Perspectives & Strategy

When I read that a Unified Data Centre Infrastructure can reduce costs and improve the management of the data centre then I was certainly interested to find out more. Solutions are available from at least two vendors, Cisco and Brocade. One thing that is always important to identify are the expected benefits and potential risks associated with any new technology.

According to “Why ‘Unified’ Is The Hot New Idea For Data Centers” by George Crump, published in Information Week on March 16th 2009 the benefits and risks associated with the approach are:

Benefits:

• The Unified Infrastructure will eliminate the many distinctions particularly in networking, storage, and computing bringing a more cost effective IT organisation.
• IT should be more responsive to business organisational needs to change the IT infrastructure as it will be managed through a single console.
• Delivers a more agile computing infrastructure at a lower cost to bring in IT cost savings.

Risks:

• The amount of change required is considerable and it is important to manage the risk of dissatisfaction as more IT staff worry about their jobs.
• An ability to deliver on ROI without disrupting existing operations while implementing the new architecture.

Organizations increasingly rely on IT to help enable, and even change, their business strategies, they need their IT infrastructure to be more powerful, agile, and cost effective than ever. This is the key argument in favour of implementation. However to implement any significant infrastructure change I think there will have to be a very strong business case and a positive ROI in a matter of a few months. My fear with such a project is that the costs associated with the new unified architecture can outrun the projected savings if the project is not tightly managed. From the perspective of corporate management a project of this type may be seen and only giving an advantage to IT, something that is currently low on their spending priorities at the current time.

One of the challenges faced has been availability of Real-time infrastructure monitoring tools can help data center management to predict faults, and to respond quickly to any faults that do occur. Over 95% of all IT managers have a keen interest in the performance of their data centres. Whilst there is is a large amount of monitoring many in IT feel that the tools that use are inadequate for the job.

In the average Data Centre the server landscape is becoming more complex particularly with the implementation of multi-core computing and virtualization. There is an ever present  need for higher-bandwidth, lower-latency, increasingly dense, rack-mount and blade servers and increased power and cooling demands within the large data center architectures. Virtualization has enabled higher utilisation ratios, but it also brings more complexity in server management. According to Cisco one of the major argument in favour of the Unified Data Centre Infrastructure is that “even modest improvements to total cost of investment (TCO), energy efficiency, and complexity can have a significant cumulative effect on the data centre”. The unified infrastructure can strengthen any argument for virtualization, but also introduces an infrastructure that can bridge any silos. An ability for the IT operations team to prioritise traffic can be a significant bonus.

Can we see a positive cash flow in a matter of a few months? Many elements of the new architecture can reduce the redundancy within the data centre, e.g. running storage and IP traffic through a single unified network interface. Crump suggests that we can reduce the number of cables from 12 to 2. I am intrigued but am mindful that network traffic concerns are part of the reason why multiple interface paths were setup in many data centres in the first place. Of course our infrastructure is always evolving and many of the technologies have rapidly advanced since the data centres were setup. For example the introduction of a 10 gigabit ethernet interface being a key advance.

One further risk that need to be added to Crump’s list is ratification of standards. My concern here is that the concept is currently only supported by two vendors. Many of the corporations I work with are very concerned about standardisation and are unlikely to adopt new technologies ahead of ratification.

I am then led to think about whether this is the only route to resolve many of our data centre problems. The question of whether there are other viable solutions always crosses my mind, none moreso than in this discussion. Many IT managers have been focusing on delivering extra value through virtualization. In most cases this has added complexity to the data centre landscape. Can advanced configuration controls provide a solution? CiRBA and HP offer solutions in this area. I am not looking here to demonstrate the supertirity of one solution over another I am merely wishing to highlight some alternative options.