Having an increased number of remote users accessing applications on fewer servers can introduce significant bandwidth and latency issues. Applications are typically designed to operate in LAN environments, and may not function well when accessed via WAN. In this article, we'll look at using WAN acceleration technologies to address latency issues, types of WAN accelerators, and key issues when deploying WAN accelerators.
Overcoming WAN Performance Problems
To overcome WAN performance constraints and address latency and delay issues many have turned to WAN acceleration solutions. WAN accelerators speed the delivery of applications by eliminating redundant transmissions, staging data in local caches, and compressing and prioritizing data. WAN accelerators generally perform their task via three methods:
Tokenization: Saves bandwidth by ‘remembering’ chunks of data at each accelerator and forwarding tokens as reference points for data previously transmitted, rather than sending the same data over and over. When a token is received, the local accelerator swaps the received token for the referenced data to be forwarded from its local memory. This method works well for most applications with the occasional exception of digital scanner systems.
Compression: Likely the most effective method of acceleration, it compresses the raw data before sending. Data is then uncompressed on the remote machine unbeknownst to the user.
Caching: Data sent to a remote site is cached locally and synchronized at scheduled times, thus allowing content to be sent at off-peak times. This method can greatly reduce the amount of data sent over WAN links during normal work hours. Having data stored locally speeds its delivery to remote offices, and can allow them to function even if the WAN is down. When WAN service is restored, there can be issues with recompiling the newly cached data at each remote site with the core data.
The type of acceleration used will depend upon your goals, the applications you're optimizing, and network devices and configuration. For example, if you have implemented Quality of Server (QoS) for WAN prioritization, you'll want to understand whether the WAN accelerator will impact application QoS settings.
Deployment Preparations
Now let's look at five key considerations your network team will want to make before rolling out any type of WAN optimization device.
1) Define the applications traversing the WAN and identify the underlying protocols and codecs. You'll want to understand how each of these protocols is impacted by different types of acceleration. In the case of VoIP, for example, g.711 codec's quality will be impacted when traversing the WAN, whereas g.729 would only be minimally affected.
2) Understand which stations are communicating across the WAN and their locations. Ensure that equipment and stations are placed correctly for optimizing WAN performance. If you have multi-tiered applications with web frontends interfacing with SQL servers accessing a database to pull objects, it is best to keep the SQL packets local to one network and not flooding the WAN with 80-byte packets.
3) Define baseline measurements for application utilization and performance, including utilization throughout the day, application response times, and when operations should occur. Are backups taking too long, or occurring during peak demand times?
4) What is the actual latency for WAN links? It's important to understand whether latency is significant enough of an issue to justify WAN acceleration. A rule of thumb is if latency is above 45 ms, it is worth considering WAN acceleration.
5) Optimize applications by pinpointing delay locations. The WAN is an easy target to blame, however, the backend core processes are frequently the more likely culprit. Issues experienced by users when accessing the WAN may actually be attributed to the LAN, but due to proximity and bandwidth issues they may not be easy to detect. Acceleration in this case may help but not be as effective in resolving delay issues as making configuration changes at the core.
In Conclusion
As more users access applications from remote locations over the WAN, it will be critical for you to ensure a positive user experience. WAN accelerating technology is an attractive way to enhance application delivery while reducing bandwidth needs. The key to successfully deploying WAN optimization technology is to be aware of how acceleration will impact application delivery and quality and understanding the source of delay to improve troubleshooting. In the Tech Spotlight, we'll discuss using Observer® to prepare the network for a WAN acceleration deployment as well troubleshooting WAN performance problems |
