| Product |
TMX5000 |
NetScaler
Application Delivery System 6.0 |
CN-5080E |
| Vendor |
Array Networks |
Citrix Systems |
Crescendo
Networks |
| Price as tested |
$33,800 |
$52,500 |
$48,000 |
| Pros |
High transaction rates; economical. |
Highly scalable in tests of concurrent
connections, goodput*, TCP multiplexing with long think times;
rich feature set. |
Screaming performance delivers top placements
in most tests. |
| Cons |
Only one interface each for client- and
server-facing networks vs. four for some competitors; compression
cannot be applied selectively; no per-URL filtering.
|
Sensitive to transaction rates in TCP multiplexing
and HTTP compression tests; only two interfaces each for client-
and server-facing networks, vs. four for some competitors. |
No support for caching, URL rewriting or
IP subnet-based access control; routes rather than switches
among interfaces. |
|
| Product |
BIG-IP 6800 |
ServerIron
450 |
DX 3600 |
| Vendor |
F5 Networks |
Foundry
Networks |
Juniper Networks
|
| Price as tested |
$66,000 |
$38,000 |
Single unit, $55,000;
Four-way unit, $220,000 |
| Pros |
Decent performer; rich feature set; good
re-sponse-time reduction; highly scalable. |
High port density; scalable (more than our
results suggest); Cisco IOS-like command-line interface. |
ActiveN feature allows multiple boxes to
work together for scalability and high availability; four-box
configuration is a strong performer. |
| Cons |
In goodput tests, system design requires
hundreds or thousands of users to reach top speeds. |
No support at test time for HTTP compression;
limited visibility for ICMP distributed denial-of-service attacks. |
Limited horsepower on a single box; response
time was sluggish until Juniper tuned memory settings. |
|
|
