Riverbed Steelhead Mobile

The Mobile appliance is easy to deploy as you use its web interface to configure a hostname, an IP address for the primary interface and set up the second network port for dedicated management access. Acceleration policies determine what traffic is to be optimized and that which will be excluded and for the latter you can specify IP addresses, ranges, ports and protocols.

Endpoint policies specify what settings are to be applied to the Mobile client. The client is preloaded as an MSI package and you can decide whether it should run transparently or be accessible from the System Tray, how much physical space is allocated to the local Data Store and which Mobile appliances it should contact.

For testing we used the lab's resident Network Nightmare WAN simulator (www.networknightmare.com) and configured it for a 1Mbps WAN link with a 40ms latency. We placed Windows XP SP2 clients on one side of the WAN simulator to represent remote users and on the LAN side we deployed Steelhead Mobile and Steelhead 100 appliances.

For performance testing we employed a standalone Windows Server 2003 R2 system on the LAN running file sharing, FTP and mail services. We opted for a standalone server as Windows domain controllers have SMB signing switched on and experience has shown that this can have a massive impact on file transfer performance.

To test optimisation we used a 4.9MB Powerpoint presentation, which we copied and emailed to and from the test server, opened remotely on the client and saved a modification and used FTP to transfer it as well. These tests were run twice - once with optimisation switched off and then with it turned on.

On the first run remotely opening the file took 58 seconds and saving a small modification took 50 seconds. With optimisation turned on these times were reduced to only 10 and 8 seconds respectively. Mailing the file as an attachment to the server took 186 seconds whilst sending the same file to the client took 59 seconds. With Mobile in action these times were reduced to only 37 and 36 seconds.

Simple file copy operations saw big improvements with the presentation copied from client to server and back again in 46 and 49 seconds and only two seconds in both directions with optimisation activated. FTP operations were similarly affected as the FileZilla client utility copied the test file to and from the server in 42 and 43 seconds and only 3.5 and 2.5 seconds with optimisation on the case.

We also found the Mobile appliance's management interface provides excellent reporting facilities. You can see all connected clients, view total LAN and WAN traffic and check on the traffic reduction percentage users are enjoying. Reports are provided for general bandwidth optimisation, pie charts show traffic summaries for selected endpoints and you can apply date ranges to each one. With access to the client utility granted, users can view optimisation statistics, change the size of their Data Store and turn off optimisation if required.

The main drawback of the Steelhead Mobile is the fact that it isn't a standalone solution and is aimed primarily at sites already using Steelheads. Those new to Riverbed will need to factor in the extra cost of the appliance but it's clear from our tests that this solution delivers in the performance stakes with all our file transfer tests showing a marked improvement with optimisation activated.

Email to a friend

Print this page