The Next Wave of Network Convergence

Jeff Aaron, vice president of marketing for Silver Peak, has more than 16 years experience in the networking and telecommunications space and is responsible for communicating the company's Wide Area Network (WAN) optimization solution.

JAaron

In the late 1990s, something significant happened that forever altered the way we do networking — the Internet Protocol (IP) became the de facto standard for communications. With this convergence came an entire industry of routers and switches, which enabled large, reliable networks (such as the Internet) to be built for anytime, anywhere communications.

It didn't take long after the world standardized on IP for a second wave of convergence to take place.  People realized that it can be expensive to maintain separate networks for voice and data traffic, so they began to merge the two (using Voice over IP). Shortly thereafter, IP-based videoconferencing and video streaming emerged as well, giving real substance to the whole "unified communications" movement.

Today, we are experiencing a third inevitable wave of networking convergence, which is having a very profound impact on businesses and data centers throughout the world. It is becoming increasingly difficult to justify a separate IP network for storage traffic as a way of ensuring the performance and reliability of off-site disaster recovery operations. It is simply too expensive — and no longer needed given advances in WAN optimization. As a result, remote backup and data replication traffic is increasingly deployed alongside one another, and in conjunction with file, email, web, VDI, voice, video, and a plethora of other applications that are critical to business operations.

This is a huge transformation within many companies as it affects the most strategic enterprise applications. In addition, it has an enormous impact on costs by enabling enterprises to save over 50% in ongoing IT expenditures. Lastly, this third wave of convergence can break down silos between storage and networking organizations, which is often appealing to CIOs.

Are enterprises ready to handle this next wave of convergence?  What can be done to prepare?

Challenges of Convergence

There are several challenges that can hamper an enterprise's ability to effectively use a converged network for storage and non storage traffic. These can be broken down into the following areas:

Congestion - Shared WANs will inevitably be over-subscribed during periods of peak usage. This leads to periods of congestion where packets are dropped (or delivered out of order).  This is not an issue when dealing with traffic that isn't time sensitive, like file or email. However, it is very problematic when traffic is interactive, such as virtual desktops, and when a high sustained volume of data is required, as is the case with data backup and replication. The issues associated with packet loss become worse as WAN capacity increases.  High capacity WANs carry a high volume of data. This means that during a given period of high congestion, more data will be adversely affected. It is like hitting a pothole at 100 miles per hour vs. 5 miles per hour. In the first scenario, the damage will be significantly higher.

Latency - You cannot defy the laws of physics.  As distance increases, it takes longer to send data back and forth - a concept known as latency. For many applications, latency will not impact an end user's experience. However, there are many applications that are extremely sensitive to latency, such as SAN extension, voice and video. In these environments, latency must be mitigated to ensure optimal performance.

Bandwidth - On a converged WAN, there are many applications vying for a very limited resource - bandwidth. As data volumes increase, bandwidth becomes increasingly scarce.

It can be quite expensive and time consuming to add more bandwidth to an existing WAN infrastructure. As a result, measures must be taken to optimize the bandwidth that is available to ensure maximum application performance while delaying costly IT expenditures.

Benefits of WAN Optimization

Various network optimization techniques can be employed to overcome the challenges of deploying storage and non storage traffic across a converged WAN.

To overcome the ill effects of congestion, for example, enterprises can use adaptive Forward Error Correction (FEC) to reconstitute lost packets at the far end of a WAN link, avoiding delays that come with multiple-round-trip retransmissions. Similarly, real-time Packet Order Correction (POC) re-sequences packets on the far end of a WAN link "on the fly" to avoid re-transmissions that occur when packets arrive out of order. These techniques enable WANs to easily recover from packet loss due to a variety of network layer conditions, such as queue overflows and constrained bandwidth links.

In addition, Quality of Service (QoS) techniques are popular for ensuring real-time or mission critical applications are prioritized over other applications. Traffic shaping ensures that applications receive the appropriate amount of WAN bandwidth. To be effective, WAN optimization solutions should be able to honor honoring existing QoS markings that may be created elsewhere in a network, plus be able to create new tags based on sophisticated application classification logic with Deep Packet Inspection (DPI).

TCP acceleration helps to overcome latency between source and target locations, which is exacerbated by chattiness in the Transmission Control Protocol (TCP). By adjusting the TCP window size and performing selective acknowledgments, more data can be sent in less time. As a result, TCP acceleration techniques mitigate the impact of latency across the WAN.

WAN de-duplication eliminates the transfer of duplicate information sent across the WAN during data transfer. This is achieved by inspecting all inbound and outbound WAN traffic in real-time, and storing a single local instance of data on each WAN optimization appliance. Prior to sending information across the WAN, WAN optimization appliances compare real-time traffic streams to patterns stored locally. If a match exists, a short reference pointer is sent to the remote WAN optimization appliance, instructing it to deliver the traffic pattern from its local instance. Repetitive data is never sent across the WAN, saving bandwidth and enabling LAN-like application performance.  By working at the byte level, WAN de-duplication typically benefits all IP traffic, delivering an additional 60% to 90% more "virtual bandwidth" across the WAN.

graphic-silverpeak

Graphic courtesty of Silver Peak.

Getting a Rapid Return on Investment

Convergence saves money. But, those savings can easily be squandered if the performance of a shared network adversely impacts application performance.

WAN optimization can help. By overcoming common bandwidth, latency, and congestion challenges within converged networks, WAN optimization maximizes application performance while minimizing IT expenditures.

Just as the convergence of voice and data onto a single network completely transformed the way enterprise IT must function, so too is the convergence of storage and non-storage traffic onto a common WAN. By taking the appropriate steps, enterprises can embrace this trend and ensure that networking evolves with emerging business needs.

Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library.