Maintaining a piece of hardware is 4/5 times more costly than actually acquiring it. This will, of course, come as little surprise to CIOs and IT departments the world over who, particularly in times of economic stress, are focusing attention on reducing operational cost.
Consolidating any line-of-business process has a role to play in potentially delivering infrastructure and human capital savings. However, consolidation can also encourage greater visibility and tighter control, assure more consistent and efficient workflows, and generally make the department or application easier to manage.
What consolidation does not do is solve the problem in the minds of many IT professionals today - that they are simply running out of space. Neither does it necessarily reduce OpEx or CapEx. Consolidating a host of application servers from branch offices is a major job, one that could very well cause application performance to drop away. Security can suffer too. Far from delivering an immediate solution, consolidation and virtualisation have brought more challenges.
It is precisely this issue of adding kit and complexity that is pushing expected return on investment (ROI) dates back, and requires businesses to embark on a journey to consolidate which begins with the outcome and travels backwards. This outcome must be excellent user experience and application accessibility – irrespective of its hosted location. The true mark of success is a consolidation programme that has no impact (other than positive) on the user experience.
Pitfalls and Potential
In the very best of scenarios, the application is constantly available – whether from a consolidated or virtualised host. Costs have been reduced through rationalisation of kit, human capital costs have fallen and management of the server and application portfolio has been simplified.
In the very worst case, the reverse is true. Costs have risen, there’s more kit ‘optimising’ the network. It’s more complex and requires additional management hours. And at the end of the process, the user knows something has changed because they can no longer access their application as reliably as they once did.
Going into the Cloud
Moving to the cloud to avoid the cost of re-distributing access is an option. Crucially, this model also provides an unprecedented level of service flexibility – enabling capacity to grow and shrink on demand in a way not possible with manual systems.
Cloud’s automation and orchestration, underpinned by a high level of intelligence at the application layer, is absolutely critical to data consolidation strategies. For example, if a user makes a request of an application at a local data centre, rather than saying ‘no’ because the data centre has reached its capacity, this request becomes the catalyst for a whole set of processes to build and deploy a virtual machine into the network, and begin routing users to it.
It is the automation piece that triggers the creation of the virtual machine while orchestration ties all the automated pieces together with, for example, an application delivery controller that is told the machine is ready to be serviced and users can be directed to it.
So, rather than changing how the data centre will look, the change will be how it is put together; with application intelligence at its very core.
Virtualising the Data Centre
Centralising control over the data centre has as much to do with security and continuity as it does with application performance and accessibility. Today, many enterprises build redundant sites as a backup and manually manage data replication and failover to the secondary site when needed. It is, in essence, an insurance policy. It’s also a non-performing asset.
By virtualising data centre resources at both sites, it’s possible to turn these non-performers into ongoing available assets that will function in a distributed scenario to achieve maximum reliability and performance, regardless of location. For example, in an active-active data centre configuration, it’s possible to carry out data replication, upgrades and maintenance on a more-frequent basis, increasing overall uptime and decreasing time-to-market for services.
So, with high availability achieved, activity can be refocused on getting the most from the data centre virtualisation effort. And here, IT must look to the layer between the switches and servers, where proxy, cache and SSL assets abide. This layer can account for up to a third of the cost and complexity that data centre teams must manage. Getting virtualisation right is critical.
The problem is that, despite consolidating servers into a blade environment, the hardware itself is often not virtualised. This means the density of virtual machines per blade is not as high as planned. Indeed it’s not uncommon to have a single virtual machine running on a single blade, when the hardware in capable of supporting multiple machines.
This lack of dynamic movement and poor density is one of the major reasons why expected ROI from consolidation programmes is being pushed back.
The key is to virtualise this environment too - encouraging best practice and enabling further cost reduction right across the environment.
Referencing Success
So what makes an effective data consolidation programme? As we know, consolidating your data centre estate for the sake of short or medium term ROI is a mistake.
The complexities, whether it be moving servers into a central cloud location without considering the impact of application performance, or failing to appreciate the man-hours this demands, have resulted in much-delayed returns.
The crucial element is to focus on connecting users to the application rather than computers to clouds and data centres. It is the application, and how it is used, that provides the true value to the business, as well as how easy it is to scale this application to suit users – something which cloud infrastructure can excel at.
Moving away from a network-centric view is key. Traditional networking has no concept of the application. It will know what a port is, but it will not have the intelligence to recognise the application users are connecting to. Consider Port 80, traditionally used for handling web traffic - it’s now initiating calls to thousands of applications, but lacks the intelligence to discern the characteristics of each and every one. And this intelligence is now critical to assure optimum performance.
Owen Cole is EMEA Technical Director, F5
Post new Comment