Results tables

Measuring price/performance
Hitting the ratios
Measuring cache age
Measuring response time
Measuring downtime

Measuring price/performance We normalized our raw throughput numbers to arrive at a figure that depicts how many requests per second $1,000 can buy. The higher the measurement, the better the price/performance.

Product Requests per second per $1,000 iMimic DataReactor 5000 102 IBM 3500M10 72 Dell PowerEdge 130B 67 Quantex Microsystems WebXL-4000 67 Stratacache Flyer 100 63 Compaq TaskSmart 900 60 IBM 5000 54 Microbits Intelli-App Pizza Box 53 IBM 5600 52 Compaq TaskSmart 2500 47 Squid 37 Swell Tsunami CPX-1000 33 Lucent IPWorX WebCache 100 33 InfoLibria DynaCache X1 cluster 32 Lucent IPWorX WebCache 150a 28 InfoLibria DynaCache 30I 25 Network Appliance NetCache 25 Cisco Cache Engine 590 21 Pionex PCA-1000 23 Cisco Cache Engine 7300 18

Hitting the ratios Hit ratio measures the number of times a cache produces a hit vs. the times it passes the request onto the Web. The higher the hit ratio, the better the performance.

Product Document hit ratio Squid 55.0% Swell Tsunami CPX-1000 55.0% Cisco Cache Engine 590 54.8% Cisco Cache Engine 7300 54.5% Lucent IPWorX WebCache 150a 54.2% iMimic DataReactor 5000 53.2% Pionex PCA-1000 52.9% Compaq TaskSmart 2500 52.7% Lucent IPWorX WebCache 100 51.1% Network Appliance NetCache 50.3% Microbits Intelli-App Pizza Box 44.4% IBM 5600 42.1% IBM 5000 40.8% Quantex WebXL-4000 40.3% Compaq TaskSmart 900 38.7% IBM 3500M10 37.1% Dell PowerEdge 130B 36.0% InfoLibria DynaCache X1 cluster 34.4% InfoLibria DynaCache 30i 32.3% Stratachache Flyer 100 32.3%

Measuring cache age Cache age shows the maximum age an object reaches before it is dumped from the cache. A higher score demonstrates better performance because a device that can store an object longer has a better chance of producing a hit.

Product Cache age Swell Tsunami CPX-1000 19.1 hours Squid 15.6 Cisco Cache Engine 590 11.3 Cisco Cache Engine 7300 8.6 Lucent IPWorX WebCache 150a 8.4 Pionex PCA-1000 7.6 iMimic DataReactor 5000 7.2 Compaq TaskSmart 2500 6.5 Network Appliance NetCache 4.7 Lucent IPWorX WebCache 100 4.3 InfoLibria DynaCache X1 cluster 3.6 InfoLibria DynaCache 30I 3.2 IBM 5000 2.4 Microbits Intelli-App Pizza Box 2.4 IBM 5600 2.3 Quantex WebXL-4000 2.3 Compaq TaskSmart 900 1.7 Dell PowerEdge 130B 1.3 IBM 3500M10 1.3 Stratachache Flyer 100 1.2

Measuring response time To measure response time of these products under peak load conditions, we introduced an artificial delay on the server side to see how each product would react. The lower the measurement, the better the performance.

Product Mean response time iMimic DataReactor 5000 1.35 seconds Squid 1.41 Network Appliance NetCache 1.44 Swell Tsunami CPX-1000 1.44 Compaq TaskSmart 2500 1.45 Cisco Cache Engine 7300 1.48 Cisco Cache Engine 590 1.50 Pionex PCA-1000 1.51 Lucent IPWorX WebCache 150a 1.58 IBM 5600 1.65 Microbits Intelli-App Pizza Box 1.69 Quantex WebXL-4000 1.70 Lucent IPWorX WebCache 100 1.71 Compaq TaskSmart 900 1.72 IBM 5000 1.75 IBM 3500M10 1.78 Dell PowerEdge 130B 1.81 Stratachache Flyer 100 1.89 InfoLibria DynaCache X1 cluster 2.06 InfoLibria DynaCache 30i 3.08

Measuring downtime Time Until First Miss (TTFM) score measures how long it takes a cache to recover from an unexpected condition like a power outage. When a cache is able to deliver its first miss, end users are able to access the Web. The lower the measurement, the better the performance.

Product Time to first miss (TTFM) InfoLibria DynaCache 30i 0.1 minutes InfoLibria DynaCache X1 cluster 0.7 Network Appliance NetCache 1.2 Microbits Intelli-App Pizza Box 1.6 Lucent IPWorX WebCache 100 1.8 Lucent IPWorX WebCache 150a 1.8 Pionex PCA-1000 2.2 Squid 2.4 Dell PowerEdge 130B 2.5 iMimic DataReactor 5000 2.9 IBM 3500M10 3.0 Stratachache Flyer 100 4.0 Cisco Cache Engine 7300 4.2 Quantex WebXL-4000 4.2 IBM 5600 4.8 IBM 5000 5.1 Cisco Cache Engine 590 6.9 Compaq TaskSmart 900 7.8 Compaq TaskSmart 2500 9.2 Swell Tsunami CPX-1000 n/a

Back to the main review

Cache panache
A well-placed caching device can boost Web site performance and cut WAN costs.

How we did it

Law firm defends move from software to hardware caching
Better performance drives Perkins Coie from software proxy servers to CacheFlow's Internet Caching Appliances.

Interactive buyer's guide
Find the caching device that's right for you or compare two or more devices in a variety of categories, with our detailed database of product specs.