These economic development stories where states pass some special legislation making it attractive for a company to build a datacenter in their location are fairly common news these days. The story is basically that the states tend to offer short-term economic benefits to the datacenter builder with the hopes of getting their share of the economic development in the long term from property taxes and such after the term the legislated incentives is over, and in the short-term from taxes on the earnings of local workers and suppliers in the build stage of the project.

As tracking the datacenter industry is part of my job I read these news stories almost weekly, usually in the form of a small blurb in the online version of the local newspaper, and had been following this story, waiting for something definite to happen, but surprisingly this inter-state battle has made the national news, with CBS News and CNN picking up on what looks to be a very large investment by someone in one of these two states.

But even the national press isn’t asking the hard questions; how much, exactly, are these states willing to concede in tax revenues, and spend in infrastructure updates to attract these datacenter construction projects.  And can they point to any successful projects, so far, in places that have made these concessions and gotten the payback they expect, which is usually some sort of technology renaissance happening in a rural community with additional technology vendors opening shop there.

The amount of money spent on building and equipping a datacenter can be mind-numbingly huge to areas that are not used to seeing those kinds of dollars spent on technology projects. But there still seems to be little understanding in these areas as to how the money is actually spent and the impact of that spending, or rather the lack thereof, on the local community.

The website http://www.FDCCIConnect.com contains a number of documents that can be used to evaluate the progress of a consolidation effort and hopes to attract a significant number of users from the Federal datacenter teams working on the consolidation effort in order to build a viable birds-of-a-feather type community.

I’d really like to see this site be successful. There is significant value in learning from the experiences of others working in the very specific area of datacenter consolidation for the federal government.  Almost all of the issues being faced by the teams dealing with this first round of consolidation efforts will be applicable to the ongoing consolidation and closure process and taking advantage of the experiences of those who have gone before can only be of benefit to those government IT folks who are just now begging the planning and evaluation of their datacenter consolidation and closures.

But the Health Insurance Portability and Accountability Act (HIPAA) means that the security of medical data is an absolute necessity for any vendor that deals with medical information. And this isn’t just a set of suggestions; datacenters have to meet very strict standards for data protection to be HIPAA certified. The certification steps range from specific training for datacenter workers who have access to protected data, to government audits by HIPAA inspectors that assure that the requirements in the Code of Federal Regulations are met. Additional reporting requirements are required and guarantees must be in place for the security of the data. Breaching those guarantees can result in a variety of penalties.

The problem that cloud service providers will face in delivering services to the medical industry is that each datacenter that holds any patient data will technically need to be HIPAA certified. So there is no simple way of making sure that identifiable components of the patient data will ever be exposed when that data may be distributed throughout the cloud.  This doesn’t mean that their won’t be HIPAA certified clouds, it just means that the broad promise of cloud delivered services being able to be a best of breed choice from among all available selections won’t be one that will be available to medical services that need to deal with patient data.

But this is America, and where there is a perceived need there will be vendors who will supply that need.  Datacenter providers like Colocation America, who just announced the HIPAA certification of their datacenters, will undoubtedly team up with other datacenter providers to begin to offer networks of HIPAA certified datacenter back ends to enable application service providers to offer their services to hospitals and medical practices throughout the country.

These SM10000-XE microservers, which can also support up to 64 SATA disks or SSDs without reducing the CPU density as well as up to 16 10 GbE or 64 1 GbE ports in the same rack are a similar, but higher performance solution than SeaMicro’s original SM10000-HD server solution which was based around Intel’s very low end Atom CPU. By moving to the Sandy Bridge Xeon processor they have delivered a scale-out solution for customers needing the higher performance CPUs while trying to control energy consumption and space utilization within the datacenter. By utilizing Samsung’s low energy consumption Green DDR3, they have tried to minimize the most common energy sinks in their servers.

SeaMicro’s own Freedom fabric ASIC is the key component to interconnecting all these CPUs, memory, and storage at a very high speed with low latency. The same fabric is used in their Atom CPU based offerings.

This design takes microservers in a new direction. With their original idea being to use low-power, low capability CPUs in large numbers to take on workloads that traditionally didn’t require a lot of CPU power, SeaMicro has switched directions in an attempt to bring the energy and space savings concepts to a part of the datacenter market with a need for higher performance servers that can still minimize datacenter energy and space utilization.

• Excellence in Data Center Consolidation Leadership
• Innovative Application of Technology to Support Data Center Consolidation
• Data Center Consolidation Program with Maximum Savings for the Organization

As the awards are focused on government programs, they are only open to government agencies and employees involved in the current datacenter consolidation plans that have been mandated across all Federal agencies. The published criteria for the awards are focused on the types of conditions one would expect; how well the programs worked, how well all stakeholders were involved, sustainability of the approach, etc., with the codicil that the comparisons are strictly for public-sector operations.

And therein lies the rub; the metric for the awards isn’t set against what is possible to be done, but simply how well the nominee has done compared to other public-sector efforts. While I applaud MeriTalk for making the effort to recognize the progress that is being made, all of the publicly announced results I have seen so far are less than impressive, even within the narrow announced goals of the consolidation program.

For the government to truly take the leadership role in consolidation and optimization that the original announcement seemed to be positioning as the Administration’s effort in this field, the performance of the consolidation and optimization program needs to be measured in more than just the number of datacenters closed or the so far pitifully small savings that have been achieved when compared to the $70+ billion dollar budget for Federal IT spending.

Because the official definition of a datacenter as used by the consolidation effort includes spaces as small as a departmental server closet (see http://www.cio.gov/documents/FDCCI-FAQs.doc) looking at gross numbers for “datacenters closed or consolidated” can be very misleading to the public or anyone who has a more traditional understanding of what a datacenter physically represents.

Show me a 10% cut in overall datacenter expenditures with no reduction in capabilities or a 10% increase in work performed achieved through operational efficiencies while consolidating excess or legacy equipment and I’ll be impressed.

The vStart products show up at your site preconfigured to your specifications and ready to run. The installation support includes a Dell field service engineer to assist in the installation. Dell’s goal is to deliver a system that fits seamlessly into your existing environment. The system is self-contained in a rack mount that hosts the Dell servers, Dell EqualLogic storage, UPS,  pre-labeled cabling, all neatly wired and attached, and all the  networking interconnects for the hardware devices that make up the pre-packaged cloud.

The base operating system with the vStart is Microsoft Windows Server 2008 R2 datacenter edition which allows unlimited VMs under the terms of the Microsoft software license. The system can be managed using both Microsoft and VMware management solutions as well as Microsoft System Center Advanced Infrastructure management plug-in from Dell for the vStart solutions. Dell is also a partner in the Microsoft Private Cloud Fast Track program.

So in 2009, when Cisco released their Unified Computing System, it didn’t seem to bring a lot of concern to the big players in the server hardware business, like HP and IBM. Despite the tight integration the UCS offered between Cisco’s server and networking infrastructures that was all that Cisco seemed to be able to offer.  The other critical hardware and management components of their soup-to-nuts approach had to be provided by other vendors, such as EMC, HDS, CA, and BMC. HP and IBM, on the other hand, offered a complete line of their own storage hardware and software management tools, all tightly integrated with their own servers and networking. This put Cisco firmly in the role of a sort a super systems integrator, with all the baggage that carried when compared to HP and IBM.

But Cisco has surprised their competitors and the industry as a whole with the success of their UCS approach. They announced this week that in the last 30 months they had built a base of over 10,000 UCS customers, proving that there was a definite market desire for their approach to the single source, unified computing environment. Ranging from the sole service provider to smaller enterprises, to point product solution s in larger environments, UCS looks like it’s here to stay.

While there is some question in the general purpose computing world over the value of a niche performance product like InfiniBand when 10 GbE and 40GbE products are much more mainstream and can meet the majority of business computing performance and network latency needs, the core HPC market for InfiniBand isn’t going away. And from that base, Intel has the opportunity to move the technology that is InfiniBand into more general purpose network and switching fabrics.

But it’s not as simple as building ASICs and chipsets for InfiniBand devices.  By acquiring this technology Intel also has a jumpstart in Integrating InfiniBand into their motherboard chipsets and eventually integrating the InfiniBand support directly into the CPU.

This isn’t a farfetched or even an uneconomical move by Intel. They predict that the HPC market, in the form of the top 100 supercomputers, would alone use a number of CPUs equivalent to today’s entire CPU market by 2018.  That’s a lot of processors, and direct support for the top switching and interconnect fabric would make Intel CPUs a dominant player in that market.

This would also be the logical continuation of the basic idea that motivated the InfiniBand concept; to integrate desktop, servers, and storage into a single high performance switched fabric. And end-to-end common interconnect system from user to datacenter with very high performance. And the acquisition of these QLogic assets also increases Intel’s portfolio of patents and engineering skills in an area where they see significant importance.

While QLogic is a major player in the InfiniBand market, the market leader is Mellanox Technololgies, the provider of choice for Oracle and their Exadata database clusters.  Oracle likes the Mellanox product so much that they took a 10% stake in the company last year, hoping, I’m sure, to get some control over this enabling technology for their premier database product.

This isn’t much of an issue when you are faced with large or small urban areas, as one industrial or office building pretty much just looks like another.  And as we have seen, datacenters have little trouble co-existing inside existing office towers or industrial parks. But Fiberpop’s first datacenter plan, designed by Keith Waters & Associates, a Minnesota fine home building firm, sets a new standard for the concept of fitting your business into the local community.

From the proposal filed with the local planning commision

The plans they have submitted for approval are for a 36,000 square foot datacenter  that fits easily into a community filled with upscale homes. Looking like a large, modern home, complete with gabled roofs, the exterior look of high-ceiling great rooms, and even a chimney to imply that warm, cozy fireplace, the facility is designed as a modern building that will be able to co-exist in a residential community without notice (beyond that, hopefully, the occasional comment of “Wow, that’s a cool looking place.”)

Business structures often are faced with dealing with the local residents NIMBY desires, but by hiding their community-based datacenter in plain sight, and not simply trying to stick a traditional office building in the middle of a residential area, Fiberpop and their architects might well have hit on the future design model for smaller datacenters servicing local business and community needs.

The eight components that make up System Center 2012 are:

App Controller (beta) - This application is designed to be a single point of control for applications across both private and public clouds. System administrators can both deploy and manage these applications through this interface.

Configuration Manager (RC)  - Microsoft completely redesigned the rule model of Configuration Manager for this release., It was a necessary change in order to support a broader range of devices, in including support for mobile devices running Android, iOS, and, of course, Windows Phone 7.

Data Protection Manager (beta) - DPM continues its real-time data protection capabilities and adds single-window management across security and other services.

Endpoint Protection (RC) - Improvements in the anti-malware, signature detection, vulnerability warning system, and a focus on user controls. In addition to those in place for device control

Operations Manager (RC) - The core product of the Systems Center line, Ops Manager provides centralized, single console management of your public and private cloud resources.

Orchestrator (RC) - management for process automation with hooks to the major automation player’s management tools (IBM, BMC, EMC, CA, and most notably, VMware).

Service Manager (beta) - The primary tool for deploying services across your cloud infrastructure.

Virtual Machine Manager (RC) - With support for more than just Microsoft’s basic hypervisor, VMM now also handles VMware, along with Xen and Azure.  It can combine capacity from all the different virtualizations into a single cloud.  As they are not the leading player in the VM market it makes sense to support alternatives to their own system.