HPC in the Cloud: Hybrid Cloud Ready for Takeoff in 2012

Dedicated to covering high-end cloud computing
in science, industry and the datacenter

Translation Disclaimer

Subscribe | Sign In

January 17, 2012

Hybrid Cloud Ready for Takeoff in 2012

Denoid Tucker, Vice President of Technology, StrataScale

If 2011 was the year that cloud computing gained public acclaim as a revolutionary new way to use computing and storage resources more efficiently, 2012 may be the year that the IT industry's predominant focus on the public cloud begins to give way to a greater appreciation of hybrid cloud computing.

futuristic arrows Public clouds – which allow multiple users to share massively scalable arrays of servers, storage and other resources through the Internet for maximum cost savings – dominate the current cloud computing market. Most enterprises, however, still rely largely on either internal or externally hosted private clouds that reserve computing resources exclusively for their own organization.

Hybrid clouds, on the other hand, combine aspects of both these models, providing the control and legacy compatibility of dedicated internal resources where needed, while shifting other less-demanding or less-critical applications to the off-site cloud. Hybrid clouds, as I view them, can encompass not just a blend of internal and external computing resources, but also both dedicated physical servers and virtual servers. This inherent flexibility can make hybrid solutions a good choice for adding new capacity or capabilities to existing legacy systems while still gaining some of the cost and scalability benefits of the public cloud.

Amazon Web Services, with revenues approaching $1 billion a year, has become far and away the industry's largest cloud services provider while concentrating almost exclusively on the public cloud. Yet Amazon, even when combined with a host of lesser-known public cloud providers, is barely scratching the surface of cloud computing’s ultimate market potential.

While the public cloud market is probably at least 10 times larger than the hybrid cloud market today, I expect the hybrid cloud to begin closing that gap. It wouldn't be surprising, in fact, to see hybrid implementations reach 50 percent or more of the total market as mainstream enterprise users grow increasingly serious about moving more of their applications and infrastructure to the cloud.

Enterprise cloud concerns

The truth is, most enterprise computer users aren't willing, or able, to entrust their mission-critical applications to the public cloud. Their reluctance is partly due to concerns about security, reliability and regulatory compliance in a shared computing environment. Equally important, however, is the realization that there are some things the public cloud simply won’t do.

The problem is that while some applications run just fine on virtual servers, others don’t. If you’re using multiple virtual machines (VMs) or sharing servers between multiple users, you run the risk of one VM using too much of the computing, storage or bandwidth resources. And when the remaining users must vie for a small, fixed pool of resources, applications are prone to crash or perform poorly.

Unfortunately, users in a public cloud environment simply don’t know what demands other users are going to place on their shared computing resources. As a result, enterprises often don’t know whether their applications will even work in the public cloud.

My experience shows that some applications – particularly large enterprise resource planning databases, accounting systems and other I/O-intensive applications – require the full capabilities of a dedicated hardware system. That’s why offering both physical and virtual dedicated server hosting, in addition to a full range of public and private cloud services makes sense. I have learned that when it comes to building and operating effective cloud-based systems, “one size fits all” is often not the best approach.

There are situations where it makes sense to keep at least some data and IT resources in-house. But there clearly are significant benefits to be gained from moving web servers, storage and other easily managed functions into the cloud. Just remember that whenever you’re ready to venture beyond your own data center and into the cloud, there are a wide variety of options available.

Start by identifying your specific needs and objectives, and then consider what makes sense to move into the cloud and what doesn’t. And unless you’re ready to make a major change, be sure to choose a solution that gives you the flexibility to run all the applications you already have.

About the Author

Denoid Tucker is Vice President of Technology for StrataScale, Inc., a Sacramento, Calif.-based provider of public and private cloud servers, dedicated servers and hybrid hosting services.

Share Options

Subscribe

Subscribe to HPC in the Cloud

Discussion

There is 1 discussion item posted.

Hybrid clouds using Amazon can be very straightforward
Submitted by Mark Blessing on Jan 18, 2012 @ 10:23 AM EST

Interesting writeup, Denoid. There is a growing demand for hybrid clouds as you describe, and with it, an opportunity to transform the cloud into a transparent extension of the on-premise servers.

For example, Bright Cluster Manager 6.0 automatically cloud enables Linux servers, providing the capability to extend out to AWS EC2 with a few mouse clicks, and then manages these nodes as part of the local cluster. Bright provisions, schedules, monitors and manages these nodes in the same GUI or cluster management shell as the local cluster. And to your important point that not all applications are suitable to run the cloud, the sys admin can easily designate jobs as cloud-appropriate, and the scheduler can automatically allocate cloud-based nodes when additional capacity is required.

This capability also works for private clouds that use the EC2 APIs.

Here is a brief demo on Youtube: http://bit.ly/xTfeYE
and a short online description: http://bit.ly/u1fkWo

Post #1

Join the Discussion

Become a Registered User Today!

Registered Users Log in join the Discussion

RSS Feeds

Feeds by Topic

Feeds by Content Type

Subscribe to All Content

All

February 22, 2012

February 21, 2012

February 20, 2012

February 17, 2012

February 16, 2012

February 15, 2012

February 14, 2012

February 13, 2012

February 10, 2012

Around the Web

UK Government Opens CloudStore

Feb 21, 2012 | UK's CloudStore aims to change the face of public sector ICT procurement by offering 1,700 services in an app-store format.
Read more...

Unified Computing Pioneer Transforms Business Model

Feb 20, 2012 | CEO Pete Manca details Egenera's unusual journey from hardware vendor to software provider.
Read more...

Biotech Firm Doubles Down on Amazon Cloud

Feb 14, 2012 | Connecticut-based biotech firm Metrum Research Group moves entire IT infrastructure to Amazon cloud.
Read more...

NSF Report Details National Cloud Agenda

Feb 13, 2012 | The National Science Foundation (NSF) describes its support for cloud computing research in a 21-page report, submitted to Congress last week.
Read more...

Read more headlines...

Introducing LRDIMM – A New Class of Memory Modules

01/17/2012 | Inphi | This paper introduces the LRDIMM, a new type of memory module for high capacity servers and high-performance computing platforms. LRDIMM is an abbreviation for Load Reduced Dual Inline Memory Module, the newest type of DIMM supporting DDR3 SDRAM main memory. The LRDIMM is fully pin compatible with existing JEDEC-standard DDR3 DIMM sockets, and supports higher system memory capacities when enabled in the system BIOS.

Optimizing Data Centers for Big Infrastructure Applications

12/08/2011 | StackIQ | Today's application workloads have grown to encompass hundreds or thousands of servers and massive amounts of data. Organizations such as Google and Facebook have developed proprietary tools and techniques to manage their compute environments. But what about the rest of us? This white paper provides an overview of the growing wave of large-scale, ‘Big Infrastructure’ applications and explores the challenges inherent in provisioning, configuring, and deploying software for clusters running massive data workloads.