What a Difference a Layer Makes
Mar 30, 2011
The application layer of the OSI stack has long been regarded as the focal point for WAN optimization (WANop). Efforts to maximize application delivery existed here because this is where communication for the network is synchronized and all communications between applications is managed.
But the times, they are a changin’.
David Greenfield recently noted over at IT Knowledge Hub, “I think there’s a similar battle going on right now in the WANop space.” He says that historically the WANop market has been led by a single vendor that has focused on optimizing and improving the performance of TCP-based protocols via application layer optimizations. He is quick to add, however, that in today’s market there’s “a growing need to provide broad application support, not simply a few specific TCP applications.”
Greenfield concludes his piece by noting that, “I think enterprises will prefer a WAN optimizer that can improve the functionality of all applications–transparently.”
We couldn’t agree more, which is why Silver Peak approaches WANop from the Network Layer rather than the Application Layer.
Application Layer Limitations
The problem with approaching things from the application layer in today’s climate was best summed up by Asheesh Raina, principal research analyst for software at Gartner, in a recent email interview with ZDNet Asia when she said that, “in the application layer, ISVs will also need to consider which server platforms they want to develop for.”
While some may choose to develop platform specific solutions for WANop, the current IT climate seems to want more. Workforces are becoming not only increasingly dispersed, but are also often accessing data via a multitude of platforms, devices and application versions. This cries out for the type of platform agnosticism that enterprises could only find at the network layer.
Optimize all Enterprise Applications
Of all the reasons for performing WANop at the Network Layer, agnosticism is probably the most obvious because it benefits all current and future enterprise applications as well as their individual versions. Silver Peak was the first vendor to practice deduplication at the network layer and is the only vendor to address bandwidth, latency and loss characteristics in real-time to improve the performance of all enterprise applications.
Not only does working at the Network Layer make the WANop solution compatible with a wide breadth of applications and versions, it also improves the transferability, changeability and robustness of the solution because as enterprises upgrade their systems, whether to new versions or new software, the WANop solution will continue to work for them.
More IP Applications
Unlike other WAN acceleration vendors, Silver Peak does not bypass any IP traffic – if it runs over IP, Silver Peak can help optimize its performance. Again, this is a result of working at the Network Layer, which provides the ability to scale higher in terms of flow count, enabling more traffic to run over the WAN while utilizing less bandwidth.
In addition, Silver Peak is the only vendor to address Network Integrity issues in real-time with its patented Forward Error Correction (FEC) and Packet Order Correction (POC) technologies. Other WANop technologies rely on aggressive TCP retransmission methods to address packet loss, which do not work with time-sensitive applications like voice and video because they add additional latency. Furthermore, these methods only work on TCP traffic, rendering them useless for applications that leverage UDP and other protocols. Lastly, aggressive retransmissions are often quite “unfriendly” to other applications trying to use the WAN at the same time as it consumes all available WAN resources.
The agnosticism provided by working at the Network Layer in and of itself would provide a long-term ROI for an enterprise, merely by not having to customize their optimization solution every time a new version or application is installed on their network. The savings in terms of the time and money spent on IT personnel in and of itself would serve as a good argument for its implementation.
But there are a number of other key ROI drivers including the number of locations that comprise the WAN, the total size of WAN traffic, the nature of the traffic, and the ability to de-duplicate data transmission. Being able to address all these concerns at the Network Layer means they’re handled by a single appliance, which significantly reduces hardware costs, and which in turn reduces costs for power, cooling, and the amount of space required for the physical data center.
Proof is in the Performance
Increased ROI, higher throughput and scalability through agnosticism all make an awfully strong argument for working at the Network Layer rather than the Application Layer. The argument is so strong that the “battle” mentioned by David Greenfield above seems like it should be rather one-sided…that side being the one pioneered by Silver Peak.