Mixed Objects Matter

Scale at Line-Rate Throughput for Demanding Web & Cloud Services   

Customers consistently choose the Scality RING for:

  • Hardware Freedom: Pure software, runs on ANY x86 standard Linux based servers
  • High Performance: Line-rate performance, scales linearly across mixed workloads
  • 100% Reliability: Reliability with minimal intervention at massive scale
  • It Just Works: Production proven in most demanding environments


The adoption of Web & Cloud applications by service providers is driven by cloud economics, low TCO, and technology to support the most demanding mixed object workloads. The Web & Cloud use case verticals cover web application providers, service providers, media and entertainment and financial enterprise.

Scality enables these industries to have a storage solution to help them run their own Web & Cloud storage infrastructure. Dailymotion, Comcast, and many others achieve superior ROI compared to the Amazon S3. They support a wide array of content, focusing on purpose-built applications and services leveraging any industry standard x86 hardware.

This Performance Spotlight highlights why mixed objects and high throughput performance matter for Web & Cloud services.

Scality RING’s performance matches perfectly with Web & Cloud workloads with storage-challenging mixed object sizes. The RING demonstrates that native REST connectors can scale linearly from 300 MB/s to over 1000 MB/s, and are capable of over 6000 MB/s aggregate performance on a six (physical) node system. A very challenging ratio of more than 50% of the workload containing small files was used with a single REST connector to saturate the 10GbE link reaching maximum bandwidth utilization. The high throughput performance of over 1000MB/s was reached while scaling linearly – even across multiple generations of hardware using HDD. This supports the typical growth patterns of Web & Cloud data centers and underscores the RING’s software-based flexibility and performance capability to realize the benefits of high performance with a parallel design.

Web Application and Cloud Service Providers are building architectures similar to Google and Amazon. They are seeking OPEX benefits of deploying their own infrastructure at cloud scale to avoid the hidden costs of using public clouds such as AWS S3, where the bandwidth cost rises faster as the data grows.

Web & Cloud data comes in a wide range of unstructured content; from the small to very large file sizes, anywhere from documents, thumbnails, images, audio to high-res videos. There is a higher ratio of smaller-size file distribution compared to other use cases (see the table on typical application and data size distribution). Key performance indicators using the native REST connector are throughput, bandwidth utilization, latency, consistent performance across mixed workloads, and linear scalability. Other object storage systems cannot be deployed for such type of workloads as they are optimized only for larger (1MB+) object size workloads.

Why choose Scality RING for Web & Cloud services?

  • Simple application integration via a range of storage protocols
  • HTTP REST (S3, Swift, CDMI, Native)
  • File (NFS, FUSE)
  • Limitless storage
  • Linear scalability
  • Scales under load without performance degradation across mixed workloads
  • Lower TCO than AWS S3
  • Flexible, geo-aware platform
  • Use ANY x86 standard hardware

Web-and-Cloud-table

The performance was validated at the Scality lab, and there are more details available upon request. Mixed objects ranged from 32KB to 128MB, and over 50% of the distribution included small files. The read-write workload ratio consisted of 75% read to 25% write. Performance scaled linearly using a fixed number of read-write ratio of threads over an increasing number of native REST connectors.

Web-and-Cloud-chart

Leave a Reply

Your email address will not be published. Required fields are marked *