just how much progress has been made – sometimes with all the frantic day in, day out activity, one doesn’t reflect on the big picture.
Customer adoption has been through the roof and is keeping the VCE and vSpecialist folks VERY BUSY… In Australia alone, and just listing SOME of the PUBLIC examples:
- Enterprise Using Vblock: Westpac, Curtin University, Curtin University, TPI, Komatsu, Melbourne IT, Thomas Duyera
- Service Providers delivering Public Cloud IaaS on Vblock: Optus
- Systems Integrators who offer public and private cloud on Vblock: CSC (IaaS and DaaS)
- Vblock resellers: NetstarLogicalis, Dimension Data
Outside Australia, the customers are LEGION. Some of my personal favorites: Kemet – a neat enterprise that builds capacitors based out South Carolina, Alcon – a neat enterprise that’s in the eye-care business. I like these ones because they’re businesses like many. Not huge household names (we have tons of those too) or massive service providers (check – have those too), but enterprises like many who simply want to accelerate their virtualization and private cloud journeys.
To put it in perspective – we’ve been building Vblock manufacturing capability (yes, there is a Vblock manufacturing line – as you would expect with any product) to be able to produce a HUGE amount of Vblock in 2011.
Speaking of Vblock customers… check out this HOT OFF THE PRESSES kick-a$$ customer example :-)
You can download the high-rez MOV version here for your offline-viewing pleasure :-)
What I loved (seeing early versions of this video) was when the folks on the VCE team working with Lockheed said “I want more cowbell” in the video. I think they got it :-) You can read more directly from Lockheed Martin here.
What are the key takeaways I’ve learnt over the last year?
- The biggest internal (i.e. in bringing the value proposition to market) challenges have proven to be business (eg. getting to the point where we can provide a single quote and pre-stage Vblocks – as one example), not technical.
- The biggest customer adoption (i.e. reasons why customers say “NO”) challenges have proven to be organizational (Vblock is a product that breaks the traditional “server/network/storage” team silos)
- The customer response has simultaneously been very different, and much louder than expected. Early on the Vblock wins were MASSIVE, and looked/felt a lot like “managed services” rather than “a product”. As 2009 progressed, we realized that far more common were people who wanted an integrated product more than anything else – and those came in every shape and size.
- there is unbelievable customer demand for the category of “integrated x86 infrastructure to support virtualized use cases” (lump in Vblock with other real alternatives).
- it makes our competitors go loony (and try to make things that are “Vblock-like” – but are not REAL alternatives), which along with customer demand is a good sign we’re doing something right.
What can you expect to see more of?
- More around orchestration. UIM v2 is now GA. UIM v2.1 (which adds Vblock 2 support) is right around the corner – this is ACCELERATING. It represents something that doesn’t exist anywhere else – a turnkey, integrated full-stack orchestration tool for an “integrated infrastructure” product. More importantly than anything though is that JUST LIKE vCloud Director, it abstracts out many APIs into one higher-level API set. There’s now a single API for a Vblock as a whole (while of course, just like with vCloud director you can still use vCenter APIs and ESX APIs – you can continue to optionally use the UCS XML APIs and the EMC storage APIs).
- More around efficiency. With EMC supporting Native FCoE, and “right around the corner” updates on UCS providing fabric services in the UCS 6100, the end-to-end 10GbE FCoE design will simplify the config, and also enable us to squeeze out components. Note how there’s a second customer benefit here – before it’s available in a Vblock, customers trying to do this would run into lots of problems. Just because something seems like it can be done, sometimes it can’t practically. You know that when the Vblock product is updated, we have put it through it’s paces.
- More around functions. Yup, the ideas behind VPLEX Geo and FabricPath and OTV are linked ideas. The Vblock product roadmap is a good forcing function on the parents roadmaps. Since there must be windows where major functional pieces come together (largely governed by VMware’s roadmap schedule), it forces the parents (who in effect “OEM the components to VCE”) to work closely on the roadmap.
- More around use cases. We have published boatloads (SAP on Vblock, Exchange 2010 on Vblock, SQL Server on Vblock, Oracle on Vblock, Sharepoint on Vblock, View on Vblock, Greenplum on Vblock, Cisco Unified Communications on Vblock – on and on and on) of application use cases – and it’s speeding up, not slowing down.
- More on better partner communications. We know we need to get better on this front.
What is something that’s a clue you have someone trying to make something “Vblock-like” – but isn’t analagous?
- Vblock is more than a reference architecture. Cisco Validated Designs are great, but are a document, not a product. You don’t “assemble” a product.
- Vblock is more than “brew your own”. Vblock 0, 1, 2 are designed products for different functional and scaling models. You can of course buy EMC kit, Cisco UCS kit, and VMware software and brew-your own, and do stuff before it gets into the Vblock, or make very different design decisions. WE SUPPORT THAT in the existing “mix and match” way we always have. Heck, there will always be a little lag from the parents products being updated and them being in the Vblock. That’s not a BAD thing, that’s a GOOD thing. When people try to make a comparison between a reference architecture (which is there to try to help you integrate technologies) they are mistakenly still operating in the “mix and match” model that Cisco and EMC still of course support. If you want to use the Vblock BOM and reference docs as the starting point, and make your own Vblock-like thing, that’s OK. The difference is a Vblock is a product. Think: when I build my whitebox servers, I can pick ANY components I want. That’s very flexible. When I buy a server, I can pick some components. One is “mix and match”. Vblock is there to accelerate the outcome.
- •Vblock single support is more than “joint escalation”. Our early customers were clear. “If you want to compete for this ‘integrated infrastructure’ business, you need to have a single place we call when we have a problem, and they own the case and case resolution from beginning to end”. That was one of the first signs to us that we needed to create the VCE Joint Venture, because “business as usual” (joint escalation) wasn’t going to cut it. Single support = no hand-offs, period
We still have much, much more to do – and are far from perfect. But we aren’t resting on our laurels. Feedback is always welcome.