Realizing the benefits of cloud storage

Despite some concerns over the cloud’s ability to serve as a reliable and cost-effective storage solution for business and consumer applications, we are seeing a shift toward it. The equipment and services markets are growing rapidly for cloud storage services. For example, the Taneja Group estimated the total cloud storage hardware market in 2010 was $3.2 billion, growing 31 percent per year to $9.4 billion by 2014. And 451 Research calculated the cloud storage services market was $769 million in 2010, growing 47 percent per year to $4.4 billion by 2014.

The rationale for such growth curves is certainly worth exploring. A study from Forrester points to a significant cost advantage of cloud storage solutions over traditional internal enterprise storage. In that study, the annual cost of cloud storage for 100Tb of data was estimated to be $231,600 vs. $955,500 for internal storage, a 76 percent reduction in storage costs.

Some media and entertainment applications could benefit from the lower-cost cloud storage model because the creation, distribution and conversion of video content in digital form is a huge driver of storage and corresponding data protection.

Media and entertainment is seeing an unprecedented growth in storage due to higher-frames-per-second (fps) cameras and video resolution. For example, fps has increased from a historical 24fps to 48fps in the upcoming film The Hobbit by Peter Jackson, to 60fps in the latest Avatar movie, with increases to 300fps expected in the near future. At the same time, digital video production continues to increase resolution from 2K initially to 4K today, with 8K and 16K distinct possibilities for the future. 

As a result, the raw video content produced by high-resolution digital production can quickly get into the Tb/h — even more for stereoscopic capture from multiple cameras. For example, Coughlin Associates predicted a more than fivefold increase in entertainment-industry storage between 2012 and 2017, and will reach more than 87 petabytes (Pb) of storage shipped per year. Even when only considering cloud services for vaulting capacity to complement internal data centers, this represents a significant cost-savings alternative.

Data protection

Data protection and archiving measures are more complicated because of this massive video data generation. Data protection typically involves making copies of the original data, and ensuring copies are distributed, both across multiple disk arrays and geographically. This ensures the survivability of the vaulted content and its availability whenever it is required. Typically, nonlinear editing (NLE) post-production storage is a mix of direct and network-attached local storage. Coughlin Associates recently found that 15 percent of participants are now using cloud-based storage in their post-production, and around a quarter of these had more than 1Tb in the cloud.

Recent service outages from cloud providers, such as Amazon EC2, Google and Microsoft, only reinforce the point that even well-tested backup plans can occasionally fail and cause downtime. Although these are becoming less frequent as the cloud model matures, the basic data protection practice of multiple data sets in multiple locations should still be deployed for maximum reliability. Failing to follow these principles can lead to major production delays when the cloud is used in a collaborative workflow and negatively impact the sale and distribution of content to downstream users.

Although the cloud services discussed until now are applicable across various industries, from media and content production to financial and government-use cases, let’s now explore different use cases of the cloud for the media and content production and distribution industry and their networking implications. Focusing on cloud storage, let’s consider three typical use cases: backup/archive, content and I/O. Among the three, each is defined by different parameters, such as I/O specifications, including bandwidth and response time; geographical diversity; annual downtime; and recovery point and recovery time objectives (RPO and RTO).

In the case of backup or archive storage, one can consider how such an approach can help the industry meet vaulting requirements of raw and/or produced content. Given the enormous amount of storage required for applications such as sporting events footage or the next blockbuster hit (tens or hundreds of petabytes of data), the cost savings of cloud storage are extremely attractive. In this case, characteristics such as geographical diversification and recovery consistency and assurance would typically have more influence in the service definition than the bandwidth and response time.

Content storage, on the other hand, could be characterized by higher availability and I/O rates. It also benefits from a certain degree of geographical diversification for storing and distributing large files. Distribution of produced content, such as television programming, movies, trailers and clips to various partners, thus, becomes simplified and more cost-effective. Higher availability here ensures content can be sold and distributed when needed, and in a timely manner with sufficient bandwidth.

A third use case for cloud storage would be better served by I/O storage. Optimized for performance with greater bandwidth, reduced response time and highest availability specifications, with less emphasis on geographical diversification, this type of cloud is primarily geared toward active development. This would be ideally suited to content ingestion and post-production workflow collaboration. When we consider the size of individual video frames (roughly 50MB for 4k frames), this translates into high bandwidth requirements, whether we consider real-time content ingestion (from 1.5Gb/s to more than 12Gb/s for current formats) or faster-than-real-time transfer of stored content using 40Gb/s or 100Gb/s transmission rates. The low-latency characteristic of this type of storage also makes it suitable for online post-production work.

Although the compute offering of the cloud creates an interesting alternate and/or complementary solution for the media and content industry, such as transcoding and watermarking, it relies on the same type of networking infrastructure used for other parts of the cloud offering.

Network architecture

For this potential to be fully realized, however, the various types of cloud services described in the preceding paragraphs can deliver only the performance the underlying network architecture will allow. The foundation of this architecture starts with a high-speed, high-capacity network that can withstand the enormous amount of data being uploaded, edited and retrieved from the cloud. Whereas 2.5Gb/s or 10Gb/s speeds were deemed plentiful less than a decade ago, today’s applications benefit from 40Gb/s and 100Gb/s coherent networks.

The traffic flows must also be optimized to ensure latency-sensitive applications are prioritized and routed accordingly through a service-aware architecture. For example, the speed and reliability of the network is of utmost importance for news editors and producers who must edit hours of raw footage into a 30-second spot for the next newscast. In this case, not only does the cloud need to be available and reachable, it needs to be accompanied by a highly resilient network infrastructure offering five or six “9s” of reliability (99.999 percent or 99.9999 percent uptime) as part of the overall network design for business-critical applications using available protection, mesh and control plane options.

In addition to high resiliency demands, today’s networking infrastructure must be able to respond to predictable, and unpredictable, bursts of traffic appropriately with scalable bandwidth. If not, video editors, news producers and content distributors will incur serious headaches. In addition to varying bandwidth demands, the architecture must also be flexible to accommodate various types of traffic services — whether Ethernet-based, Fibre Channel — or be able to transport video streams in their raw formats.

Given the high value of the content created and distributed in the entertainment and broadcast industry, and the inherent data protection requirements, it is also important to consider the security aspect when dealing with cloud services. Even though content is viewed as generally secure once in the cloud, (a recent survey from 451 Research found that only 10 percent of respondents considered the data in the public cloud insecure), an important aspect of protecting the proprietary information is considering the security of the in-flight content as it transits to and from the cloud. A networking solution that incorporates data encryption as part of its transport infrastructure alleviates that concern by ensuring the data is kept secure as it traverses the network and complements the cloud offering by securing the data while in transit.

Such features and characteristics of an assured network help accelerate and facilitate access to cloud solutions for a number of use cases in the broadcast industry and, along with the significant cost advantages of the cloud compared to internal data centers, are key considerations when evaluating networking options. A powerful, carrier-class production network with enhanced security provisions, along with cloud-based services, is guaranteed to assist the video production industry in overcoming the quality, latency and security challenges of today’s increasingly distributed file-based workflows.

Patrick Scully is product line manager and video transport expert for Ciena.