• vCloud Hybrid Service Dedicated Cloud: Provides physically isolated and reserved compute resources, as well as a private cloud instance. Sold on an annual term with pricing starting at $0.13/hour
• vCloud Hybrid Service Virtual Private Cloud: Multitenant compute resource model, but with dedicated allocations for customers. Sold on a monthly term with pricing starting at $0.045/hour

It would be very easy to get critical at this announcement, and gasp at the fact that a company that formerly hated on the public cloud is now turning around and offering a hybrid solution. But that attitude ignores the reality for VMware customers who are often conservative and have only just developed an appetite for taking tentative first steps into the public cloud.

For those customers – application portability, an architecture they’re familiar with and an invoice from a vendor they’re already deeply partnered with is important. That’s the bottom line and the reason that, despite what the purists will say, this initiative will gain some traction.

It’s no threat to the big cloud providers of course, but it does raise some questions for those companies who make their income by helping large organizations built private clouds – their world just got a little bit more complicated.

In terms of technology, vCloud Hybrid Service will offer a simplified approach to management, allowing VMware customers to use the same tools and processes they use today to manage both on-premise and off-premise environments.  They’ll do this via the free vCloud Connector plug-in. As an interesting aside, VMware selected cloud automation technologies from Puppet Labs for the orchestration part of the offering – not surprising when one considers that  the company recently announced that it was putting $30M into Puppet Labs.

VMware is holding out an olive branch to partners – vCloud Hybrid Service can be sold the same way as on-premise VMware licenses with a standard SKU, and partners can retain the billing relationship with customers.

Interesting times huh?

While the OCP has made great inroads at opening up many parts of the data center operation, one area they’ve seen little impact is the world of networking. This is largely due to the fact that networking disruption lags behind other parts of the datacenter. While commodity hardware and open source software for compute is a well accepted and recognized approach, the same for networking is unheard of in the public arena. This despite the fact that many webscale operators design their own networking gear – in particular Google is rumored to do so.

Networking kit accessible to the general populace on the other hand tends to be highly proprietary hardware/software combinations that give customers very little flexibility. Traditional vendors like the fact that new technologies require updating both hardware and embedded software – awesome for their profit margins, but wasteful and expensive for customers. Which is where the Open Compute Project’s announcement comes in. The aim is to product an operating system agnostic, open source switch that, in the words of Frankovsky “can be treated just like a bare-metal server when it comes on the network”. The reference for the switch would create a piece of hardware onto which customers can load their own operating systems – in much the same way as customers load OpenStack onto standard servers. In doing so the initiative opens up the opportunity for organizations to take advantage of commodity hardware directly from OEM providers – and neatly sidestep the network giants in the process.

If this initiative gains traction, it raises some real concerns for the traditional networking companies – Big Switch Networks and VMware have put their names to the initiative but so to has Cumulus Networks, a stealth company that was founded by JR Rivers, a long time Cisco veteran and a former Google networking engineer. Google doesn’t talk about it’s forays into a new type of networking device but if the rumors are true, then Rivers was likely involved in taking commodity hardware and rolling it out within Google – taking that experience and applying it at a startup raises some interesting prospects. The fact that Cumulus has joined the OCP initiative raises them even more. Anyway – the initiative looks to get going immediately There is an OCP meeting at MIT next week and word is a number of companies with an interest in disrupting the networking status quo will be there.

Of course we already have OpenFlow, a protocol that allows users to manage their hardware. But this initiative is going further and covering any operating system that may be uses. In terms of how it will work, the projects aims to create a switch that has within it a “boot loader” that will let software be installed onto the device, across the network. And therein lies the opportunity for another open source software play – currently networking OSes are proprietary to specific vendors – Cisco, Juniper and Arista – imagine a networking OS built on top of Linux for example – it’s a prospect that has these other vendors quivering in their boots.

Change is something that has been lacking in the networking world – it looks like very soon it will be the new normal

• High-Availability: Self-healing to automatically recover when hardware, software, networks or even an entire region fails – and in less than a minute from a backup made less than a minute ago.
• Intelligent Auto-Scaling: Scaling to provide and charge for scalability in response to spikes in traffic.
• Rapid Restore Data Vault: ‘Time machine’-like capability to allow hosting company customers to roll back to last good version of files themselves in case of accidental data loss or malicious attack for websites, databases and mailboxes.

Of course the proof of a product comes from customer success and here is where the case study of US web hosting business BrickStreet Data Systems comes in. CEP Andrew Skattebo says that:

We wanted to grow our hosting business but our old platform was vulnerable to systems failures, traffic spikes and end users deleting their own files. It was also exceedingly difficult to do even routine maintenance on each server. That’s why after months of research and testing and finding other solutions either too complex or too expensive, we chose and deployed HybridCluster. Today we can offer a more cost-effective, high-availability hosting platform that is easier to maintain and takes away many of the risks that normally affect a business like ours.

I’m really pleased for Luke and the team – everything seems to be really fitting into place – the product is excellent and is already seeing wins in customer sites globally. The funding is in place to fuel the development of the product and the team is being built out with some excellent hires. I’m really looking forward to watching HybridCluster in the months and years ahead.

We’ve been hearing through the back channel for quite some time that AppFog has been struggling to gain marketshare and revenue. Despite some high profile funding, and being in the sexiest of spaces, the company hasn’t managed to leverage its position to scale. I’m led to believe that the company was facing a cashflow cliff, in actual fact it has seriously downsized its headcount in recent months.

Given this email then, I see that there are two possibilities here:

1. It’s simply a cost cutting measure designed to extend the company’s short cash runway. As someone said to me on twitter, “Extensive resource and low take up (high cost and low revenue?) isn’t good for a volume play like #PaaS“
2. CEO Lucas Carlson has managed to find some money somewhere and whomever that money came from has a real commercial reason to shut off the Rackspace tap

This story will be updated as more news comes to hand.

Fist pumping however isn’t enough – the analysts attending the event needed to see some proof that OpenStack was generating excitement beyond the vendor community. The motivation for the vendors to talk it up is obvious – quite simply, every traditional (and cloud for that matter) vendor feels sheer terror at the massive elephant in the room. Amazon Web Services is innovating at an incredibly fast rate and also owns the lion’s share of current cloud adoption. Recently VMware CEO Pat Gelsinger very publicly showed just how worried he is about the threat of AWS when he told partners that a workload lost to the Seattle-based giant is a workload lost forever – this sentient is one that is shared by vendors from across the spectrum – the old faithful’s like HP, IBM and Dell who were caught napping and need to move fast to make up for lost time. But also the public cloud vendors who, despite trying every approach they know, have yet to make real inroads to the hegemony they face in the form of AWS.

So if AWS’ competitors need to have a credible proposition in the face of AWS dominance, what is the value proposition for customers who don’t have the same commercial drivers that technology vendors do? Quite simply, the success stories from the likes of Netflix, Zynga and other adopters of cloud have shown them that the economics, gain in agility and technology benefits that cloud brings do in fact have validity for their particular situation. They may not have an appetite to consume technology in the way that pleases the public cloud fanboys, but fundamentally they want to start using “this cloud thing”.

Indeed this market opportunity – the enterprise customers who wants cloud, but in their own terms – has finally hit home as the direction to head – and the truly horrible to watch, but telling in its message, rap video from the first day of the summit tells it all – enterprise is the future for OpenStack. My assessment is that for vendors other than AWS (and perhaps Google as well), enterprise is their Hail Mary pass.

Of course talking up a focus is one thing, delivering wins in that direction is another. It was interesting that the metrics OpenStack chose to crow about to the analysts were, as my friend Paul Miller points out, somewhat softer ones that customer wins:

The message was one of steady progress (a 26% increase in patches to the code since Q4 of last year), broadening community (a 27% growth in participating companies), and growing interest (a doubling in website traffic)

In OpenStack’s defense there were some interesting case studies (Best Buy, Bloomberg, Comcast, the National Security Agency and one really interesting brand that we’re not permitted to name) but these case studies tended to still be either in lab-trial form, or else were small deployments across limited parts of the customer organization. Indeed much of the hubbub around the ill-advised and ill-timed “announcement” by Mirantis’ Boris Renski that Ebay was moving off VMware and on to OpenStack (an on-again, off-again story that would seem to have been largely fiction beyond Ebay having a bit of a play with OpenStack) was from the analyst community that is just dying to see some real-life production success at scale for OpenStack.

Josh McKenty, the ever debonair CEO of Piston suggested that there were “around” 400 production deployments of OpenStack, plus perhaps 20 times that number of pilots – while that’s positive to see, the community needs far more, far larger and far more public success stories soon. On the morning of my departure from the event I went for a short jog with Luke Tymowski an engineer from one organization using OpenStack in the wild, Canadian academic network provider Cybera. Cybera have an interesting OpenStack history, they’ve built a range of projects on OpenStack – digital sandboxes, a learning management cloud and a space weather modeling platform being some examples. Interestingly enough, a couple of their projects were built on VMware and Eucalyptus backends and both have been rebuilt on top of OpenStack – in the case of Eucalyptus because of core functional lackings, and in the case of VMware unsurprisingly because of cost. It was awesome to have a chat with him about what success they’ve seen using OpenStack but, again, as positive as the story is, it’s not the sort of thing that garners attention in the way that AWS’ alleged $600M CIA cloud does.

But there’s a bottom line here that we all need to be aware of. No matter the share it has of hip young startups, no matter how rapidly it innovates and no matter how much attention AWS gets with its spy-cloud, AWS isn’t an enterprise company. While in a recent exchange with Rackspace executive Lew Moorman, Benjamin Black, founder of Boundary and proponent for all things Amazon pointed out the fact that Amazon is moving fast and out-innovating the other players. But the very real point he ignored is that enterprises would rightly forego a significant amount of “bleeding edge” for robustness, reliability and maturity. AWS has a history in super low-margin, super low-touch service provision. There is an entire ecosystem around OpenStack that realizes that what enterprises really want is very different from that – they want rock solid guarantees, they want a traditional support model and they want a roadmap that is clear and palatable.

It is a subject of much debate of course – noted Clouderati and AWS pinup boy via his role as Netflix cloud architect Adrian Cockcroft is adamant that Amazon gets this enterprise cloud stuff:

@zalez how’s the AWS enterprise sales stuff going? @benkepes isn’t sure if AWS “gets” enterprise. I figure they hired you, so they do now.

— adrian cockcroft (@adrianco) April 22, 2013

I’m yet to be convinced – but will be watching with interest to see how AWS develops its enterprise mojo over time.

Either way, we need to accept that for enterprise, yes, their sales cycle is significantly slower than a Netflix or an Instagram. Which is just the paradox that we need to remember when opining on the chances or otherwise of OpenStack. Sure we’d all love to be able to write about massive production deployments of OpenStack today – but enterprise doesn’t work like that. What we are seeing is a slow, but steady maturing of both the product and the initiative as a whole. It’s not over, as the saying goes, until the fat lady sings. But I sense that the OpenStack summit attendees who were watching carefully saw a larger woman warming up in the wings. Watch this space

My friend Krishnan Subramanian of Rishidot Research where musing over this fact and decided to do something about it. We’re holding an even ion the 9th of May in Las Vegas that is going to bring together 80 of the most influential, the most important, the most visionary and the most opinionated cloud and infrastructure people together in an invitation-only event to look at what infrastructure will look like in the year 2020.

The Cloud 2020 Summit, as the website articulates, is borne out of the fact that from chips to servers to data centers to infrastructure software, we are seeing tremendous transformation in the name of cloud computing. The shift in the landscape is occurring at a rapid pace leading to large scale confusion on where the industry is headed. At Cloud 2020 Summit, we plan to decipher where infrastructure services are headed in the year 2020.

We’re looking at a few broad topic areas:

• The data center of the future

• Chip design and server density issues
• Data center design, efficiency and sustainability
• Security – how to secure data in this brave new world
• The building blocks

• Storage – handling the mass
• Networking – extrapolating out the software defined network in the future
• The cloud operating system – the layer on top of the infrastructure
• Controlling our heterogeneous future – the control plane that pulls all this stuff together

The Cloud 2020 Summit is taking place at the SuperNAP and participants will be invited to take a tour of the facility which, believe me, is pretty awe-inspiring.

So if you think you’ve got something to offer in the discussion about the future of infrastructure and would like to attend the event, please feel free to submit an application – our advisory board will select attendees in the next couple of week.

The theory goes something like this. Cloud computing has meant that data enter operators can add or change servers and storage almost instantly – all programmatically. The network however hasn’t got the same level of flexibility. Provisioning the network still requires lots of technical planning, manual configuration and complex systems and processes. SDN aims to reduce much of this technical burden by using a software layer for much of the network provisioning.

Alcatel Lucent, or at least its apparently more nimble spin off Nuage Networks certainly sees SDN as valuable and it is today announcing a new solution that is aimed to help organizations optimize their data centers for their future needs. The open software based solution called Virtualized Services Platform, is focused on some distinct verticals – heathcare, banking and utilities, as well as more general sectors. Trials of the solution are set to begin next month and three organizations that have signed up to be guinea pigs include UK cloud service provider Exponential-e, French telecoms service provider SFR, and US regional health service provider, University of Pittsburgh Medical Center (UPMC). According to Alcatel Lucent, worldwide commercial availability is planned for mid-2013.

So, what will Nuage Networks’ VSP solution actually do? According to the company it:

• Provides an open, software-only solution that works with OpenStack, CloudStack and VMware cloud environments and any datacenter network switches
• Delivers network connectivity inside the datacenter, between datacenters as well as datacenter to Enterprise VPN connectivity
• Improves server utilization and efficiency by up to 40 percent according to Bell Labs studies, removing the problem of available server resources becoming stranded because the network wasn’t configured to deal with a particular need or change
• Makes the network available instantaneously no matter where the application is running – even if the application moves to a virtual machine in another rack of servers or to another datacenter
• Delivers full programmability using a set of standardized APIs in clear IT-friendly language

Which all sounds excellent. The only concern being that this is a pre-announcement of a technology offering that will go into trial next month. I’m not disputing that VSP exists, I’m only saying that I’m really looking forward to a follow up call in a couple of months from Alcatel Lucent to report upon the results of the trial. SDN is a very valuable thing – but the proof of the pudding is, as they say, in the eating.

CloudVelocity

CloudVelocity’s aim in life is to extend the enterprise data centre to the public cloud. They promise to enable multi tier applications to run in the public cloud – securely and with no impacts on performance. CloudVelocity was founded by engineers with a background at NeoPath Networks and Sun Microsystems. The promise was attractive to backers with Mayfield Fund stumping up with $5M in Series A cash. The CloudVelocity approach is to clone private applications and then forklift them, in one swoop, onto AWS. Subsequent releases are supposed to support other public cloud providers.

CloudVelocity sees a continuum in the functionality it wishes to offer, at the low end it is simply a case of easing the migration of workloads from traditional to cloud infrastructures, thereafter comes failover. From here it is, theoretically, a short step to true cloud bursting. CloudVelocity isn’t there yet but is promising live cloud bursting by the end of 2013. At the same time they’re also talking about their offering providing a degree of management. It is interesting to hear these vendors start with high level talk of true cloud bursting but then slowly modify their language to be more about management – it seems the engineering realities around cloud bursting are an under appreciated obstacle.

Aside from the technical barriers however, it’s often argued by enterprise practitioners that there is no real desire to move legacy workloads between infrastructures – yes those applications need to be exposed to the outside world at the read/write layer, but in terms of the plumbing they sit upon, the usual approach is to simply leave them where they are – “if it ain’t broke, don’t fix it”. CloudVelocity has a different view and believes that the overwhelming majority of enterprises will want to take advantage of new platform architectures rather than the traditional views they currently have – time will tell.

Cliqr

I also talked with another startup in the space, Cliqr. Cliqr was founded by a bevvy of ex VMware executives and has also picked up its own funding – this time from Google Ventures. Cliqr is positioning itself as a platform to move, manage and secure applications. They’re aiming to offer application templates, application benchmarking and, eventually, the ability to run applications anywhere. More unicorns and rainbows?

I asked Cliqr how they thought they differed from the plethora of other players in the space. Their response was interesting, in their view there are three distinct takes on this application portability space:

1. First-Generation Script-Based Migration Tools such as RightScale and the recently released AWS OpsWorks. These platforms hard-wire applications to Specific Cloud API’s. Because of this approach, using these platforms for multi cloud operations is possible but requires multiple migrations to do so
2. Newer Image-Based Tools Provide Good Portability like Ravello, Cloud Switch. In Cliqr’s (admittedly biased) view, these platforms typically introduce a migration and performance overhead. They’re also not really focused on the management side of things and hence don’t offer much in the way of application visibility or management. With these platforms, it is problematic to deliver secure scaling, unless scripts are used. In moving to a script based approach however the portability of applications is restricted
3. Application-Centric Cloud Management Platforms. The holy grail where Cliqr believes they play. Offering the ability to on-board once and run anywhere, to move in real time between cloud and between private infrastructure and clouds. These platforms also have a management and security aspect to them.

MyPOV

Bottom line? No one has yet cracked it. Nobody offers multi cloud portability in real time, with reasonable scalability and robust security. All the players have a different take on how they work and have different strengths and weaknesses. They’re all proxies for another contentious topic, open and widely adopted standards. After all with standard architectures and approaches there would be significantly less barriers to cloud migration – and therein lies the rub, these vendors are essentially working against the cloud vendors somewhat underhanded attempts to increase the stickiness (or lock in if you will) of their platforms. it’s not a new trait for technology – the industry has been built on this passive/aggressive way of talking open but building closed.

In an idea world therefore cloud bursting would be an easy thing to deliver – the reality is and will always be quite different. Vendors such as those above will continue to try and solve the weighty issues around moving workloads. No one is there yet and the reality is that true cloud bursting is a long way into the horizon.

First a recap of what the organizations looked like before, and what they’ll look like post-spin out. Currently VMware, the market leader in virtualization, also has some other assets – end user solutions like Zimbra and Horizon Suite and PaaS offerings centering on Cloud Foundry. EMC dominates in the storage sector, but also has some peripheral offerings like Greenplum and Pivotal Labs. Essentially this new structure sees both VMware and EMC offload their more edgy solutions into a new entity, Pivotal, which will be spun out. Pivotal will also gain 1250 engineers (500 from VMware and 750 from EMC) and will be owned 69% by EC with VMware taking the remaining 31%. The new entity will also have $300M in existing revenue to fuel its growth.

So far so good right? EMC and VMware get to remain in their super profitable existing markets, without having the confusion of disruptive technologies in the fold. At the same time they get to own an independent entity that is immersed in the new paradigm of cloud, PaaS and big data, and get to do so in a well resourced way. A match made in heaven? Except:

VMware and End User Solutions – Ditch ‘Em

VMware focusing on end user solutions just doesn’t make any sense to me. I was pretty skeptical when the company bought Sliderocket and I’m happy to see they’ve recently offloaded it. But they still have the email application Zimbra in-house, seemingly floating along as an orphan. Ditto for social networking solution SocialCast – both solutions that should have either been offloaded or spun out to the Pivotal business. You then have the Horizon Suite – a kind of uber VDI-on-steroids that arguably isn’t a particularly good fit within a company which derives the vast majority of its revenue from core infrastructure.

Those Fighting Word About AWS

At a recent partner conference, VMware CEO Pat Gelsinger came out with all guns blazing about AWS. In a statement that left no doubt in anyone’s mind just how worried VMware is about AWS, he said that:

We want to own corporate workload, we all lose if they end up in these commodity public clouds. We want to extend our franchise from the private cloud into the public cloud and uniquely enable our customers with the benefits of both. Own the corporate workload now and forever

Clearly this fighting talk was an effort to rouse their partners and motivate them to get out and fight the risk that AWS poses. But in trying to motivate their own partners, VMware has a real problem:

About That VMware Public Cloud

And this is where it gets really complex. At the same time as it was offloading its more disruptive assets so that it could focus on core revenue, VMware was also preparing to announce its move into the public cloud. The service is designed to compete with all the big names in public cloud provision – AWS, Rackspace, HP etc – but is yet to be fully unveiled. What has been announced is that the initiative will be headed by former president of Savvis, Bill Fathers. Apparently VMware is really committed to this public cloud offering and it will be getting “the level of investment appropriate to that priority and to capitalize on a $14B market opportunity,” according to Matthew Lodge, VP of Cloud Services

While VMware is trying hard to placate the public cloud naysayers by calling this a “hybrid” offering, in reality this has more to do with placating the hundreds of VMware partner who, frankly, are panicking about now that VMware will start to directly compete with its own ecosystem. VMware is always going to be able to offer higher efficiencies in a public cloud than it’s partner ecosystem can do using VMware tools – as such they introduce some fragility into the entire channel apparatus. All of a sudden that idea of offloading the cloudy assets in an attempt to shore up ongoing revenue of core products has been hit a double whammy – the company doesn’t get to bask in the glory that initiative like Cloud Foundry brings, but neither does it enjoy extreme channel buy in that a single minded strategy would deliver.

All of a sudden Gelsinger’s comments about AWS can be seen in their true light – he doesn’t care a fig where workloads are placed – so long as wherever they’re places they’re sitting on VMware solutions – chalk that one up to pure self-interest. But the economics of a VMware fueled public cloud are problematic, as one commentator said on Twitter:

I did a cost analysis for a big [integrator] last week – VMware is $6 per GB ram per month – adds about 30% in some cases to price – and do you think customers care what the hypervisor is? I can’t see how VMware’s core revenue maintains in any shape or form

Predicting What it Means for Pivotal

This entire setup is strange. We now have three operations with significant areas of overlap and sales channels that, to an extent at least, overlap and compete. We have VMware that on the one hand is exiting its Open Source and high-stack offerings, while at the same time is starting to talk a public cloud story. We have Pivotal trying to go it alone, but still having to pursue a strategy that sees it embrace choice and heterogeneity – traits that are toxic to EMC and VMware. The pivotal setup is interesting, but still too confusing – if VMware had put Horizon in with the mix, and not strayed into the public cloud arena it would make more sense. Similarly there was an opportunity to leverage DynamicOps to really deliver a heterogeneous hybrid cloud offering but, again, this is only tenable in a standalone entity. As I said back in July:

(VMware needs to) Create a standalone entity that is focused on cloud infrastructure – include DynamicOps in with this and direct the existing infrastructure business to “play nicely” with the new unit. The new business “VMCloud?” can  be a stellar vendor of public and private cloud technologies and the DynamicOps inclusion can mean that the products have a compelling story when seen alongside customer use of existing VMware products and services. In doing so VMware disrupts the IT management and service provision space before other players (BMC anyone?) are able to do so themselves

Summary

God, what a tangled mess we weave…. In fairness, the CEO’s of EMC, VMware and now Pivotal have a very difficult job They need to disentangle a vast array of solutions, competing go to market strategies and differing revenue streams, at the mean time keeping up momentum and a positive partner vibe. That’s not an easy task. While there are things I would have done differently in this Pivotal move, overall it’s an interesting play. If these companies can just resolve the issues around channel conflict, this could be an interesting situation to watch.

Oracle announced it has agreed to acquire Nimbula, a provider of private cloud infrastructure management software. Nimbula’s technology helps companies manage infrastructure resources to deliver service, quality and availability, as well as workloads in private and hybrid cloud environments. Nimbula’s product is complementary to Oracle, and is expected to be integrated with Oracle’s cloud offerings. The transaction is expected to close in the first half of 2013.

Yes, the home of closed, proprietary, heavy and aggressive technology has acquired a lightweight open source product that is part of a “mutual gain” initiative. OMFG indeed!

MyPOV

Nimbula’s decision to join OpenStack was, in my view, partly inspired by the difficulty they had in gaining real marketplace traction. Other than Russian search giant Yandex, the enterprise deals were slow in coming to Nimbula – this despite the fact that their solution was actually very solid (no surprise given the founders of Nimbula built Amazon’s EC2 service in the past. By joining OpenStack, Nimbula got some much needed attention and recognition.

This recognition however may not have turned into significant revenue and I suspect the investors were starting to itch for a short term exit before cash levels started to get painful – fundamentally in the private cloud market it’s hard to convince people not to just “keep buying VMware”

In acquiring Nimbula, Oracle gets some private cloud chops, the interesting thing will be to see whether they continue the approach of deploying Nimbula on commodity hardware or try and fudge it into their long standing and frustrating “cloud in a box” story. Much is unknown about the deal (not least of all the price paid) but it will be fascinating to see where it goes.

Disclosure – Nimbula’s VP of Marketing is an investor in a company that I’m also an investor in. I have no inside information on the Nimbula/Oracle deal however.