Bruce Moxon
Senior Director of Strategic Technology and Grid Guru

Bruce Moxon works with enterprise customers deploying grid computing solutions. He brings more than 20 years of experience in scale-out computing architectures for both scientific and commercial applications and writes, speaks, and teaches extensively on the continuing evolution of grid computing. Bruce has architected and developed solutions for a number of high-throughput computing environments, including Perlegen Sciences' SNP discovery system, Bank of America's CRM and analytics systems, and NASA's Earth Observing System.

---

The below article was originally published in the April 2006 edition of the Tech OnTap newsletter. To receive the newsletter monthly and enjoy other great benefits, sign up today.

In an era of flat IT budgets, tight power and space constraints, oft underutilized resources, and increasing competitive pressures that require more agile business practices, virtualization has become the mantra of modern IT architects.

The ability to deploy commodity compute, network, and storage resources; to provision them on demand; and to efficiently manage and effectively share those resources is at the core of the grid—or utility—computing movement.

Over the last few years, we've seen significant growth in the deployment of server virtualization technologies such as VMware, Microsoft® Virtual Server, the open source Xen project, and others. These technologies provide a framework for application consolidation in a multiple-OS environment to more efficiently use shared server resources.

To date, these approaches have been applied successfully in quasi-static production and development environments. For example, production application servers and database servers running on disparate operating environments can be cohosted on a single physical server. And multiple virtual machines (running multiple OSes) can be made available on a shared basis throughout a development organization to reduce physical infrastructure costs.

These approaches have certainly led to increased utilization of server resources. In some cases, application migration capabilities have been leveraged to dynamically balance load across shared computing resources.

However, server virtualization alone cannot fully deliver on the core promise of grid computing—true application virtualization—in which applications can be rapidly provisioned, migrated, and even replicated to support a variety of production and development requirements.

"The missing ingredient in the application virtualization 'stack' is storage services."

For the most part, today's applications maintain their "state" on persistent storage—in file systems or databases. These applications increasingly reside on networked storage (SAN and/or NAS architectures). Moreover, it is becoming increasingly common, especially in scale-out server environments, to also maintain operating system and application software "stacks" on shared storage (network boot configurations).

In this environment, the acts of provisioning new applications, migrating them from one physical server to another, and replicating them for development, test, and QA purposes all require support of shared storage services.

We see a variety of trends impacting the move toward a fully realized application virtualization model including:

• Nondisruptive operation. With increasing application consolidation in the server and storage layers, the impact of a system outage is no longer limited to a single application. There will be increasing pressure for true 24x7 operations and nondisruptive upgrades. Storage Magazine does a great job of exploring the progress—and work still to be done—in this area in its March 2006 issue (registration is required, but it's worth it).
• Global namespace (GNS) solutions and file-level migration in tiered storage environments. GNS solutions provide a uniform view of files across many physical file servers. Additionally, they support transparent data migration—often at the file level—to facilitate effective storage tiering. These capabilities exist in clustered storage systems such as the recently announced NetApp Data ONTAP® GX, virtual file management solutions for the Microsoft® Distributed File System (DFS), and inband NAS aggregation switches (think of them as "file routers") from the likes of NetApp partners Acopia Networks and Neopath Networks.
• Diskless boot. An increasing number of organizations are exploring the use of file server-based system "images" and diskless boot as a way to rapidly provision and switch between different software stacks, especially for dev/test environments. With iSCSI software initiators now readily available for Windows® and Linux® environments and emerging support for software initiator-based diskless boot using standard Ethernet cards instead of costly iSCSI HBAs, I fully expect this trend will accelerate.
   
  (A real-world example of a large-scale diskless boot environment is the 1,000-node NetApp Kilo-Client project, which was profiled in last month's issue of Tech OnTap. Software-initator-based diskless boot capabilities were demonstrated last August by Emboot and String Bean Software using the latter's iSCSI target, WinTarget™. Microsoft acquired the WinTarget technology in March of 2006, and NetApp has recently announced support for iSCSI software boot-from-SAN capabilities from Microsoft.)

Over the next few issues, I'll take a deeper dive into the topic of enterprise grid computing and explore ways in which the companies I talk with are leveraging the unique capabilities of Data ONTAP. Upcoming topics will include:

• Use of MultiStore® to develop a dynamic provisioning and load balancing approach for a storage service provisioning model
• Application image cloning and diskless boot for rapid application provisioning in rack-mount server environments
• Database cloning for improved development, test, and QA efficiency and out-of-band reporting and ETL activities
• Whole application migration strategies (server and storage) for application right-sizing

The information in this article is useful to me. 1 2 3 4 5 Strongly Disagree Strongly Agree What other types of information would you like to see included in this article? Additional comments:   Thanks!