Cisco® Wide Area Application Services (WAAS) allows IT organizations to dramatically improve Citrix performance by optimizing Citrix client-server traffic over the WAN. Citrix allows IT organizations to centralize application deployment, processing, and management from remote-client workstations to a Citrix server farm located in one or more data centers. Each remote Citrix client requires WAN bandwidth-potential significant amounts of network capacity are needed in medium and large offices that rely on Citrix for enterprise applications. The optimizations provided by Cisco WAAS bring dramatic throughput improvements, better response time, and WAN bandwidth savings for remote-office locations. This document examines the optimizations that Cisco WAAS offers and discusses how Cisco WAAS facilitates optimization of Citrix services.
CHALLENGE
Enterprise IT organizations are challenged with centralizing distributed server infrastructure because of the need to maintain compliance, improve data protection, and control costs-without compromising the productivity and experience of the remote-office user. Citrix Systems, through the use of the Citrix Independent Computing Architecture (ICA), helps IT organizations provide complete desktop functions or specific applications to remote users through a terminal connection to a Citrix server in a server farm in the data center. Citrix facilitates server consolidation, because all applications can run on remotely accessible Citrix servers, including Microsoft Office, Outlook, Web browsing, enterprise applications (such as Oracle, SAP, and Siebel), and many others. Although Citrix is highly efficient, each user consumes a significant amount of WAN bandwidth, thereby saturating heavily used WAN links. Accessing applications that are rich in content over the WAN where high latency, packet loss, and low bandwidth are commonly found constraints creates performance challenges for end users and can quickly overwhelm the WAN infrastructure.
With Cisco WAAS and the Cisco Wide Area Application Engine (WAE) Appliances and router-integrated network modules from Cisco Systems®, enterprise IT organizations can now improve Citrix performance and minimize WAN bandwidth consumption.
HOW DOES CITRIX WORK?
Citrix Systems provides server-based solutions that enable the processing of applications to be moved from the client to the server. Citrix Presentation Server (formerly Citrix MetaFrame) is a remote access and application product built on Citrix ICA, the Citrix Systems thin client protocol. ICA is broadly similar in purpose to window servers such as the X Window System, but has a wider scope in that it also provides feedback in terms of user input from the client to the server, and a variety of means for the server to send graphical and audio output from the running application to the client. Citrix transmits high-level window display information as opposed to purely graphical information.
A critical challenge of such an architecture is performance-a graphically intensive application (as most are when presented using a GUI) being served over a slow network connection requires considerable compression and optimization to render the application usable by the client. The client machine may be a different platform, and may not have Windows GUI routines available locally-in this case the server must send the actual bitmap data over the connection. To date, such challenges have been met with only partial success, a fact that might explain why the widely predicted switch to an application server model over the Internet has not occurred.
Citrix environments do not usually have print services in the remote office, but most remote offices include printers and have users who require local printing capabilities. Therefore, every time a document is printed inside a Citrix client session, the rendered document print job is spooled over the WAN through the terminal session, and then sent back from the data center to the user's local printer. A spooled print job is often 10 times the size of the original document, and many Citrix client users end up having a significant portion of their WAN link bandwidth spent on highly redundant print traffic.
CISCO WAAS PROVIDES OPTIMIZATIONS FOR CITRIX
Cisco WAAS is a multilayer application acceleration and WAN optimization solution that improves application performance over the WAN. Optimization for Citrix and the associated protocols is achieved through the following Cisco WAAS features:
• Cisco WAAS Transport Flow Optimization (TFO)-Cisco WAAS TFO provides standards-based, field-proven throughput improvements for TCP-based applications while maintaining packet-network compatibility and safe coexistence with other network nodes communicating by using standard TCP implementations. TFO terminates TCP sessions locally and transparently optimizes flows that traverse the WAN, thereby shielding communicating nodes from WAN conditions. It includes the following components, each providing specific acceleration for Citrix services:
– Large initial windows-Citrix client connections exit the TCP slow-start phase more quickly and enter congestion avoidance, thereby allowing the connection to more quickly leverage available WAN capacity.
– Window scaling-Cisco WAAS transparently increases the window capacity of optimized TCP connections to allow more data to be in transit, thereby improving Citrix throughput.
– Advanced congestion handling-Through intelligent handling of congestion scenarios, Cisco WAAS can more efficiently retransmit lost data when necessary, and return to higher levels of throughput on the network much more quickly, resulting in better Citrix client-server and application performance.
• Cisco WAAS Data Redundancy Elimination (DRE)-DRE is an advanced form of network compression that allows Cisco WAAS to maintain a database of data that has been seen previously traversing the network. This information is used to remove redundant transmission patterns from the network. For repeated patterns, only instructions need to be sent, and the original message is rebuilt in its entirety by the distant Cisco WAE, providing significant levels of compression and ensuring message and application coherency because the original message is always rebuilt and verified by the distant Cisco WAE. Because DRE is application-independent and bidirectional, it is effective regardless of the application being used or direction of traffic flow. Therefore, data patterns that are identified for one application protocol can be used by other applications, and patterns that are identified for one direction of traffic flow can be used to remove redundancy for traffic flowing in a different direction. The Citrix ICA protocol does not eliminate redundant data, meaning that when a user requests previously viewed screen content the whole set of data is resent. With DRE, viewed screen content is stored as previously seen transmissions in an application-independent format, and if redundant segments are seen (that is, content or part of content that has been viewed before), significant levels of compression can be achieved. Data redundancy is especially beneficial when many Citrix ICA sessions are running concurrently over WAN links and when users are viewing content or parts of content that they have viewed before. DRE can eliminate up to 99% of redundant network traffic and provide up to 100:1 compression.
• Persistent Lempel-Ziv (LZ) compression-Persistent LZ compression is a standards-based compression coupled with a long-lived compression history for the connection that can be employed to minimize the amount of bandwidth consumed by a TCP flow. It can be used in conjunction with DRE or independently, and it can provide from 2:1 to 5:1 compression based on the application and data being transmitted. This level of compression is especially helpful for data that has not been previously seen and suppressed by DRE.
The Cisco WAAS default policy includes a policy to apply the full suite of optimizations to Citrix ICA traffic. Although Cisco WAAS can optimize Citrix, additional performance improvement can be gained by configuring the Citrix server for login encryption only (no session encryption) and no compression. Cisco WAAS advanced compression (DRE and persistent LZ) provide far better compression levels than can be realized with native Citrix compression.
CITRIX OPTIMIZATION PERFORMANCE
When a Citrix client accesses an application through a Citrix Server, Cisco WAAS applies optimizations to improve performance and reduce bandwidth usage, thereby providing a faster application response time with less WAN usage. Optimized for WAN environments, the Citrix ICA protocol includes functions to queue mouse movements, cache bitmaps, and compress traffic. However, Cisco WAAS brings additional improvements through WAN optimizations, including DRE, persistent LZ compression, and TFO, as described previously.
By employing Cisco WAAS for optimizing Citrix, IT organizations can realize 85-percent or better bandwidth savings as compared to uncompressed Citrix flows, and up to 50-percent better bandwidth savings than what Citrix provides with native compression. In a test performed on a Citrix system with Cisco WAAS, a Citrix client and server were deployed in separate physical locations and separated by a T1 WAN connection with 80 ms of round-trip latency, as shown in Figure 1.
Figure 1. Citrix Deployment with Cisco WAAS
This environment included Citrix Presentation Server Version 4 running on Windows Server 2003. The applications being accessed were hosted on Microsoft Windows Server 2003 with Internet Information Services (IIS) Version 6. Applications tested included standard Microsoft applications such as Microsoft Word, Excel, Web content, and Macromedia Flash Player. The client workstation was a standard Windows XP Professional SP2 client with Citrix ICA Client Version 9. Figure 2 shows the performance improvement provided by Cisco WAAS as compared to native Citrix access (uncompressed versus Citrix compressed versus Cisco WAAS compressed).
As the results show, Cisco WAAS delivers a significant reduction in uncompressed Citrix network traffic, and far more compression than can be realized with native Citrix compression. In addition, Cisco WAAS provides a perceived improvement in overall performance and stability. Citrix sessions optimized generally seem noticeably smoother with less wait time for operations such as screen refreshes, while providing a reduction in lag and jitter. As shown in the figure, Cisco WAAS can provide up to 90-percent compression (10:1) for Citrix ICA traffic, which can be as much as 75-percent better (4:1) than what native Citrix compression provides. Cisco WAAS also provides additional optimizations for Citrix not directly related to bandwidth consumption:
• Stability for client ICA connections-When Cisco WAAS optimizations are employed, response times are stabilized, predictable, and less susceptible to influence of network load or packet loss.
• Normalized bandwidth consumption-Not only is bandwidth consumption minimized because of Cisco WAAS compression, but usage remains more linear and controlled.
Furthermore, the Citrix ICA protocol does not optimize TCP/IP packet flows, resulting in an unusually high level of unnecessary 64-byte acknowledgement packets, and the protocol does not efficiently use the TCP over WAN. Citrix ICA does not provide data redundancy elimination functionality, so when users request previously viewed screen content, all content is resent. With Cisco WAAS, Citrix ICA benefits from the following unique optimization technologies:
• TFO improves the ability of the client and server to more efficiently communicate, mitigates the effect of WAN conditions, and allows for more efficient use of WAN resources.
• DRE learns Citrix ICA and other application traffic patterns and stores them locally to eliminate redundancy from future transmissions; it identifies repeated sequences even within the transfer and suppresses them. When a user requests previously viewed screen content (even if only previously seen by another user), the repeated screen content can be safely suppressed to minimize bandwidth consumption.
• Persistent LZ compression minimizes the size of all messages being exchanged, even those that have been optimized by DRE.
Cisco WAAS optimizations result in superior Citrix performance, noticeably smoother sessions and application stability, less wait on screen "painting", and a noticeable reduction in lag and jitter.
SUMMARY
Cisco WAAS provides the tools necessary for IT organizations to safely use Citrix servers in the data center and provide Citrix ICA acceleration to remote-office users, thereby improving the performance of Citrix remote clients in WAN environments, providing a smoother session feeling to the end user, and reducing network traffic by 60 to 75 percent. By applying intelligent optimizations such as LZ compression, TFO, and DRE to the Citrix ICA protocol, Cisco WAAS provides remote-office users with a LAN-like experience when working with Citrix servers.
