Coyote Point load balancer sticks to the essentials
Equalizer E550si doesn't match the traffic acceleration features of high-end competitors, but delivers most everything else at a winning price
Coyote Point Systems was one of the first vendors to produce a commercial load balancer in the mid-Nineties, and the company has always offered good basic features at very competitive prices. The Equalizer E550si is no exception. Starting at $10,995, it is not the least expensive load balancer available, but few of the products with lower price tags have 20 gigabit Ethernet ports or many of the advanced features found in the Equalizer. While it does not have all the features such as compression and traffic acceleration that you'll find in an F5 BIG-IP, Juniper DX, or Citrix NetScaler system (click the links for my reviews), it's also much more affordable. It does offer all the features needed to deliver Web-based applications, including excellent local and geographic load balancing, persistent sessions, SSL termination, and sophisticated health checking.
Click for larger view.
The Equalizer is quite capable. You can set it up between the internal and external networks, with one external server port and 20 internal ports, or inside the firewall with both servers and virtual cluster addresses on the same network. You can establish unlimited numbers of virtual clusters, with up to 128 physical servers in each cluster. The E550si should be able to handle up to 8 million simultaneous users (according to Coyote Point). In my testing, I found that the E550si was able to maintain consistent loads on a variety of servers regardless of the number of rules or persistent connections I created.
Exercising the Equalizer
As for my previous tests of load balancers, I tried out the Equalizer E550si by setting up a virtual cluster of several servers running a demo version of an e-commerce site. I then used an Ixia 400T traffic generator and IxLoad software to simulate a large number of users accessing the virtual Web site, then compared the loads generated on each Web site. The Equalizer was able to keep actual loads on the servers consistent even though their processing power varied considerably. I then enabled a number of features, such as SSL and persistent sessions, and attempted to overload the Equalizer by simulating many concurrent users. As with the other solutions I've tested, this was only possible with artificially small sessions. When simulating actual traffic, the gigabit connection became saturated before the limitations of the device were reached.