IBM IaaS goes global

IBM's $2 billion acquisition of SoftLayer last summer has given IBM a highly competitive cloud services delivery infrastructure platform, one architecturally capable of competing with Amazon, Microsoft and others. Now it is massively (and...


IBM's $2 billion acquisition of SoftLayer last summer has given IBM a highly competitive cloud services delivery infrastructure platform, one architecturally capable of competing with Amazon, Microsoft and others.

Now it is massively (and globally) building on that foundation with plans to dramatically scale up the SoftLayer global footprint, committing $1.2 billion to more than double the number of SoftLayer data centres by the end of 2014 (from 13 to 28).

It is important to note that IBM is not pouring any concrete or doing extensive remodeling of existing IBM datacentre facilities. Rather, it is taking advantage of large-block contracts for datacentre space, power, and network capacity from data facilities operators like Digital Realty Trust (a current Softlayers provider). This is one key element in enabling such a rapid expansion of IT assets.

The other element is adoption of a highly modular datacentre build and deployment model. IBM's aggressive datacentre expansion/interconnection program not only provides additional capacity for current and planned SoftLayer services but also creates an immediate point of differentiation with Amazon Web Services and other hyperscale IaaS providers. The latter's current cloud architectures feature relatively large regions and availability zones while IBM's more distributed and modular datacentre approach, which will drive the deployment of tens of thousands of servers in the coming year, features multiple country-level deployments supporting local in-country (as well as globally distributed) implementation. The expanded footprint gives IBM customers the option of leveraging "local" cloud datacentres to accommodate data residency, compliance, security, and control/visibility concerns.

The acquisition of SoftLayer, this accelerated datacentre expansion, and IBM datacentre integration onto SoftLayer's backbone network represent a significant stake in the ground for IBM and its cloud ambitions. SoftLayer brings its homegrown Infrastructure Management System (IMS) cloud platform, providing the automation and orchestration necessary for cloud-scale operations and single-tenant/multi-tenant deployment options, as well as a demanding Internet-centric customer base (e.g., gaming companies) whose requirements for uptime, performance, and low-latency often exceed those of enterprises. The combination of the SoftLayer assets and IBM's deep pockets, middleware/SaaS solutions, managed services capabilities, and enterprise customer base set the stage for an enterprise-oriented Web-scale cloud platform.

The pace and scale of IBM's planned expansion of its Softlayer datacentre assets also provides some important insights into one of the most discussed developments in the IT infrastructure marketplace, the Software Defined Data Centre (SDDC). From the perspective of datacentre operators, SDDC is not a product, or even a suite of products. It is an emerging datacentre operating model; one that is increasingly at the core of creating a successful and sustainable cloud service business model.

The emerging SDDC operating model is based on three major characteristics:

  • Deployment of highly homogenous (though not necessarily commoditised) IT hardware assets in modularised pools
  • Implementation of a highly automated service creation and resource management framework for rapid provisioning and reconfiguration of virtualised and (increasingly) bare metal compute, storage, and network resources
  • Availability of a secure, reliable, and flexible network backbone to provide low friction connectivity between a large numbers of identical SDDC datacentres dispersed across metro, national, and global locations

Each of these characteristics has major implications for how operators of SDDC datacentres such as Softlayer approach IT hardware acquisition, deployment, and migration as well as where they will make investments in purchasing versus developing software. For example, it is interesting to note that IBM completed its acquisition of a high speed data movement software company, Aspera, which would help enhance Softlayer's inter-datacentre interconnect facilities.

Some of the most fundamental changes are in how an SDDC approach affects IT hardware choices. As more service providers and enterprises adopt SDDC operating models for some, and eventual nearly all their IT assets, three critical changes to anticipate are:

  • Accelerating purchase and deployment of IT hardware (compute, memory, storage, I/O network) in pre-defined, highly standardised, and pre-modularised bundles (current converged systems bundles represent the lower end of this modularisation curve) to dramatically reduce data centre launch and expansion time windows
  • Growing expectations that the cost of migrating (decommissioning) IT hardware assets will trend towards zero (essentially falling to the hourly cost of staff that unplugs and rolls out the old modules, and then rolls in the new ones)
  • Closer matching to the component (primarily compute, but also storage) technology enhancement cycle as SDDC operators adhere closely to rigid 3 year technology refresh cycles for hardware assets (with the possible exception of assets used for long term archiving services)

These changes also has interesting implications for how most organisations, those SaaS and end customers not opting to adopt an SDDC operating model themselves, will gain access to the latest IT component technology. SDDC-based datacentres will be able to much more rapidly (within in weeks versus months) incorporate new processor, memory, storage, and network technologies into their environments. They will also be able to quickly, and cost effectively end of life older, inefficient technologies. IDC predicts that that the impact of this transition will emerge by the end of 2014/beginning of 2015 and will become quite apparent in 2017 as the installed base of hardware assets reach the end-of-life phase.

Developers and IT organisations that have business models best served by quick access to the latest technology will increasingly turn to cloud service providers like IBM/Softlayer that fully adopt SDDC operating models versus traditional OEM equipment suppliers and channel partners, like IBM's STG business unit.

Many factors influenced IBM's decision to sell its System X business to lenovo, but this development will be seen as a earlty indicator of this change. Managing this transition will be a critical challenge for IBM, its traditional OEM competitors, and the wide array of current cloud services providers. It will also have implications for financing providers that are faced with both rapid deployments and assets with limited end of life value. For IT leasing and financing providers it's important to build the impact of SDDC models into their portfolio planning. Also, recycling and refurbishment operators will need to understand how to refocus their efforts for this new acquisition strategy.

Posted by Rick Villars, Vice President, Storage and IT Executive Strategies, IDC

"Recommended For You"

IBM tackles datacentre energy problem Uptime's advice to datacentre owners: involve operations from the start