IBM’s $1 billion commitment to take its entire software portfolio into the cloud and charge for it monthly will appeal to many of its existing enterprise data center customers, including broadcasters and pay TV operators. Broadcasting and content distribution have increasingly become IT driven and subject to the same economic and logistical considerations as other enterprise computing applications, although with specific infrastructure and security requirements.
IBM has cloud customers in this category, such as TP Vision, the Amsterdam-based joint venture between Philips and Chinese TV manufacturer TPV responsible for the development of Philips’ Smart TV. IBM provides TP Vision with cloud IT capacity services on a pay per use basis for its smart TV based OTT services available in 30 countries in Europe, Latin America and Asia Pacific.
At the same time some major content distributors and pay TV operators are exploiting their infrastructures to provide flexible cloud services themselves and on this front are partnering with major cloud technology and capacity vendors such as IBM. On this front, France’s leading MSO Numericable announced a partnership with IBM early in February 2014 to launch a cloud-based service aimed at business customers, following in the footsteps of other cable operators such as Virgin Media in the UK. The deal was to jointly develop an Infrastructure-as-a-Service offering to enterprises via Numericable’s Completel unit, exploiting some of the technology and software developments just announced by IBM.
The $1 billion announcement was made at IBM’s Pulse 2014 cloud conference held in Las Vegas this week. For IBM itself it represents an urgent, some would say desperate, attempt to revive its recently flagging fortunes by belatedly matching Amazon, the runaway market leader in provision of elastic cloud services on a pay per use model. IBM can draw some consolidation here in that HP, its major rival in the enterprise data center, has been hit if anything harder by the recent decline in the big systems market driven by the migration towards the cloud.
HP’s 2013 revenues were down 5% after allowing for currency movements at $112.3 billion while IBM’s fell 2% to $99.8 billion on the same measure. Both have responded with job cuts but HP’s are more drastic, shedding 27,000 jobs or 8% of its workforce during 2014, while IBM has announced 15,000 losses, or 3.4% of the workforce.
IBM was once the king of big iron as the world’s dominant manufacturer of mainframe computers for enterprises, but after the enterprise world made increased use of smaller systems, based more on the client/server model accessed via PCs, the company was on the verge of bankruptcy in the early 1990s after losing $5 billion in one year. IBM was rescued under CEO Lou Gerstner with a reinvention of the mainframe as a more flexible machine enabling virtualization to consolidate proliferating servers that were incurring rising management costs, while also migrating away from being hardware centric to focus increasingly on software and services.
This reflected the course of enterprise computing and the restructuring was largely successful, overseen most recently by current CEO Virginia Rometty, to the extent that software was IBM’s second largest business in 2013 and the most profitable by some distance, generating $25.9 billion of that $99.8 billion in revenue, with margins of 89% compared to 48.6% overall. At the same time software has become pivotal to IBM’s $18.4 billion services business.
But 2013 was also the year revenues started to decline, even from software, driven by the fundamental change spearheaded by Amazon but also pursued by the likes of Google, Microsoft and Cisco, from a capital purchase or licensing model to renting or paying by usage. For IBM and HP this means shifting wholesale to a cloud model in order to return to revenue growth, which both are frantically doing in slightly different ways.
While on the surface both HP and IBM face an uphill struggle to dent Amazon’s momentum and start closing the gap in terms of public ‘pay as you use’ cloud services, both have some advantages. The main one is that both have large bases of enterprises with data centers that currently are expensive to run and consume large amounts of power. IBM and HP are well placed to help these customers migrate to a hybrid cloud model, where the in-house data center is slimmed down to its bare essentials, reducing ongoing support costs and power consumption. This involves migrating in the first instance less critical applications such as data archiving to the cloud, as many have done with Amazon’s Simple Storage Service (Amazon S3), while retaining more critical applications using operational customer data in house, with greater perceived security.
This is where HP and IBM feel they can score by starting to migrate critical applications and data to the cloud, under what is effectively a three way hybrid model. The three parts are remaining in-house IT operations, private cloud services hosted by the provider via this emerging PaaS (Platform as a Service) model, and thirdly fully commoditized public services like Amazon S3.
HP has been moving that way with the latest version of its Virtual Private Cloud (VPC) unveiled in January with the aim of enabling enterprises to exploit the economics and scalability of a public cloud, while retaining the security and control of a private cloud. HP’s latest Managed VPC allows customers to choose between virtual or physical configurations, multiple tiers of storage, hypervisors and network connectivity types. Previously private clouds tended to involve physical configurations run by the third party in more like a traditional outsourcing arrangement.
Until now HP’s cloud strategy has been largely tactical, delivering services and features almost on a drip feed basis as and when they become available, coaxing customers along gradually towards an increasingly external virtualized platform. IBM by contrast has taken a more strategic approach, presenting cloud as a long term solution for major enterprises, playing on its greater strengths in R&D and associated technologies designed to make innovative use of Big Data.
A key part of this strategy revolved around a new initiative launched April 2012, which was called IBM Expert Integrated Systems and then IBM Pure Systems, in a move over hyped by some analysts as being the company’s most significant announcement since the inaugural mainframe launch half a century ago. Cynics on the other hand derided it as just another hardware launch, as Pure Systems was a line of factory pre-configured servers that did not on the face of it look that radical.
The key point was that they incorporated software called the IBM Flex System Manager combined with so-called “Patterns of Expertise” for automated configuration and management, which together have become an important plank of the company’s hybrid cloud infrastructure for management “out of the box’.
The initiative also aimed to capitalize on IBM’s R&D around intelligent software designed to learn and adapt, so that as well as automated set up the resulting infrastructure would continue to reconfigure itself in response to usage patterns in order to optimize storage, CPU and bandwidth costs, while reducing administrative overhead and accelerating time to market for new services and applications.
But while IBM continues to roll out the produce of its R&D as appropriate, to some extent this strategy has been overtaken by events and so the company has had to become more tactical and converge towards HP’s approach. There was little mention of ‘Pure’ or ‘Expert Integrated’ systems at Pulse 2014 in the context of the cloud investment, which to an extend has had to be cobbled together quite hastily.
For one thing IBM is actually selling some of the core hardware activities that had been folded into Pure Systems, but are now being divested in order to generate the cash for the accelerated push into hybrid cloud. IBM announced in January 2014 that it was selling its x86 Business including server blades to Lenovo for $2.3 billion. This came nine years after Lenovo bought IBM’s personal computer business including the ThinkPad brand for $1.25 billion and turned it into the world’s leading PC business by revenue by 2013.
It was IBM’s failure to become a market leader in x86 servers that contributed to the decision to sell, just as was the case with PCs. In EMEA, IBM was the third largest maker of x86 servers after HP and Dell, with a market share of 13% and revenue of around $900 million in the first three quarters of 2013. This was not enough to create the economies of scale IBM now demands in hardware, with gross margins in the region 15% to 25%, compared to 40% to 50% for higher end servers, where IBM is still in the game.
However the sale disrupts IBM’s Pure Systems since x86 servers and blades were a key part of that, which of course by the same token means that Lenovo gets a leg up into some of the larger IBM accounts. It also begs the question of how Lenovo and IBM will work together to manage the ongoing transition from traditional UNIX servers to denser blade based configurations, given that some of these of these, based on the Power architecture, will still be supplied by IBM.
Meanwhile IBM contends that the sale of the x86 division is part of the drift away from being a supplier of proprietary hardware towards being a provider of cloud services that run on commodity hardware from a variety of manufacturers. This is likely to involve close cooperation with Lenovo in a number of accounts for the short term at least.
IBM has already committed more cash than it raised from Lenovo to various projects around hybrid cloud and will spend a great deal more. IBM prepared the ground for this renewed dash for cloud in June 2013 with the $2 billion acquisition of Dallas based SoftLayer Technologies, whose software supports rental of hardware and services capacity such as servers and networking equipment, with support for hosting via this model of services such as database or website management. Widely used by startups for renting capacity on a scalable basis during their early growth, SoftLayer enabled IBM to compete with Amazon in the smaller enterprise cloud sector. The company was founded in 2005 and has about 21,000 customers supported via 13 data centers in the US, Europe and Asia.
SoftLayer is an integral part of IBM’s new PaaS cloud offering, which is focused on what has become an important aspect of competition in the cloud field, attracting developers. As part of this initiative, IBM is making its whole enterprise software portfolio available on a rental basis from the cloud and launching an open beta codenamed BlueMix, which incorporates a platform called DevOps for supporting software development in the cloud. DevOps supports planning, development, testing, deployment and monitoring of applications, while turning systems management into a service that can be delivered from the cloud.
As part of BlueMix IBM is bringing its middleware portfolio to SoftLayer through these pre-defined software “patterns” designed to enable existing applications to be ported to the cloud. IBM regards this as a differentiator against Amazon, since there are more than 200 of these pre-defined software ‘patterns’ that enable existing enterprise applications to be ported readily to a hybrid cloud environment.
Another component was added at the last hour through the acquisition of Cloudant, which is described as a database-as-a-service company, enhancing the SoftLayer capabilities in that area. Cloudant already has a rival hybrid cloud provider, Rackspace, not just as a client but also an original backer. The need to appeal to developers has become essential for all cloud platform vendors aiming to compete with Amazon. They need to assemble a full stack of components that will help clients in their transition to a modern energy efficient dispersed data center, emulating Amazon’s Web Services in creating an API driven model for business computing. To do this the vendors need to draw a sufficient mass of leading developers into their platform to provide essential parts of this overall hybrid cloud infrastructure. Even IBM cannot do this all itself, but does believe it has the edge in being able to support the transition of its existing datacenter customers at least.