Earlier this year we presented our thoughts about the Blade Server Market emergence and its’ potential effects upon distributed computing and data centers. We, at that time, focused upon the blade server market dynamics, O/S selections and decision processes that would define the growth potential for this packaging of computing power and the extensive availability of multiple applications such as CRM, ERP, Data Warehousing and Database Management.

The economic dynamics and technologies have driven the market acceptance even faster than we estimated.  The number of players in this segment has increased and most “name” server competitors have offering providing significant cost savings with the anticipated small footprints. Wireless capabilities have also been incorporated at the Blade level. The availability and efficiency, plus mainstream acceptance of Linux in core corporate applications helped drive the increasing placement rates of flexible blade server systems in the enterprise, SMB, and government sectors. Our prognostication that the qualitative elements of installing and operating under the Linux mantle would be as important if not more important than the realizable out of pocket cost savings when calculating TCOT (total cost of ownership over time) has been validated. Software availability is always a question for sophisticated applications.

With the tightening of IT capital budgets there are increasing volume of viable customer RFQs for “hundreds of blades per installation” with approved budgets versus the high incidence of information gathering requests early last year. This clustering approach has moved well beyond the early stages of market development.  The market is still young but as increased functionality is incorporated in this blade geography, the number of active increased dramatically. As we have indicated continuously, the use of multiple operating systems in this environment – UNIX, Windows, and Linux – is typical even within the same enterprise.

We continue to meet with many of the leading industry participants as well as attending relevant shows but again most importantly with customers, prospects and down to the board level component suppliers to validate our noted trends. This primary research methodology combined with secondary sources provides us with a reasonable and realistic overview of the segment and its ongoing evolution.

We strongly feel we are well along the “second wave” of blade server technologies and availability, i.e., well beyond the early adoption phase. We anticipate additional features and functionality beyond computing capacity toward the incorporation on the board of NIC and Fiber Channel “HBA type” capabilities (there are products available in the market today). The vendor product roadmaps do provide interesting and pre-emptive marketing discussion points although we feel many of the plans are ambitious to incorporate on the limited geography of the blade form factor.  However, as the sophistication of the ASICs increase with density evolutions, roadmap reality may remain one or two “turns” away. We continue to advise our readership/client base in the strongest terms, that they must closely review the sector opportunity to leverage their core competencies. Failure to do so will result in an untenable “catch up” mode for them as laggards while the competitor reaps a significant revenue stream in this vertical hardware and horizontal market application segment.

We currently estimate (conservatively) the blade market to be about 45-55,000 unit installations per month (note: actual installations tend to lag sales somewhat). The growth rate is only tempered by IT department budgetary constraints in this continuing (but improving IT spending) weak economic recovery period. However, we continue to project the potential exists for a multi-billion dollar market in the next few years – our best estimate at this time has been $2.54 Billion worldwide at year end 2005 with a 25-40% growth rate potential through the 2006-7 time frame. This does not include the expected ancillary market for related services that is inexorably evolving and could add an incremental 15-20% to this market in each year. Software sales are difficult to estimate at this time for there is considerable migration from existing “big box servers” to the blade and distributed markets. Grid computing is another factor.

Why continue to consider blades versus standard “iron”?

The primary drivers remain --- COST and TCO – and scalability on demand. The blade server clustering approach offers a proven resolution to:

• Scalability Options – Increase or decrease – Simply add or inactive a blade in the rack to move in the desired direction at minimal cost.
• Provisioning challenges – Automated capabilities simplify global provisioning
• Space constraints and costs  -- Significantly reduces footprint requirements
• Qualified personnel shortages – A few can manage many devices centrally
• Management for multiple sites worldwide – Simplified via point and click remote Web management.

As many have now found in practice, Linux offers further advantages to these stacks of computation power not only in the initial  the web server type applications but now at the work group level. The cost per seat (acquisition, deployment, ease of upgrades, and an ever increasing option of software no longer tethered to one company) is dropping dramatically. At some point, the shadows cast by the legal ownership questions posed in the Linux area will be resolved.

What applications have we seen?

We prefer to continue to view core applications in the following manner for consistency.

1. At the edge as static web servers, group computation nodes; or as distributed utility servers.
2. As front end devices such as ASP hosting, dynamic web servers, cluster computing resources, terminal server nodes, or possibly AV and Media streaming applications.
3. As back end devices such as data base servers, High availability and failover clusters, CRM, ERP, Data warehousing, SCM or mail and messaging servers.

Many others are emerging as the user universe exploits the inherent flexibility and capability of cluster blade servers.  Another issue will explore the emergent applications.

What configurations?

In mid 2003, we wrote that depending upon the configuration of a rack for the blades, a typical minimum configuration of 16 blades utilized the same space as a single standard server box. This seems to continue as a basic practice in our sample universe. Depending upon the design, the servers can be stacked vertically (see below) or in a horizontal chassis with a significant reduction in cabling complexity.  This allows a company can create massive computing centers in minimal space and in highly distributed environments.  This “reach” allows a company to create distributed data centers – mini SANs – at regional offices as an example. This can be an attraction for regional sales offices, distribution centers, or small banking offices.

Rack Mount Blade Servers

Blade densities and performance continue to increase with the evolution of 4 way and 8 way processors. As they evolve in this form factor, the TCO and cost savings in space can be massive when compared to prior operating modes. 

What is the TCO potential?  Is it a reality or merely a “sales” pitch?

Our sample universe feels it has become a reality and in some cases exceeding the initial estimates within their organizations. Where have they realized the cost reductions that add to their bottom line? The impact points are along the lines we indicated in early 2004 and 2005.

• Systems -- The system expands by adding another blade or cluster (conversely, computing power no longer necessary can be easily removed/transferred). The servers are manageable and do provide a simple upgrade path. The networked systems yield the infrastructure and data storage virtualization that is inherent in network attached systems. The scalability – hardware and software – remains a key driver.

• Management and Human Resources – This aspect is mentioned most frequently in our sampling for Web based management simplified and centralized provisioning and management from with fewer skilled resources. The blades are managed locally and remotely as needed, an option that is critical for servicing and upgrades of software and systems. In some cases, the need to add human resources at all locations is eliminated for one or two individuals at the home location can manage multiple locations. The personnel savings is a significant component of the reduced TCO and is a function of the organization size and number of sites.

• High Availability and Failover – In some industry segments, they must take advantage of continuous availability of crucial information resources. This feature is intrinsic in this blade clustering design. The blades are typically hot swappable in case of failure, have trend functionality and diagnostics built in so that when a trend from the defined norm is noted on any device, it can be identified and scheduled for replacement without major operational interruptions. The approach offers dynamic scaling and virtualization for further availability and continuity options.

• Business Continuity/Back-up – Again, as previously outlined last year, business continuity is simplified from both cost and operation perspectives. With a now reasonable (low) cost of creating a data farm and distributed SANs, a complete back-up installation can be installed and managed anyway geographically, as a routine operating mode to react to potential catastrophic failures due to earthquake, hurricane or other forces. The inherent virtualization now provides a simple and reliable operational policy for business continuity. Security, of course, in the information infrastructure is addressed in compliance with corporate policy. Of course some of the more astute blade server vendors are incorporating encryption capability or chips in their offerings.

Summary and Opinions

Our opinions that the blade server approach and clustering mode is an invaluable and highly adaptive infrastructure is unchanged and is significantly strengthened by the latest offerings and user acceptance we see in the marketplace. We feel the evolution pace of feature and functionality will increase for the approach offers (1) continuous and secure operations, (2) applies intelligent and automated management to significantly reduce operational costs, (4) simplified O/S selection with Linux acceptance in critical applications along with its’ security aspects and the (4) yield of dynamic resource optimization for the customer universe.

We continue to strongly urge that those in the computing sector to evaluate expanding their server portfolios consider adding a blade server line [Intel or AMD processor based, operating systems – Windows, UNIX (various flavors), and Linux as soon as feasible to participate in this emergent and viable sector. As always, an new products should incorporate backward and forward compatibility within the installed base and rack systems, be hot swappable, utilize improved GUI and Web manageability, address copper and optical fiber interfaces, multiple connectivity protocol capability or consider embedding connectivity protocols such as  Fibre channel and GigE/IP on the blade at a minimum. The competitive landscape is changing but the keys remain HP, IBM, and SUN along with roles for new entrants.

We feel our client base should focus on how applications can be distributed easily with the flexible computing model with blade servers and grid computing techniques versus the hardware only. Operational requirements - ERP, CRM, SCM and so forth - are only one aspect for consideration. Compliance (HIPPA, Sarbox and others) and data management issues (convergence of Data, Voice, Video) for the every increasing information flows. Now that the market is well beyond critical mass in our opinion, strategic perspectives need to focus upon the application software as well as hardware improvements.

Issue No: 2006-06b \

06-16-06

If there are questions about the content or opinions expressed, contact

Ventures Technology Watch Editor:    Jeri Trippe  at jerit@venturestechwatch or

Technology Editor – E. Poshkus at edposhkus@creativestrategiesvc.com