Load balancing technology, which took off in 1990s with the rise of the Internet, continues to find behind-the-scenes work in the enterprise — including a supporting role in the current mobile boom.
Take the case of Richard Fleischman & Associates, an IT consulting firm that uses load balancing products from Kemp Technologies in its disaster recovery (DR) line of business. RFA’s financial services industry clients include hedge funds and broker-dealers, and financial industry regulations require a disaster recovery plan. (New York City customers can cut over to RFA’s Purchase, N.Y. disaster recovery center or the company’s Boston center.)
Stevens Demorcy, senior systems engineer at RFA, says Kemp Technologies provided a solution to a tricky DR issue: Making customers’ iPhones, Android and Windows smartphones quickly available in the event of a disaster. In the last few years, many customers had migrated from BlackBerry and BlackBerry Enterprise Server (BES) to smartphones that use Microsoft’s ActiveSync mobile data synchronization technology. (ActiveSync is a Microsoft Exchange Server feature that lets users access email, calendar and contacts on mobile devices.) During Superstorm Sandy, many clients had to go to DR and wait hours for DNS updates to propagate and reach their devices.
Related: Gimme Shelter: Wall Street Braces for Next Superstorm
Analysis: Did Wall Street’s Business Continuity Plans Fail During Sandy?
In the BlackBerry world, BES handles the back-end communications task of redirecting blackberry devices to the DR site. But that tool no longer exists in the smartphone/ActiveSync environment. As a consequence, it can take up to a day for some smartphones to get an updated DNS record, which redirects the device to a new IP address, Demorcy notes.
RFA built its smartphone solution using a combination of Kemp’s LoadMaster local load balancers and GEO global load balancers. The GEO appliances are key in this application, since they redirect smartphones when customers go into DR mode. Demorcy says the GEO acts as a DNS server, repointing traffic wherever the server administrator needs it to go. RFA typically places one load balancer and one GEO at the customer’s site and the same number at a disaster recovery center, though Demorcy says the particular arrangement may change depending on a client’s needs.
Load balancing is an “old concept” that “hasn’t changed much,” Demorcy says, but using the global feature to quickly give people access to their email during an emergency has reinvigorated the technology. “That’s the part that makes it worthwhile. The global part is where you benefit the most if you want to do DR.”
Market Shift Pushes Load Balancing Into Application Delivery Controllers
The first flowering of load balancing coincided with the rapid rise of Internet properties, which found a need to distribute workloads from one Web server to another as traffic dramatically expanded. Software and appliance-based load balancers became mainstay tools for business seeking scale and high availability.
More recently, however, load balancing to some degree has been morphed into the newer technology category of application delivery controllers (ADCs). Those devices provide load balancing as part of a broader set of features that may include compression, traffic management, network and application protocol optimization and Layer 7 content switching. Secure Sockets Layer offload is generally an ADC function but can also be found in more traditional load balancers.
Heman Tailor, director of IT at SCAinteractive, a company that provides interactive promotions such as customer loyalty programs, cites the additional features as a plus. The company currently uses the Coyote Point E450GX appliance. (Fortinet acquired Coyote Point Systems earlier this year.) The technology manages the Web server cluster in SCAinteractive’s network operations center, which the company uses to run its promotions.
“Using a load balancer is very typical of running a Web cluster — or any other cluster, for that matter,” Tailor says. “There are many solutions out there, but what makes it stand out is features such as hardware compression and SSL offloading, and Layer 7 routing and load balancing.”
Case Study: Thanks to ADC, Carfax Accelerates Data Delivery to Meet Demand
Those features, Taylor says, “reduce the load significantly on our end, along with making the management of the servers behind the clusters much easier.”
As for future developments, Tailor points to the incorporation of global load balancing as most server farms add second and third sites to boost reliability.
Load Balancing’s Enduring Qualities Work Well for Failover
While load balancers have grown in sophistication as they meld into application delivery controllers, customers continue to cite the old-fashioned virtue of reliability as the important draw. The capability to shift workloads away from oversubscribed servers in a data center or to failover from one server to another remains a core feature.
Now Nerd, based in Beaverton, Ore., deployed Barracuda Networks‘ Load Balancer 440 to support its NerdDeck offering, a white-label remote technical support service. NerdDeck serves as its customers’ remote support department, providing a phone and chat service as well as remote access tools.
Geoff Turner, Now Nerd CEO, says installing the load balancer paid off within weeks of purchasing it. A server failed overnight, but the load balancer detected the outage and redirected requests to other servers. Now Nerd staffers never noticed the problem until they came in the next morning, checked email and discovered the server alerts. Customers detected no change in service.
News: Load Balancing Among Issues in Healthcare.gov Enrollment System Crash
“Different people have different purposes for load balancing,” Turner says. “Our goal was not offloading SSL. It was to give customers a consistent experience.”
That said, Turner notes that his company may use some of the newer features such as global load balancing down the road.
Embedded Load Balancing Meets Virtualization Challenges
Load balancing technology in recent years also has had to contend with virtualization and cloud computing, which complicate the environment in which workloads are distributed. Accordingly, some enterprises have adopted server application virtualization software with embedded load balancing.
That was the case for W&W|AFCO Steel, a structural steel fabricator in Oklahoma City. The company runs key applications such as labor scheduling and enterprise resource planning on its SQL Server database platforms, which operate as VMware virtual servers.
Analysis: Will Fault Tolerant Servers Improve Your Virtualized Environment?
Tips: How to Virtualize Mission-Critical Applications
At one point, W&W|AFCO Steel was running SQL Server 2005 and 2008 and planning to migrate to SQL 2012. But an outage affected a significant portion of the company’s IT infrastructure before that transition could occur: The Citrix environment on SQL Server 2005 went down for three days, says Todd Park, vice president of IT at W&W|AFCO Steel.
A the time, the steel fabricator lacked a high availability capability. But after its brush with failure, the company deployed DH2i‘s DxConsole server application virtualization software. DxConsole includes a resource manager that load balances virtual instances, providing a failover capability.
W&W|AFCO Steel put DxConsole into production earlier this year. The company’s SQL Server 2008 R2 and SQL Server 2012 virtual instances failover to one another in the event of an outage, Park explains.
The software requires little intervention to operate, which is a boon for a company with a small IT shop, Park says. “We don’t need to manage anything on a daily basis. It just works.”