I’m making my way home after spending a week in San Francisco attending GigaOM’s Structure event. Structure is one of the highlights of the cloud conference calendar – bringing together old friends, industry super stars and exciting customers in a refreshingly intimate two day event. Despite pretty much permanently suffering from conference fatigue, Structure actually left me pretty excited. I sense a real sea-change going on, and it’s a change that is immensely satisfying. Today with fellow GigaOM Pro contributors Paul Miller and David Linthicum, and joined by head of GigaOM Research Jo Maitland, I took part in a mapping session that looked at the multi-cloud opportunity.

First some context. Moderator Paul Miller defined the various terms in the following way:

  • Hybrid cloud is generally taken to mean a mixture of one public cloud resource and one private cloud resource
  • Multi-cloud on the other hand is taken to mean a more complex composite of cloud resources with potentially multiple public and private elements

Thinking about it in the hours after the event, I’m not so convinced by the differentiation between these two terms, rather it seems to me that multi-cloud describes a situation that combines resources from a number of different vendors, this in contrast to, for example, HP’s Converged Cloud notion that includes public and private cloud resources, all from HP themselves. Hybrid Cloud would seem to me to describe a situation where workloads could be moved between public and private clouds (and, by extension, that the various cloud resources shared a common stack). At previous cloud events there has been much conceptual discussion about a possible hybrid cloud situation that would eventuate sometime in the future, and given this portable workload notion it was generally articulated in terms of a cloudbursting opportunity with workloads miraculously moving from private resources to the public cloud as demand spikes dictated. This notion of cloudbursting has long been hotly debated with many suggesting it was little more than a pipe dream – the “unicorns and rainbows” stuff of marketing literature.

Multi-cloud on the other hand is a far more mature and pragmatic approach towards cloud infrastructure and, in my view, directly delivers upon the “composable enterprise” notion that Warner Music group CTO Jonathan Murray spoke of at the event. An organization embracing a multi-cloud strategy will likely have infrastructure resources spanning multiple vendors, potentially public and private, and likely to be across a variety of operating stacks. In the multi-cloud scenario it’s not a shock to see organizations using AWS, some public OpenStack and a private cloud build on one o the stack products.

Indeed two separate organizations at the event spoke openly about their multi-cloud approach. PayPal got on stage and finally delivered some much needed clarity around their previously hotly debated move from VMware to OpenStack. Indeed PayPal confirmed that they are using OpenStack and that around 20% of production workloads sit on the open source product. The interesting thing though is that PayPal articulated an expectation that they would continue to use a mix of VMware and OpenStack products for the foreseeable future – it seemed that different workloads have different requirements met be different solutions – a very pragmatic and broadminded approach indeed. Box was another organization that spoke of the multi cloud opportunity – they currently use AWS extensively but are also in the process of running a proof of concept using OpenStack – again it seems a “horses for courses” attitude prevails.

Naysayers would hold up the example of Netflix that would appear to be completely wedded to AWS. However the reality is somewhat different from this. For a start Netflix uses a significant number of its own tools to augment missing functional areas in the AWS portfolio. Add to this the fact that Netflix has time and again gone on record beseeching other cloud players to up their game and catch up with AWS and it’s pretty clear that a multi-cloud approach would appeal to the company – I believe that Netflix is closely looking at Google’s Compute Engine product lineup to see what is fit for its use.

In our multi-cloud mapping session, there seemed to be a significant amount of support for a “best of breed” approach where customers cherry pick modular solutions that best meet particular needs. True there are some very real issues to deal with in terms of Data Gravity and the impacts it has on the range of choices that are practical, but this best of breed multi-cloud approach would seem to be a very strong enabler of the more modular, more flexible and more contextual IT architecture of the future. The fact that even the infrastructure vendors in the room seems to buy into the notion suggests to me that we’re nearing a welcomed maturity when it comes to the cloud – it is perhaps an acceptance of the fact that not one size fits all and that allowing customers to use the components that best meet their needs will end up being beneficial for all vendors.

Structure 2013 marked something of a watershed – a more mature approach by customers, a more flexible approach by (most) vendors, and a focus on the varied, complex and individual needs of organizations today and into the future. It is indeed a multi-cloud world.

Ben Kepes

Ben Kepes is a technology evangelist, an investor, a commentator and a business adviser. Ben covers the convergence of technology, mobile, ubiquity and agility, all enabled by the Cloud. His areas of interest extend to enterprise software, software integration, financial/accounting software, platforms and infrastructure as well as articulating technology simply for everyday users.

4 Comments
  • Thanks for the nice debrief!

    Interesting pattern I see for VMware & OpenStack and AWS & Google ( for Public ).

    Any more details on ” VMware and OpenStack products for the foreseeable future – it seemed that different workloads have different requirements met be different solutions – a very pragmatic and broadminded approach indeed” ? What work load is favored for VMware for instance?

    When we met at Switch for cloud2020, I thought Box had some presence there? Wondering if thats their OpenStack PoC?

  • @benkepes grt read. multi-cloud which should probably be called multi-vendor cloud is inevitable IMHO. As abstraction and automation evolve and enable true application portability it begets a “brokerage mentality” where cloud resources are procured more on economic than technical terms (insert the entire cloud futures discussion from tuesday). Honestly, I never considered that “hybrid cloud” was constrained to one vendor as you suggested EG with HP being both a customer’s private and public cloud solution. I always considered the “hybrid cloud” as heterogeneous – today it’s looking like Openstack/private AWS/public for now right? — As things continue to evolve I think the abstraction layers will MEDIATE all of this and it’s not going to really be architectural at a certain point – agents will be making decisions and selecting vendors in real=time based on both economic and technical requirements. vendor allegiance for cloud resources at what we currently call the IAAS level will be nonexistent… Cheers! GREAT POST. Josh

  • Nice Post.Multi Cloud has its own tendency complex composite of cloud resources with potentially multiple public and private elements

  • Very nice. I discovered this after my post, which focusses more on composabilty of the cloud applications.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.