What's new in Red Hat Enterprise Linux 6.2
After the release, RHEL 6.2 began translating " What's new in Red Hat Enterprise Linux 6.2, " but since more interesting materials were constantly appearing, I finished the translation just now (I shortened it a little). I hope that the relevance of the post is not completely lost.
At the end of 2011, thirteen months after the release of Red Hat Enterprise Linux 6, the second update of the corporate Linux distribution was released. It is a logical continuation of the first update and contains not only fixes released after RHEL 6.1, but also an extensive list of new features - virtualization, resource and file system management have been significantly improved.
Resource management
The RHEL kernel scheduler now supports the function of “control over groups of CPU threads” (cgroup). It prevents applications from using more CPU resources than are allocated to this group even if some of the CPU resources remain unused. This is true, for example, for “cloud” environments available via pay-per-use (paying only for used resources), because it does not allow users to capture more resources than they bought. Similar restrictions on resource consumption are also interesting for providing a guaranteed level of service. Now you can be sure that user-acquired CPU resources are available in the guest system quickly and at any time.
')
In addition, Red Hat improved the scalability of the cgroup-code. Now you can use hundreds of groups working simultaneously without any reduction in performance. Red Hat developers also note the improved performance of cgroup controllers for I / O and memory tasks.
Core
The RHEL kernel is based on linux kernel version 2.6.32 and offers the pstore function, which allows you to analyze data about the system crash (on some of the modern hardware platforms) stored in a special non-volatile memory.
As usual, Red Hat has added a variety of drivers to support new hardware. Especially a large number of changes affect drivers for storage systems and network devices. The improvement of support for 10-Gigabit network devices deserves special attention. Audio drivers, USB 3.0 and PCI Express (PCIe), have also been updated.
Red Hat, as a rule, does not change the main components of the distribution kit during the life cycle of the major version of RHEL (ie, for example, 6.0). However, this time an exception was made for the graphical stack. There was a change of Mesa 3D to version 7.11 and X Server on 1.10. Developers have changed their rule in order to be able to update graphics drivers for AMD, Intel and NVIDIA. Release Notes also includes support for Intel graphics for the next generation of chipsets. Probably meaning the platform Ivy Bridge, the appearance of which is expected in 2012. The updated Synaptics touchpad driver adds support for multi-touch technology.
Red Hat has updated the perf utility used for performance monitoring, and its kernel components, where, among other improvements, cgroup support has been added.
File systems
Improved scalable file system XFS supported within the Scalable File System add-on. Optimized processing of loads associated with intensive processing of metadata, which are, for example, when accessing a directory with a large number of small files. Previously, such operations could greatly affect performance.
As a technology preview , the new release includes Parrallels NFS (pNFS) as part of NFS 4.1. This technology is designed to significantly increase the speed of access to data through parallel data processing.
Thanks to “Lazy Inode Table Initialisation”, such operations with the Ext4 file system as disk partitioning can now be performed much faster, since Initially, only the most important file structures are created during the formatting process (and the rest later). Asynchronous data writing allows you to write data to the CIFS file system up to two times faster than before.
Network
Now fully supported work with cluster Samba (CTDB) on a cluster file system GFS2. The High Availability Addon (HA) allows you to support HA configurations for applications running VMware virtualization on RHEL 6.2 guest OS.
The newly added Stream Control Transmission Protocol (SCTP) to support multiple addressing allows hosts on a network to be accessible under different IP addresses. The RHEL 6 kernel now has a sendmmsg system call, which allows a whole batch of messages to be sent to use one call, which reduces the overhead of system calls and increases network throughput. Also, to increase bandwidth in RHEL updates, the Transmit Packet Steering (XPS) feature is included.
Virtualization
Added a number of new features related to virtualization, which will not be superfluous for Red Hat Enterprise Virtualization (in RHEV 3.0 the same core as in RHEL 6.2).
Starting with version 6.2, RHEL will use “virtual CPU time sharing” for guest KVM systems that have access to multiple processors. This technology allows guest systems to use the resources allocated to them more efficiently and, therefore, work faster. This is especially advantageous on large systems with multi-core processors, they have the problem of quality and performance, which are reduced due to the problems of preferential access. This is usually associated with locks that provide exclusive access to data structures. However, it is worth noting that this functionality requires hardware support. At AMD, it is called “Pause Filter”, and at Intel “Pause Loop Exiting” (PLE).
Red Hat developers have increased the speed of processing UDP data running through virtio-net drivers; RHEL 6.2 will more efficiently handle network packets that are smaller than 4 KB. Optimization of network communication "I / O path" are designed to improve performance. Red Hat also points out that the speed of macvtap and vhost network technologies has increased. However, they are still Technology Previews and, therefore, are not included in the support contract. The same applies to live snapshot support, which allows the guest system to save a file system image while running — for example, to back up data.
On NUMA systems, using libvirt, you can now not only determine the processors used, but also the memory areas for the guest OS. Ideally, this allows the use of memory, which is most accessible to applications running on a NUMA node. Qemu, which is used for KVM, now also offers USB 2.0 emulation. When working in a paraverted Xen guest OS, RHEL 6.2 supports ballooning (“swapping”) of memory, a feature that allows you to adjust memory during program execution.
Also recently, Linux Container (LXC) applications have become supported. This allows you to separate applications from the rest of the system and regulate the use of resources using cgroup and namespace. This functionality can be enabled through the libvirt API and virt-manager; however, it also has the Technology Preview status at the moment.
More details?
More information about these and other new features of RHEL 6.2 can be found in the Red Hat documentation. A quick tour is available in Red Hat Enterprise Linux 6.2 - What's New . Complete information can be found in the Release Notes , and for those who want to dig even deeper - Technical Notes .
At the end of 2011, thirteen months after the release of Red Hat Enterprise Linux 6, the second update of the corporate Linux distribution was released. It is a logical continuation of the first update and contains not only fixes released after RHEL 6.1, but also an extensive list of new features - virtualization, resource and file system management have been significantly improved. Resource management
The RHEL kernel scheduler now supports the function of “control over groups of CPU threads” (cgroup). It prevents applications from using more CPU resources than are allocated to this group even if some of the CPU resources remain unused. This is true, for example, for “cloud” environments available via pay-per-use (paying only for used resources), because it does not allow users to capture more resources than they bought. Similar restrictions on resource consumption are also interesting for providing a guaranteed level of service. Now you can be sure that user-acquired CPU resources are available in the guest system quickly and at any time.
')
In addition, Red Hat improved the scalability of the cgroup-code. Now you can use hundreds of groups working simultaneously without any reduction in performance. Red Hat developers also note the improved performance of cgroup controllers for I / O and memory tasks.
Core
The RHEL kernel is based on linux kernel version 2.6.32 and offers the pstore function, which allows you to analyze data about the system crash (on some of the modern hardware platforms) stored in a special non-volatile memory.
As usual, Red Hat has added a variety of drivers to support new hardware. Especially a large number of changes affect drivers for storage systems and network devices. The improvement of support for 10-Gigabit network devices deserves special attention. Audio drivers, USB 3.0 and PCI Express (PCIe), have also been updated.
Red Hat, as a rule, does not change the main components of the distribution kit during the life cycle of the major version of RHEL (ie, for example, 6.0). However, this time an exception was made for the graphical stack. There was a change of Mesa 3D to version 7.11 and X Server on 1.10. Developers have changed their rule in order to be able to update graphics drivers for AMD, Intel and NVIDIA. Release Notes also includes support for Intel graphics for the next generation of chipsets. Probably meaning the platform Ivy Bridge, the appearance of which is expected in 2012. The updated Synaptics touchpad driver adds support for multi-touch technology.
Red Hat has updated the perf utility used for performance monitoring, and its kernel components, where, among other improvements, cgroup support has been added.
File systems
Improved scalable file system XFS supported within the Scalable File System add-on. Optimized processing of loads associated with intensive processing of metadata, which are, for example, when accessing a directory with a large number of small files. Previously, such operations could greatly affect performance.
As a technology preview , the new release includes Parrallels NFS (pNFS) as part of NFS 4.1. This technology is designed to significantly increase the speed of access to data through parallel data processing.
Thanks to “Lazy Inode Table Initialisation”, such operations with the Ext4 file system as disk partitioning can now be performed much faster, since Initially, only the most important file structures are created during the formatting process (and the rest later). Asynchronous data writing allows you to write data to the CIFS file system up to two times faster than before.
Network
Now fully supported work with cluster Samba (CTDB) on a cluster file system GFS2. The High Availability Addon (HA) allows you to support HA configurations for applications running VMware virtualization on RHEL 6.2 guest OS.
The newly added Stream Control Transmission Protocol (SCTP) to support multiple addressing allows hosts on a network to be accessible under different IP addresses. The RHEL 6 kernel now has a sendmmsg system call, which allows a whole batch of messages to be sent to use one call, which reduces the overhead of system calls and increases network throughput. Also, to increase bandwidth in RHEL updates, the Transmit Packet Steering (XPS) feature is included.
Virtualization
Added a number of new features related to virtualization, which will not be superfluous for Red Hat Enterprise Virtualization (in RHEV 3.0 the same core as in RHEL 6.2).
Starting with version 6.2, RHEL will use “virtual CPU time sharing” for guest KVM systems that have access to multiple processors. This technology allows guest systems to use the resources allocated to them more efficiently and, therefore, work faster. This is especially advantageous on large systems with multi-core processors, they have the problem of quality and performance, which are reduced due to the problems of preferential access. This is usually associated with locks that provide exclusive access to data structures. However, it is worth noting that this functionality requires hardware support. At AMD, it is called “Pause Filter”, and at Intel “Pause Loop Exiting” (PLE).
Red Hat developers have increased the speed of processing UDP data running through virtio-net drivers; RHEL 6.2 will more efficiently handle network packets that are smaller than 4 KB. Optimization of network communication "I / O path" are designed to improve performance. Red Hat also points out that the speed of macvtap and vhost network technologies has increased. However, they are still Technology Previews and, therefore, are not included in the support contract. The same applies to live snapshot support, which allows the guest system to save a file system image while running — for example, to back up data.
On NUMA systems, using libvirt, you can now not only determine the processors used, but also the memory areas for the guest OS. Ideally, this allows the use of memory, which is most accessible to applications running on a NUMA node. Qemu, which is used for KVM, now also offers USB 2.0 emulation. When working in a paraverted Xen guest OS, RHEL 6.2 supports ballooning (“swapping”) of memory, a feature that allows you to adjust memory during program execution.
Also recently, Linux Container (LXC) applications have become supported. This allows you to separate applications from the rest of the system and regulate the use of resources using cgroup and namespace. This functionality can be enabled through the libvirt API and virt-manager; however, it also has the Technology Preview status at the moment.
More details?
More information about these and other new features of RHEL 6.2 can be found in the Red Hat documentation. A quick tour is available in Red Hat Enterprise Linux 6.2 - What's New . Complete information can be found in the Release Notes , and for those who want to dig even deeper - Technical Notes .
Source: https://habr.com/ru/post/142518/
All Articles