The Politics of Converged Infrastructure

Politics Highlights

• Software Defined x sits atop Converged Infrastructure including Bring Your Own y – separated by the hypervisor
• Data centre outsourcing saw a conflict between CFO and CIO
• Cloud Computing saw a conflict between LOB and CIO
• Converged Infrastructure adds a potential battle between 4 distinct data centre constituents managed by secret sauce orchestration software or firmware
• Suppliers address a high-level purchasers, but need buy-in from the data centre if possible
• Suppliers will succeed by improving efficiency and reducing internal IT costs
• The success of Converged Infrastructure and Integrated Systems will be limited by internal politics

At the beginning of the last decade data centre outsourcing was all the rage, but the potential savings for customers were often limited both by the inflexibility of most solutions and internal politics. We found at that time that there was a considerable difference of opinion between CFO and CIO constituents. As Cloud Computing has grown similar conflicts of interest have developed between Line Of Business (LOBs) – often the first to adopt SaaS offerings such as salesforce.com – and the CIO. In general terms:

• The CFO wants to limit unrestricted spending and keep things within budget
• The LOBs want fast application development, production agility and strong links between their business objectives and the technology used to serve them, while
• The CIO wants to serve the organisation effectively, addressing the requirements of legislation, adopting relevant new technology, recruiting and retaining key specialist staff and an appropriate budget.

We expect an even greater conflict of interest within the data centre as users consider the adoption of Converged Infrastructure and Integrated systems. We’re currently in the process of interviewing a number of these organisations and don’t want to pre-empt our findings: however it is appropriate to consider where they’re likely to occur and how to minimise them.

Software Defined x, Converged Infrastructure And Bring Your Own y

We’ve drawn a picture of a generic ‘data centre of the future’ in Figure 1. There are essentially 2 sides – separated by the hypervisor layer the North can be loosely described as ‘Software Defined …’ and the South, ‘Converged Infrastructure’. We’ve put virtual machines and the systems management of virtualised elements in the former; server, storage, networking (including ‘Bring Your Own’ equipment) – in the latter. The savings proposed by Converged Infrastructure suppliers such as Cisco, HP, IBM, Oracle, Fujitsu, HDS, Huawei and others are based on building integrated systems and/or the automated ‘secret sauce’ software to orchestrate how they work together. They’re using their deep expertise in ‘what works with what’ (dare we say W⁴?) to improve overall system performance, time to deployment, power and cooling costs and business agility. In addition they also offer ‘one throat to choke’, which helps the user reduce the time and money spent on supplier management.
Of course each supplier version comes with its own flavour: Cisco builds out from the Network space, HP from industry standard x86 servers, IBM from a mix of mainframe, Power, x86 servers and its Tivoli software, HDS from its storage heritage, VCE by combining EMC storage and Cisco networking with VMware’s dominant North-side virtualisation, etc. All of them are moving up a layer of management within the user organisation to propose their new offerings.

The User Constituents Of Converged Infrastructure

We show the user view of Converged Infrastructure in Figure 2. It shows a number of distinct constituents, each typically with its own staff and budget. In particular:

• Application Developers and LOB – working to provide technical support for business process, these employees are helped by the standardisation of the platforms they work on and the ability to write and test applications in virtual containers before putting them into production; they are users of PaaS from Cloud service suppliers; over time their need for technical knowledge of the underlying technology is diminishing and many are now part of the LOB, rather than data centre staff; this constituency is likely to grow if users managed to shift their spending from infrastructure maintenance to business innovation
• Logical Infrastructure and Virtualisation Admin – this group has grown with the increased deployment of Enterprise virtualisation; their tools are systems management suites such as VMware’s vCenter or Microsoft’s Systems Center; they are effectively the providers of IaaS to the application developers, parcelling and virtualising up the underlying infrastructure elements; this constituency is likely to grow as a controlling element of a Converged Infrastructure
• Server, Storage and Network Admin – data centres today are often fragmented, using a number of solutions introduced in different years and offerings requiring specialist knowledge and vendor-specific management tools; a strong feature of ‘horizontal integration’ was the use of different supplier solutions in each layer; these employees are involved in deployment of new products and upgrades, ensuring business continuity, protecting the physical environment and in some cases moving tape in and out; the number of these roles within an organisation is a function of its size and technology fragmentation; the deployment of Integrated Systems is likely to have a negative effect on employment here, with one admin type (server, storage or network) retained and the expertise of others largely replaced by a vendor’s autonomic management software; an alternative will be for each element to stretch from South to North sides with the use of storage hypervisors, Software Defined Networking and the like, with the integration of the ‘platform’ done at a higher level
• Physical Infrastructure, Maintenance and Facilities Admin – all data centres have staff looking after the physical environment and there has been investment in improving efficiency as electricity prices soar and governments introduce carbon taxes; 15 years ago a number of suppliers talked up the advantages to users of combining facilities and IT management budgets, but – despite making significant savings themselves by doing so – they have largely failed to persuade their customers to do the same; the introduction of Converged Infrastructure is likely to have a neutral effect on this constituency, since new systems management tools expanding from the management of IT to include power and cooling

We note that there may be important generational issues here, because the South side staff tends to be older and more experienced than the North side ones.
Vendors use different arguments to sell at different levels within an organisation: talking about improving server utilisation or the number of TBs manageable by each storage admin is designed for a different audience than the need to improve the balance between infrastructure maintenance and business innovation. Selling Converged Infrastructure to large organisations will involve selling to the CFO, CEO and CIO – especially where the ‘I’ stands for Infrastructure rather than Information. They need to be careful with the South side constituents when making new claims, proving that their autonomic software can reduce the leg work and justifying what they used to sell if its taken them this long to come up with systems which are properly integrated.

Some Conclusions – Reducing Internal IT Staff Spending Is The Nub Of The Problem

A number of suppliers euphemistically talk about ‘redeploying’ staff from technical to business roles, but we shouldn’t be taken in – the adoption of Converged Infrastructure is about firing people and saving costs as a result. We expect declines in business ITC spending by all company sizes in 2013. In particular:

• Large companies (those with over 1k employees) will spend 1.0% less ($1.4 trillion)
• Medium (those with between 100 and 1k employees) 2% less ($503 billion) and
• Small (those with less than 100 employees) 4.8% less at $1.5 trillion.

However the biggest savings will be made by reducing, or at least controlling, internal staffing costs, which we estimate as 47.4% of current data centre spending.
The supplier politics are about comparative performance and cost, their credibility and experience and the use of standard interfaces (such as Open Stack), ease of adoption, etc. The user politics will include infighting between the constituents we identify. Business process re-engineering, data centre outsourcing, Cloud Computing each failed to reach their full potential for vendors as many users were unable to make the necessary internal changes. We also expect internal politics to slow – and sometimes put barriers in the way – of the success of Converged Infrastructure sales.
If you’re a supplier, why not sponsor our multi-client study in this area? We’re including 30 interviews with users and hosters deploying Converged Infrastructure, allowing you to learn much more about the barriers and drivers of adoption.

If you’re a user, let us know your thoughts on the internal changes brought about by Converged Infrastructure deployment