Load Balancing over the Long Haul

Arthur Cole

One of the more far-reaching goals of virtualization and cloud computing has always been the idea of regional or even global load balancing.

How convenient would it be to be able to shift workloads not only from server to server but from data center to data center, taking advantage of off-peak hours, lower energy costs and the like, all at the touch of a few buttons?

While that vision has generally taken a back seat to more immediate concerns, such as resource consolidation, a couple of key developments this week indicate that the global data center might be closer than we thought.

The first was Cisco's announcement of its Overlay Transport Virtualization (OTV) system that provides for massive migration of virtual images across the company's Nexus 7000 switch. Technically an addition to the company's Data Center Interconnect (DCI) platform, OTV is an add-on to the NX-OS operating system and essentially extends current migration capabilities, such as those provided by VMware's VMotion system, beyond the LAN onto IP networks. From there, VMs can be rebuilt elsewhere, as long as the target data center is also equipped with a Nexus 7000.

Company execs report that linking up multiple data centers will take all of five minutes, with VMware users gaining particularly simplified management capabilities through a new plug-in that brings the Cisco Application Control Engine (ACE) to the vCenter virtualization management stack. The move provides for control of load-balanced images and ACE instances directly through vCenter rather than a separate management dashboard.

Meanwhile, application delivery specialist F5 Networks says it can now provide a long-distance migration system for VMware's vSphere 4 platform. The company claims it can transfer live virtualized Web applications, plus related storage, over thousands of miles without any downtime or disruption for the user. The system combines F5's Application Delivery Network -- consisting of the Big-IP WAN optimization module and both local and global traffic manager systems -- with the VMotion and VMware Storage VMotion components in a way that overcomes the bandwidth, latency and packet-loss issues that currently afflict long-distance migration efforts.

At the same time, Microsoft users should note that NetEx has added Hyper-V support to its HyperIP continuity and recovery optimization software, providing long-distance data transfers at speeds up to 800 Mbps. While the package is intended for backup and rcovery operations, there's no reason why you couldn't match it to the Live Migration capabilities of Hyper-V to set up a long-haul data center network capable of shifting virtual machines and data loads to a broader set of resources.

Shuttling data halfway around the world might seem like a function reserved for the largest of organizations. But as cloud computing works its way into everyday IT affairs, even smaller operators will see greater flexibility, and less cost, through the ability to shop around for the best deal.

It's just a matter of determining what kind of data you feel most comfortable sending over the long haul.

Add Comment      Leave a comment on this blog post
Feb 11, 2010 4:20 AM Gary Anderson Gary Anderson  says:

The issue of chasing the cheapest power across the globe is something that individual companies have to address on a case-by-case basis.  Does it make more sense to overprovision your compute capacities to create three different centers (or purchased cloud capacity) at an initial cost of 3X, or to try to save 50% OPEX by moving compute load to areas with cheap power?  This might not make sense if the customer is located in the U.S. near a cheap source of hydro-electric, but it could make sense if they're in a heavily taxed and expensive power zone or a zone that has strict regulations on power use and charges substantially more during times of peak usage (Southern California for example).  The financial tradeoff has to be understood before determining if it makes sense for each company.

Feb 15, 2010 10:35 AM Jonathan Price, CA Jonathan Price, CA  says:

Portability between Public Cloud providers, and between in-house and vendor is important.  Logicalis in the UK recently announced another interesting proposition-the Cooperative Cloud.  See details at http://www.youtube.com/user/LogicalisUK


Post a comment





(Maximum characters: 1200). You have 1200 characters left.



Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.