The new release is capable of implementing virtual machines on the largest x86 instruction set servers, including those with 126 CPUs and a terabyte of main memory.
Linux can support larger virtual machines in the recent 5.3 release of Red Hat Enterprise Linux and has gained the latest Java open source technology in place of Sun Microsystems' proprietary Java SDK.
Red Hat launched 5.3 Tuesday, eight months after its release of 5.2. "It's usually about six months between releases. This time, we took a little longer" and added major enterprise computing features, said Daniel Riek, RHEL 5.3's manager of product management.
The 5.3 release is capable of implementing virtual machines on the largest x86 instruction set servers, including those with 126 CPUs and a terabyte of main memory. RHEL 5.2 was only capable of supporting virtualized hosts with up to 64 CPUs and 512 GB of memory.
Red Hat has continued its support for nonuniform memory access for the larger systems that was initiated with the 5.2 release. A NUMA architecture means all the cache memories associated with CPUs are coordinated as a whole across a large server, but a single CPU can access data held in its own cache faster than it accesses data in the cache of another processor. Initially, symmetrical multiprocessing systems imposed uniform memory access speeds to assure data integrity, but NUMA approaches -- with operating system support -- allow large systems to speed up operations.
"The system understands how the NUMA memory architecture works," said Andy Cathrow, product marketing manager for RHEL 5.3, in an interview. Servers being built with Intel's Nehalem architecture chips will include NUMA designs, he said.
Under RHEL 5.3, virtualized hosts may also support larger individual virtual machines; a VM may be assigned up to 32 virtual CPUs and up to 80 GB of memory. "I'm not aware of any existing customers demanding more" capacity in a virtual machine, said Riek.
Red Hat Enterprise Linux supports the operation of the Xen open source hypervisor. It plans to add support for KVM, or Kernel Virtual Machine, in the 6.0 version, Red Hat has said. KVM support will come when RHEL adopts a more recent version of the Linux kernel, which is produced by kernel developers under Linus Torvalds. RHEL 5.3 remains based on the 2.6.18 kernel.
RHEL 5.3 also supports Hugepage memory paging, an x86 architecture feature that can substitute a 4-MB memory page for a standard 4-KB page that's moved in and out of memory devices. RHEL 5.3 also supports Intel Extended Page Tables, a feature of its vPro virtualization technology that allows a virtual machine to track and manage its own memory pages. Both features can be invoked to improve the performance of virtual machines and give system administrators the option of placing larger workloads on a virtual machine.
The OpenJDK, or open source Java Development Kit, has been added to the 5.3 Red Hat distribution. Open JDK uses the same code base and is a match for Sun's Java Standard Edition 6. In the past, Linux users had to make use of Sun's proprietary Java Development Kit; now they can develop applications as open source code that work with Red Hat's JBoss Application Server and other middleware, Riek pointed out.
Red Hat said the 5.3 release will support Intel Core i7 processors and successors in the 45 nanometer, hyperthreaded Nehalem architecture line. The 5.3 release running on a Core i7 server may show "exceptional performance" improvement of 1.7 times faster execution for complex business applications.
It also will be able to invoke Intel's Dynamic Acceleration Technology. When the operating system detects idle CPU cores, it's able to put them to sleep temporarily in order to save power.
Google in the Enterprise SurveyThere's no doubt Google has made headway into businesses: Just 28 percent discourage or ban use of its productivity products, and 69 percent cite Google Apps' good or excellent mobility. But progress could still stall: 59 percent of nonusers distrust the security of Google's cloud. Its data privacy is an open question, and 37 percent worry about integration.
CIOs Get Smart About BIIT’s tried for years to simplify business intelligence efforts. Have visual analysis tools and Hadoop and NoSQL databases helped? Respondents to our 2014 InformationWeek Analytics, Business Intelligence, and Information Management Survey have a mixed outlook.