Red Hat’s offering had never been closer to that of Nutanix and VMware. The publisher has released versions 4.5 of OpenShift, its Kubernetes distribution, and OCS, the OpenShift implementation of the Ceph storage system. Openshift 4.5 now orchestrates virtual machines in addition to containers and enables all machines to be managed via a single interface, as does the latest vSphere 7 Tanzu from VMware and AOS from Nutanix with its carbon extension.
Continuation of the article below
SDS OCS 4.5 then promotes OpenShift as a hyper-convergence solution with the additional option of controlling from the external storage to the server nodes from the same interface. Just like vSAN at VMware, which is now connected to containers via its new vSAN Data Persistence module, and like Nutanix memory modules, which now cover as many types of storage as Ceph.
“In the past, OCS allowed OpenShift to control its local storage in order to assign it to the containers it was running. Since the introduction of OpenShift 4 and the switch to the Kubernetes engine, however, our customers have expressed the need to operate external storage via OpenShift, which became possible thanks to the replacement of the engine from version 4.2 of OCS. Gluster from Ceph. The new version 4.5 of OCS completes this integration with a special tab in the graphical OpenShift administration console and the ability to trigger automatic processes, ”explains Yacine Kheddache, responsible for architecture at Red Hat France.
An interface to manage the entire memory
In the standalone version, Ceph is a safety data sheet that is installed on a cluster of servers with many hard drives to share storage volumes in block mode, file mode, or object mode. Its advantage lies in its performance and high elasticity. However, its mistake would be to be complex too often to be controlled from its own interface. In contrast to OpenShift, which radically simplifies the management of containers orchestrated by Kubernetes. Especially since it offers all interfaces so that developers can provide resources themselves without having to ask system administrators for help.
“The first companies interested in the integration of Ceph in OCS are using OpenShift on the one hand and Ceph storage on the other, because they can finally control Ceph with all of the functions offered in OpenShift in n ‘. You no longer have to jump from one interface to another, ”argues Yacine Kheddache.
However, he states that using a Ceph storage cluster goes beyond sharing it with OpenShift. It is clear that the volumes that are not intended for use with containers but are used for applications running on other servers still need to be managed through Ceph’s internal interface. However, we can assume that Red Hat’s goal is to have these external servers orchestrated by OpenShift as well, as virtual machines can now run.
OCS can manage volumes on internal drives of OpenShift nodes and others on drives of an external Ceph cluster. On the other hand, it doesn’t know how to mount volumes between the two, but instead provides mechanisms for replicating / moving data between the two. In hybrid cloud mode, the OCS console in a local OpenShift can also control the OCS storage volumes of OpenShift instances provided at AWS and provide automatic data migrations.
“At AWS, OCS volumes, regardless of their type, use EBS block storage services. We currently only officially support this service, as we have developed a management of these volumes with extended functions via the AWS APIs to perfect the integration. However, version 4.5 of OCS contains pre-release modules that also make it compatible with Blobs, the corresponding block storage service on Azure, ”said the Red Hat manager.
More security, live migration and faster object storage
“The installation packages are made available on a local server. In many cases, our customers provide this server with updates, which they provide on removable media. “Yacine KheddacheRed Hat France
OCS and OpenShift versions 4.5 contain many more new functions. One of them is to deploy them in “bunker” mode without having to download an additional package during installation. This functionality was developed to meet the needs of Red Hat customers who carry out sensitive activities (army, administrations, manufacturers, etc.) and want their data center to remain hermetic on the Internet. “The installation packages are made available on a local server. In many cases, our customers provide this server with updates, which they provide on removable media. “”
In terms of security, OCS 4.5 offers new end-to-end encryption protocols, particularly FIPS-140, which are very popular in the US government.
On a technical level, OCS 4.5 now supports two servers that simultaneously have write permissions for a volume in block mode. This feature is intended to accompany the arrival of virtual machines in OpenShift 4.5 by enabling the migration of VMs between two clusters, which corresponds to vMotion in VMware and Live Migration in Nutanix.
“It’s interesting to note that we never had to move an application between two OpenShift clusters on the fly because the best practice with containers is to always restart a new copy of the instance. On the other hand, we show here that we can adapt very quickly, ”emphasizes Yacine Kheddache.
More sensitive, OCS 4.5 regains a feature that Ceph lost when it integrated with the OpenShift storage engine: the ability to access volumes in block mode without going through the standard S3 protocol.
“Ceph was originally developed with access gateways to the RADOS protocol. In addition, we have installed an S3 compatibility layer. Using RADOS directly can still prove useful for applications that are not intended to be exported to the cloud, as access is much faster. If this low-level protocol is available in OCS, it can also be installed over any other object protocol, ”concludes Yacine Kheddache and mentions the possibility of migrating applications for older object systems to OpenShift.