- Windows HPC hits top 10 among supercomputers
- Ruby on Rails rolls into the enterprise
- Mobile phone chargers are energy vampires
- 10 IT security companies to watch
- Researchers getting the lead out of electronics
HP loaned us two ProLiant DL380 G3 servers for our tests. One server was used as our internal e-mail system. We first loaded Microsoft Windows Server 2003 Enterprise Edition on it, and then loaded Microsoft Exchange 2003.
The second server was used to host the vendor software when needed. Three vendors (Tumbleweed, Entrust and CipherTrust) elected to send appliances. Two of the remaining vendors (ZipLip and PostX) provided software products that we ran on the Windows Server 2003, Enterprise Edition platform. PGP provided a soft appliance, which loaded its own version of Red Hat Linux onto the server hardware.
To administer the systems, we used Firefox and Internet Explorer browsers predominantly, but also used the Linux Konqueror and Mac OS Safari browsers when we could. We installed Windows XP on a Dell Latitude D810 laptop for the one product that required it for their policy editor.
For internal e-mail clients, we used both the full Outlook 2003 client, as well as Outlook Web Access. For external e-mail clients, we sent e-mail to Exchange 2003, postfix on a SuSE Linux system and to e-mail accounts on Yahoo.
For our e-mail performance tests, we used several systems on the same network segment to quickly send 1,000 messages to each product, and the time it took to deliver all messages was recorded. The messages were sent from our Exchange system using a load simulator, which blasted the messages as quickly as possible to the products. The time, as recorded in the header, between the first and last messages arriving at the destination were used to score the products. This eliminated internal queuing delays, and provided a more comparable measure of performance.
The policy rules would identify a keyword, and then trigger an encryption event. PGP was configured to encrypt based upon e-mail address, because it couldn't do content scanning. The test was repeated five times per product, and the results averaged to produce a numeric result. The same block of 1,000 messages was used on each pass for each product to keep the tests fair. Each message was designed to trigger an encryption request.
Back to Clear Choice Test: "CipherTrust tops e-mail encryption field"
Back to Technology Insider: E-mail encryption index
Comment