Added Erasure Codes and hierarchical deployment for InktankCeph storage of RedHat

RedHat recently released version 1.2 of its InktankCeph Enterprise Edition Software, which features code erasure and cache layering and Updates tools for managing and monitoring Distributed Object Storage clusters. This release is the first product update of RedHat after the acquisition of InktankStorage in December. Focus on Cloud, backup, and archiving. InktankCeph Enterprise Edition (ICE) combines open source Ceph software to support object and block storage, Calamari monitoring and management tools, and product support servers

RedHat recently released version 1.2 of its Inktank Ceph Enterprise Edition Software, which features code erasure and cache layering and Updates tools for managing and monitoring Distributed Object Storage clusters.

This release is the first product update after RedHat acquired Inktank Storage in December. Focus on Cloud, backup, and archiving. Inktank Ceph Enterprise Edition (ICE) combines open source Ceph software to support object and block storage, Calamari monitoring and management tools, and product support services. RedHat's ICE 1.2 software-defined storage comes with commercial support products that are consistent with the latest open-source Ceph storage software Firefly, two key new features, namely code correction and deletion and hierarchical deployment, have already produced benefits.

According to Dan van der Ster, A Ceph Service Manager and storage engineer in the IT department of the European Nuclear Research Center, the Inktank client, CERN, a Geneva-based European nuclear research organization, correcting and deleting codes is an important function.

"This is the only way to build a persistent and affordable multi-byte cluster," van der Ster emailed. "However, the complex data involved causes a very significant performance loss. Therefore, we are very happy to test the new pool-tiering Ceph feature-currently available features on the Firefly release-to see the combined solution (Correction and deletion code plus layer) is it feasible for our block and physical data users." Red Hat recommends the new cache-tiering feature for customers who want to correct and delete code and high-speed performance. This feature keeps the most popular data in high-performance media while cold data in Low-performance media, ross Turk, managing director of Ceph marketing and community, said.

"This allows you to open a pool in Ceph and convert it into a read cache or write-back cache for another pool," Turk said. "If you have a very intensive support pool for de-encoding, it may not be very fast, but it is very cost-effective, you should put the SSD (SSD) cache pool in front of it, this will ensure that you can quickly access the most popular data." The default erasure library of Ceph is Jerasure. When the Administrator creates an erasure-coded backend, the data block and code block parameters can be specified. The default setting of the ICE Erasure code is 2 + 1, which means that the system divides the data into two parts, creates an additional block and stores them on three object storage devices.

However, Turk says the company does not want most people to use the default settings in production. He said that the 12 + 2 erasure-coded pool may become a more common choice, providing good data distribution, tolerating faults on two nodes, and reducing storage overhead.

Turk points out the potential economic benefits of code erasure because ICE pricing is based on capabilities. He said that the customer can store more data with the same initial capacity. In this case, the system needs more time and processing capabilities to compute the data to be restored compared to the Replicated Copy.

"When you design a storage cluster, you must always provide excess capacity. If I want to put 100 GB of data into a storage cluster, usually through Ceph, I need to buy a hard drive worth 300 GB, "Turk said. "This proportion is reduced," he added. The Ceph erasure code is at the pool level rather than the cluster level. Therefore, the customer may have the erasure-coded pool and the replication pool in the same cluster, Turk said.

Ashish Nadkarni, head of storage systems and software research at the IDC, believes that Red Hat may eventually need to provide Erasure Codes at a finer granularity, however, the new ICE 1.2 feature throws Red Hat into competition with other OSS vendors.

"This is not the main thing, mainly because it is on the right track and is better suited to other OSs competitors," Nadkarni said. "Since Red Hat has purchased them, they are likely to accelerate the startup cycle. They want to ensure that the features are more enterprise-oriented and OpenStack takes precedence over others ."

Another major new feature of ICE 1.2 is the enhanced Calamari management feature. Network-based software includes a dashboard for user identity check; Performance Statistics of each disk/pool measurement IOPS changes over time; and a diagnosis workbench; and tools for monitoring disk usage and management, adjusting clusters, pools, devices, and OSD settings.

Inktank Calamari's management tools were originally proprietary, but after Red Hat's acquisition, Calamari was open-source, Turk said.

One of the fragments discarded by ICE is file-based storage support. Turk said Ceph file systems are not mature products. Other features will be available in future versions, including performance improvements, LDAP and support for iSCSI, VMware, and hyper-v, he said.

ICE 1.2 host operating systems support Red Hat Enterprise Linux (RHEL) 6.5 and 7, Ubuntu 12.04 and 14.04, and CentOS 6.5. Devices supporting connection to the ICE client include RHEL OpenStack Platform 4 and 5, RHEL 7 kernel Rados block device (RBD), Ubuntu OpenStack, and Mirantis OpenStack.

One of the most common uses of Ceph storage software is as an OpenStack cloud storage backend. Turk said that the goal of the Capability-based tiered pricing model of ICE is 1 cent/GB, which is at pb level or above every month, but the cost for smaller devices is a bit high.

For more information about RedHat, see RedHat topic page http://www.linuxidc.com/topicnews.aspx? Tid = 10

This article permanently updates the link address: Http://www.linuxidc.com/Linux/2014-07/104645.htm