Swimming Upstream

By David Strom  |  Posted 2008-02-21

It’s not always enough to read the latest business missive; users want to play it, rewind it and share it. Online video is quickly moving from a lunchtime distraction to a bona fide form of enterprise communication, but the technology is not without its challenges.

The pervasive push of moving pictures brings additional network, bandwidth and storage needs. Understanding how video content is streamed, consumed and stored across the enterprise is vital when it comes to anticipating and planning the necessary infrastructure changes.

All Video Is Not Created Equal

Some Web video consists simply of streaming clips served up from hosted sites such as YouTube. The rest involves creation, editing and distribution of broadcast-quality clips for purposes such as executive briefings, knowledge-management efforts and all-hands meetings. Video production puts the most pressure on IT resources and demands the best planning, according to experts.

Mark Raudonis, vice president of post-production at Bunim/Murray Productions in Van Nuys, Calif., suggests sticking to streaming video and making sure users have relatively new PCs.

“A 2- or 3-year-old PC should be quite capable of handling low- to mid-bandwidth video,” says Raudonis, whose company produces The Real World, The Simple Life and other reality shows, and edits and stores thousands of hours of digital video every TV season.

Other applications require more robust video capabilities, however. At the Center for International Earth Science Information Network (CIESIN) at Columbia University, for example, intense geographic information systems and data mapping needs have administrators employing more gigabit switch ports for their higher-end desktops, and bolstering select client machines with pairs of RAID O SATA drives (separate, redundant disks with alternate sectors containing the data).

The CIESIN effort demonstrates the wisdom of limiting high-end video to users who truly need it, eliminating the need to upgrade the entire network.

At Brooks Health System in Jacksonville, Fla., CIO Karen Green instituted similar controls when moving patient records from videotape to digital media. Limiting video capabilities to authorized users on authorized workstations not only satisfied the strict Health Insurance Portability and Accountability Act (HIPAA) regulations, it “helped us control the use of the content, the quality of the content, as well as the overall impact on the network,” Green says. “We are also looking at whether the business requirement is archival use or immediate retrieval and what impact this would have on the network.”

When determining the impact video will have on your company resources, keep in mind that infrastructure requirements also depend on the quality of the video an organization consumes. How you shoot and store your video—including the number of frames per second—can result in vastly different network traffic profiles and infrastructure requirements.

SAN or NAS: Which Way to Go?

Adding video capabilities to a network nearly always requires adding storage. Even the lowest-resolution video files eat up several hundred megabytes; just a few videos can sap your existing storage capacity quickly.

Enterprises have two options: a storage-area network (SAN) or network-attached storage (NAS). Different uses will motivate different kinds of purchases. If performance is more important than capacity, go with a SAN because it can access data faster than direct storage access and is often easier to expand and manage. But choose a SAN solution designed specifically for video. It may cost more, but the low latency is a must for work with video files.

If you’re seeking a more cost-effective approach, NAS is the answer, especially if you only need to host a limited amount of low-bandwidth content on your network. NAS allows you to share a file with hundreds of clients without designing any complex clustering solutions or making any changes to client software. And a NAS system that can grow into a SAN may make sense if you expect to add a lot of video down the road.

For example, a major Washington trade association is replacing a 4-year-old SAN with a new Network Appliance NAS FAS2020. “Hopefully, this purchase will get us through the next five years with much simpler maintenance than the complex system it’s replacing,” says Adam Kuhn, IT manager for the association. “The SAN was just being used to serve files and didn’t have multiple servers or fibre attachments. It was just a big drive in the sky.”

Kuhn likes the fact that he can triple existing storage capacity and simplify support needs: The FAS2020 can notify Network Appliance when a failure is about to occur. “They can dispatch a replacement part automatically—what could be better than that?” he says. “And there is no Windows Server operating system to maintain, either.”

The organization’s older SAN used SCSI drives, while the new NAS uses SATA drives, which are less costly per gigabyte of storage, Kuhn says.

No matter which method you choose, video storage is likely to grow quickly, because the files are so large compared with most spreadsheets and other documents. To manage this, enterprises might consider a chargeback scheme for the additional storage capacity. “If [users] have to pay for extra storage, they will take the time to consider what files are really pertinent that they need to keep,” says Green at Brooks Health. “Data belongs to the individual business unit, not the IT department, and putting a price tag on it helps to instill some level of ownership.”

Upgrading Network Infrastructure

Simply upgrading storage isn’t always enough. You need to understand the limitations of your operating system to handle very large files and the existing network traffic patterns to determine whether you’ll have sufficient bandwidth to handle video.

While ordinary 10 MB or 100 MB Ethernet will get you limited resolution for a reasonable number of video streams, many corporations are upgrading to 10 GB networks for their heaviest video users, and some are replacing all their network connections everywhere.

“Video definitely stresses your network—you run into latency issues, packet-loss issues,” says Burzin Engineer, vice president of technology for Shopzilla, an Internet-based comparison-shopping service. “Having a 10 GB Ethernet network definitely helps, especially if you have upgraded within the past couple of years.”

Performance can also suffer when video traverses wide-area networks or goes across the public Internet, where there is little control over latency.

In some cases, corporations moving to voice over IP (VoIP) are also considering the latency requirements for handling video when they make these network upgrades. Brooks Health, for example, is installing all new network infrastructure along with its new SAN and is considering multiprotocol label switching (MPLS) to accommodate video and VoIP and provide redundancy.

Many experts suggest employing specialized content delivery network providers such as Akamai for handling the Internet portion of their content distribution. These service providers “have spent millions of dollars to build a network that has the kind of scale and performance you need for these types of applications,” Engineer says.

Another option: Invest in a better storage network and upgrade your Ethernet connectivity for all users.

The Big Picture, in Technicolor

Video also brings infrastructure management challenges for enterprises. And the tradeoffs between better SAN or NAS devices and better network throughput can be vexing, depending on the vintage of the existing storage servers and network gear, and the types of video users consume.

“You want to make sure the experience of the first users is positive,” says Dan Kusnetzky, an analyst with Kusnetzky Group in Osprey, Fla. Indeed, first impressions count, especially when it comes to video.