In attempting to relieve the increasing bottlenecks and performance issues plaguing cloud applications, IBM inventors have come up with a method for dynamically managing network bandwidth in cloud environments. The invention, dubbed Dynamically Provisioning Virtual Machines, automatically determines the best way for users to access cloud computing resources based on the availability of network bandwidth.
The way it works is this: As the virtual machines that serve as gateways to cloud services become overtaxed by growing numbers of users requesting access, thus constraining applications, IBM’s new methodology reassigns workloads from one system node to another based on bandwidth requirements and availability. The result, IBM says, is the removal of a major roadblock to cloud efficiency.
Today’s users “have zero tolerance for network bandwidth bottlenecks,” said Ed Suffern, IBM systems engineer and lead inventor, in a statement announcing the breakthrough. “IBM’s patented dynamic provisioning invention will help cloud service providers increase network performance and improve customer satisfaction.”
The new methodology should provide a significant improvement over existing approaches, said Cliff Grossner, directing analyst at Infonetics Research, via email.
“This innovation allows adjustment of workloads based upon new capabilities to determine network state, rather than rebalancing workloads only on server state, which can produce non-optimal results,” Grossner said.
[Startup ThousandEyes monitors the performance of cloud applications. Read how it can help IT identify and troubleshoot problems in "ThousandEyes Peers Into Cloud Performance."]
IBM said its new method is ideally suited for cloud apps that commonly experience dramatic or unexpected peaks and valleys in demand for services. Examples include online retailers facing crushing holiday traffic, search engines contending with surges in response to current events, and government and news media sites that are overrun in response to events such as elections, conflicts and natural disasters.
“This capability would enable an automated data center to quickly react and rebalance workloads to remove network bottlenecks,” said Grossner.
In addition to immediate practical applications, IBM maintains that its approach provides a foundation for software-defined networking (SDN). The invention relies on software to manage network resources by obtaining data from network switches to determine the amount of bandwidth being used by IP addresses assigned to VMs. As network bandwidth becomes constrained on one node, the system automatically reassigns VMs to another node with available bandwidth.
Grossner said IBM’s invention could prove beneficial to both traditional and SDN-architected networks in analyzing and reacting to the needs of virtualized workloads.
“This invention can be used by traditional L2 and L3 switches, and also be included in SDN controllers going forward,” he said. “This will improve the capability of both traditional and SDN-architected networks to analyze the needs of virtualized workloads and react.”
IBM’s method can run on a variety of operating systems, including Windows, Linux, UNIX and CentOS.