Adaptive Computing, known as Cluster Resources until a name change last August, brings 10 years of managing high-performance computing clusters to supervising the x86 servers found in private clouds being built in enterprise data centers.
There are many companies vying to manage private clouds, including VMware, Citrix Systems, and Microsoft, whose hypervisors are often running the virtualized servers. But Adaptive offers a broad suite of components capable of mapping the physical and virtual x86 resources into a more comprehensive whole. Version 6 of its Moab Suite, released Monday, adds policy-driven automated provisioning of servers in the private cloud. The system already includes such components as storage management, network management, a change-management database, license management, and hardware monitoring.
Systems management vendors, including BMC Software, HP, CA Technologies, and IBM already offer some of these features. But it's instructive that HP resells Moab Adaptive Computing Suite itself, Michael Jackson, president and COO of Adaptive, said in a statement. "Innovative partners, such as HP, recognize the need for intelligent resource management and rely on Moab technology to help their customers further extend the value of their cloud infrastructures," he said.
"Moab can recognize high-priority versus low-priority workload placement," Peter ffloulkes, Adaptive's VP of marketing, said in an interview. It can decide on the optimal placement of a new service on a server cluster, depending on the resources available, he added.
The system can automatically initiate live migration of virtual machines to better utilize the hardware resources, whether to free up more CPU cycles for a particular application that is in high demand or to consolidate virtual machines with decreasing traffic on fewer hardware servers. Although Moab aims to be hypervisor agnostic, the live migration feature so far works only with VMware ESX Server with its mature live migration capabilities, said ffoulkes.
Moab gathers a historical record of resource performance and combines known performance capabilities with its scheduling engine. As predictable workloads materialize, it can automatically distribute them across the server cluster, he noted.
The efficient use of x86 servers, networking, and storage, along with knowledge of licensing costs -- license expense is often determined by the number of sockets on which an application runs -- will save $1 billion in operational costs over a three-year period for one of Adaptive's largest financial services customers, claimed ffoulkes, although he said he did not have permission to name the customer.
Moab can be used to establish infrastructure as a service, platform as a service, or software as a service, depending on the goals of the customer, he said.
The Moab Adaptive Computing Suite 6.0 carries a $300 annual subscription per server socket.
The former Cluster Resources, now Adaptive Computing, was a leading management system for high-performance computing systems. Moab runs half of the systems on the top 10 high-performance computing list, such as Oakridge National Labs, Lawrence Livermore Berkeley Labs, and Los Alamos National Labs. Originally known as the Moab Cluster Suite, it evolved into a Moab Grid Suite, an Adaptive HPC Suite, and for the last two years, the Moab Adaptive Computing Suite. Release 6.0 appears about one year after the previous, 5.4 release.
Adaptive Computing received $14 million in venture capital funding in September from first-round investors. They included Intel Capital, Tudor Ventures, and Epic Ventures.