Blades are an important platform for the future of the datacentre, argues Daniel Fleischer, senior research analyst at IDC
In 2005, the server blade market topped the US$2 billion mark for the first time, growing by 84 per cent over 2004. Server blades remain the growth area in the server market and remain a key issue for IT and datacentre managers in most surveys of end users. Today we have reached a juncture in the adoption of server blade technologies. Most early adopters have now deployed the blade form-factor and the challenge now lies in how to push beyond these users into new verticals, application types and also into the SMB sector. For this to occur, there are many problems that still need to be resolved in order to demonstrate the blade value proposition and therefore stimulate adoption in these new user groups.
IT simplification is the driving force behind most IT purchases in the enterprise sector today. System volumes in the datacentre exploded in-line with scale-out strategies, with operational expenditure (OpEx) increasing accordingly. Reducing OpEx is a key focus of both users and vendors looking to release more IT budget. Server blades are positioned to play a key role here in the future, being a product designed with management in mind from the ground up. That said we are not quite there today, we still need better management tools, and vendors of server blades have certainly picked up on this.
Delivering the promise
Management is key for the blade to deliver on its promise. This is where vendor offerings will be differentiated. Automation has remained one of the key value propositions and this remains the key method for reducing OpEx. The ability to manage multiple systems at the click of a button is certainly compelling, and combining this with virtualisation and redundant pools of server blades makes for a powerful and compelling story. Users do not want to have to learn and use multiple management consoles with differing characteristics and terminologies to perform simple management operations.
Today’s blade deployments remain project-based, focused around a number of key verticals within large enterprises. In most cases we see blade infrastructure, such as the chassis, pre-deployed in order to allow rapid scaling as and when required. As such we expect continued growth as the second refresh cycle picks up steam. We anticipate the replacement of traditional rack-optimised systems when the value proposition is proven in working environments.
Ultimately new infrastructure purchases will be application driven as users look towards IT to solve business issues. With this in mind it is essential for us to create an ecosystem around the blade form-factor. Certainly in Europe we are seeing vendors partner with ISVs to develop true utility computing solutions that are gaining traction in the market, although these solutions remain application specific at present. Deploying proven solutions of this type will ensure maximum possible ROI and this should not be limited to the top ISVs but will also include smaller players too. Mergers and acquisitions have already played a role in ensuring that third-party, best of breed blade technologies are integrated into tier-one blade vendor products. What users are looking for is a unified approach.
Standard solutions
Standardisation also plays a major role in the server blade ecosystem. Customers are looking for industry standard solutions, solutions that take into account existing heterogeneous environments where forklift upgrades are not a viable option – solutions that can mix old with new in multi-vendor offerings, and which allow the integration of best of breed networking technologies. We do not expect to see, for example, standards-based blade chassis infrastructure in the near future but this does not mean that you cannot house multi-vendor blade/chassis offerings alongside each other, provided standardised management is in place. We do see a sustained trend towards industry standards-based solutions, with a migration from proprietary legacy-based RISC and Unix servers onto industry-standard x86 systems. Again, blades make an excellent platform on which to fulfil such projects.
Price point remains an issue, however. We now have to sell to new budget holders such as CFOs, where ROI becomes a more important metric. Investment protection remains a major factor as most projects rollout over a four- to five-year lifecycle. Business pain points are driving IT demand. Pricing alone is insufficient and vendors will differentiate at the management level, with blades being the essential technology platform upon which this differentiation will be delivered.
Peripherals represent a further problem. One needs to monitor the blades remotely, presumably through a KVM (keyboard-video-mouse) system. One also needs to share storage, presumably through a SAN, to reduce overall costs. As usual, the reality is more complex than the idea. And scalability is an issue, since most KVM systems can handle only a few devices.
Power and cooling issues are presenting blade users with significant problems. Today’s datacentres have not been built to take into account the increasing number of processors per U. Again, it becomes a question of economics – is it cheaper to retrofit power and cooling infrastructure or simply to build new datacentre facilities? In any case, all new datacentre facilities should be built with high-density blades in mind. This issue is increasing in prominence as we speak with large enterprises. SMBs, where server rooms are more commonplace than datacentres, have not even begun to think of this as a problem, or of the associated solution costs. Ever increasing energy costs have had a major impact in datacentre cost models.
Corporate-level management continues to put great pressure on datacentres to cut costs. Indeed we do not expect IT budgets to reach the levels or flexibility to IT managers that were afforded before the collapse of the dot.com bubble. OpEx now accounts for around two thirds of total server spend making a logical area to focus on for cost-reduction. Consolidation has been the solution of choice, reducing both the physical number of systems and management applications. Looking forward, virtualisation will become tightly integrated in the x86 space, nowhere more so than in the blade market where there is a perfect fit allowing systems to assign servers in small modular units, as users look to deploy dynamic IT solutions. Again the migration from RISC/Unix environments will accelerate this, as virtualisation is already a mature and accepted technology in this space. The future for blades is bright despite the many challenges. IDC forecasts the server blade market to reach $9.6 billion by 2009 representing a compound annual growth rate of over 350 per cent.
