The WOC: WAN Optimization in a Box
As demand for bandwidth due to video and cloud begins to outstrip supply, WAN optimization offerings from the likes of Cisco, Riverbed, Citrix are growing up to meet the need.
In today’s business environment, the network is arguably more important than the physical office. Networks are the glue that hold businesses together, particularly when operations expand beyond the boundaries of any one physical space.
Whether a business operates multiple branch offices, supports a mobile workforce, or utilizes off-site computing resources from storage to application serving, the network that binds these distributed nodes is the WAN.
In the “old” days the traffic that flowed over the LAN was relatively light and often text-based. But as network activity expands into data archiving, voice, and multimedia, the links between WAN nodes (i.e., LANs) are becoming increasingly squeezed.
With increasingly high contention for WAN resources, businesses are left with two options to prevent or improve slowing performance and diminished productivity:
- Invest in faster WAN links - Increasing bandwidth across private links is an expensive proposition, particularly for a widely distributed business with many nodes. Regardless, more businesses now rely on mobile workforces whose bandwidth is limited by providers and technology outside the business’ control, such as 3G or 4G connections or satellite links in remote areas or developing nations.
- Improve network efficiency – Simply put, send less data across the existing bandwidth, while maintaining the same level of (or more) functionality. This is where WAN optimization solutions fit into the picture.
In short, WAN optimization boils down to “technologies that actively control and modify traffic traveling across individual WAN links,” according to Jim Frey, managing research director at EMA. Improving the efficiency of these network links is the job for a WOC, or WAN optimization controller.
The original generation of WOCs were discrete hardware appliances co-located at either end of the link. For example, if Acme Corporation is headquartered in NYC with branch offices in Houston and LA, their WAN might look like this:
Segment 1: NYC<->LA
Segment 2: NYC<-> Houston
Three WOC appliances would optimize the entire WAN -- one each in NYC, LA, and Houston.
Today, WOCs are not limited to discrete appliances. In some cases, they may be blade units installed in a data center, or purely software-based virtual appliances running on an existing server.
Traditionally, most WOC deployments are symmetrical, with a WOC at each end. But there are also asymmetrical deployments where a WOC is located only at the head end. An asymmetrical WOC deployment is typically used in a specialized scenario, such as an organization where the distributed workers or offices are highly mobile or located in places with limited infrastructure. In other words, where there is no location for a co-located WOC.
Less data, more speed
What does a WOC actually do? Well, it depends. There are actually many strategies available for squeezing efficiency out of network communications. Depending both on an organization’s needs and the capabilities of a particular vendor’s WOC, just which strategies are used can vary.
Every WAN possesses several key areas of inefficiency:
- Many network applications require numerous data transactions between client and server, even to accomplish relatively simple tasks. Some common network protocols like Microsoft CIFS are especially “chatty” in this regard.
- Many network requests relate to a small number of objects. In other words, out of 1,000 network requests, more than 50 percent of transactions might retrieve the same 10 percent of resources repeatedly -- documents, graphics, databases, etc.
- Network transactions are not all created equal. For example, time sensitive activity like VoIP requires bandwidth more urgently than a large file copy.