Storage appliances take cover!

New performance and sizing guidance on Red Hat Gluster Storage with QCT hardware

By Will McGrath, Partner Marketing, Red Hat Storage

I'm an oldest sibling. They say the traits of the oldest include being mature and dependable, while rebellion and excitement are often characteristics of the youngest. Think of the Prodigal Son parable.

Brothers in arms

On the surface, it’s possible to draw a parallel to the Red Hat Storage products. Red Hat Gluster Storage was the first product in Red Hat's quiver. Distributed, POSIX-compatible, scale-out file storage software, Gluster has added several modern architectural features over the years — geo-replication, erasure coding, bit rot detection, self healing and tiering, to mention a few.

Ceph was recently (April 2014) added to the portfolio and garnered attention as the new kid on the block, but also partly because it addresses the rapidly growing object storage market and has been the most preferred platform for OpenStack¹. As younger siblings often do, Ceph tends to hog the spotlight.

This week, elder sibling Red Hat Gluster Storage made a statement about its maturity in the marketplace and open source community, while also reinforcing the message around choice that Red Hat’s software-defined storage portfolio offers to customers.

New performance and sizing guide for Gluster

QCT (Quanta Cloud Technology), whose parent company, Quanta Computer, Inc., is probably the least-known leading server vendor in the world, has worked with Red Hat to produce the industry's first Red Hat Gluster Storage Performance and Sizing Guide.

QCT and Red Hat have performed extensive testing to characterize optimized configurations for deploying Red Hat Gluster Storage on several QCT servers, with the goal of providing a highly prescriptive recommendation to end customers on how best to tailor storage to the demands of their workloads.

A slew of different configuration options were tested:

• Small, medium, and large file operations
• Standard and dense server chassis
• JBOD vs. RAID6 storage layouts
• Replicated and dispersed (erasure coding) volumes
• SSD tiering vs. non-tiering
• Self-healing with different cluster sizes

QCT has gone a step beyond co-producing the 34-page joint performance and sizing guide and created single SKUs, to make ordering much easier for cost/capacity-optimized and throughput-optimized configurations.

The following diagram highlights QCT's naming convention:

(Note: You can find more details on QCT's part-number variants in Appendix A of the Performance and Sizing Guide on the QCT website).

The performance and sizing guide and ready-to-order SKUs from QCT go a long way in cementing the market and thought leadership of open, software-defined storage in a rapidly evolving landscape.

Sister, sister!

While the new performance and sizing guide represents significant value to customers curious about the best server configurations for a particular workload, it also serves as a data point in the overarching message about the ability to control the very guts of enterprise storage, something that traditional appliances generally cannot offer.

Gluster has made waves recently, as a significantly lower cost alternative to EMC Isilon for certain workloads and for being supported within multiple public clouds.

While Gluster has gained traction in the press as providing persistent file storage for containers in Red Hat’s OpenShift Container Platform, Ceph enjoys its status as the preferred block storage platform for OpenStack private cloud builders. Each has their own niche, strengths, and, more important, their own tribe. Gluster and Ceph as upstream open source communities are vibrant and, like young siblings, growing at different rates. Gluster is a larger community (with a larger customer base), while Ceph is growing faster.

Needless to say, they are both equally loved by the Red Hat family, and continue to enjoy strong engineering and marketing focus, to help customers build world-class storage for their applications.

¹OpenStack User Survey (April 2016)