Red Hat Inc. kicked off its annual Red Hat Summit by showing storage customers a major new Ceph release and a Gluster update more tightly integrated to work with application container technology.
Red Hat Ceph Storage 2, due later this summer, features object storage enhancements, an updated storage management and monitoring system designed to improve ease of use, and a technology preview of the long-awaited Ceph Filesystem (Ceph FS).
Ranga Rangachari, vice president and general manager of storage at Red Hat, called the new product the most significant update to Red Hat Ceph Storage since the company acquired Inktank in 2014. Inktank sold a commercially supported version of open source Ceph software -- much the same as Red Hat does -- on a subscription basis, with a variety of open source technologies.
The Red Hat Gluster Storage 3.1.3 update, due later this summer, is integrated with Red Hat OpenShift Container Platform, which was formerly known as OpenShift Enterprise. The integration will enable users to serve storage from Linux-based application containers, with the containers and storage running on the same server, eliminating the need for a separate storage cluster, according to Ross Turk, global director of product marketing for storage at Red Hat.
Turk said Red Hat's convergence work represents the third phase of the company's container support. Red Hat introduced a Kubernetes volume driver in November to enable the open source Kubernetes container cluster manager to attach Red Hat Gluster Storage to a running container that needs persistent data. In March, Red Hat made Gluster Storage available as a containerized application to enable users to deploy Gluster inside containers and use the container platform to manage the cluster.
"You can only containerize software. You can't containerize hardware. So, only software-defined storage systems can become containerized like this," Turk said. "The advantage to containerizing Gluster itself is that you can deploy Gluster at the same time that you deploy the applications that need it, and using the same framework."
Ross Turkglobal director of product marketing for storage at Red Hat
Red Hat's Gluster-OpenShift integration will support the expanded use of containers for enterprise production applications that need persistent data storage, according to Scott Sinclair, a senior analyst at Enterprise Strategy Group Inc., in Milford, Mass. The current use case for containers centers on test and development and lightweight applications that need only temporary storage.
"The difference between what Red Hat is doing versus other storage products that are targeting containers is Red Hat is focused not only on the end results, they're focused on the full application development cycle," Sinclair said.
Sinclair said Red Hat has a strong history of working with developers. The open source nature of its storage products is easy for developers to work with as they transition container-based applications from test and small-scale environments to production use, he added.
"The ability to develop to and then deploy and leverage the same type of storage offers advantages, because then you don't need to redesign everything as [you] move from [a] test sandbox," Sinclair said. "As we transition from dev to ops, we want the apps to continue to work the same way and not have to recode it."
New Red Hat Ceph 2 capabilities
Red Hat Ceph Storage 2 is based on the Jewel release of the Ceph open source project. Ceph is designed to provide object, block and file-based storage under a unified system that runs on commodity hardware and scales out through the addition of server nodes.
Object storage enhancements in Red Hat Ceph Storage 2 include support for Microsoft's Active Directory and Lightweight Directory Access Protocol (LDAP) authentication -- adding to the previously available OpenStack Keystone -- improved compatibility with Amazon Simple Storage Service and OpenStack Swift APIs, and a tech preview of an NFS gateway for the import and export of object data in bulk.
Another key object storage improvement is global clustering with a single namespace. Turk said users in the past would connect to the closest Ceph cluster, and to read or write an object, the cluster would direct the request to the cluster where the object was stored. New Ceph 2 capabilities will enable the clusters to synchronize, so users can read from and write to any copy of the cluster, Turk said.
"This release definitely moves Ceph closer to the proprietary products in this space, such as IBM's Cleversafe and Scality," ESG's Sinclair said. "Absolute table stakes for some organizations is the integration with Active Directory or LDAP."
Red Hat is adding the ability to replicate virtual block devices across regions for disaster recovery and archival purposes. Red Hat previously offered the ability to take a snapshot of a block device and apply it, but the system didn't automatically send it across the wire and do the orchestration, Turk said.
New console eases management
The new Red Hat Storage Console aims to address ease-of-use issues with a new interface to manage Ceph. The console includes automation technology that Red Hat acquired last year from Ansible. Turk said the prior interface, called Calamari, was good for monitoring and providing problem alerts in a cluster, but it didn't manage full cluster lifecycles. Unlike with Calamari, the customers can spin up or shut down clusters through the new interface.
So far, Ceph usage is largely object and block storage. Red Hat did not consider the POSIX-compliant Ceph FS ready for production deployments, Turk said.
With Ceph 2, Red Hat is making available a driver to enable customers to use its Ceph storage with the OpenStack Manila file-share service. Manila is an abstraction layer that coordinates the provisioning of the shared file system service for OpenStack users, Turk noted.
Turk said Ceph FS has been in development for about 10 years. He said two of the major differentiating features of Ceph FS are its snapshot system, which can be managed from within the file system, and its dynamically scalable metadata server.
Red Hat also previewed its BlueStore storage media back end at the summit. BlueStore allows Ceph to talk directly to the physical storage media without requiring an intermediary file system, according to Turk. He claimed the technology could provide a performance boost of two times to three times in Ceph. BlueStore supports modern media, such as flash and key-value drives. It replaces FileStore, which was designed for hard disk drives.
Red Hat Ceph upgrade long planned
Red Hat teased better Ceph management
Ceph and Gluster expected to stay separate
Red Hat rebrands Ceph and Gluster