Home
100 concurrent users caching the product display page only
In this test, we enabled the DynaCache full page caching of the product display page. Since the page contains a header bar, footer bar, and a mini shopping cart, all realized as JSP fragments, we also had to enable fragment caching in cachespec.xml.
To get the benefit from product display page caching, we fully populated the cache prior to the formal test by running a browsing-only scenario and using the cache monitor to get information about the state of the cache.
Table | -5 and Table | -6, and Figure | -4 summarize the results of the test.
Table 9-5
CPU usage Minimum JVM heap usage Hard disk usage increase Commerce server 1 85.98% 350MB Commerce server 2 86.13% 350MB Database server 36.12% 40MB Web server 4.72% Resource usage in the test with 100 users and caching of product display page
Comparing Table | -5 with Table | -3, we can observe that the CPU usage on the database server has dropped to 36.12% from 100%, while the CPU usage on both WebSphere Commerce servers has increased. The database server is thus no longer the bottleneck.
Table 9-6
Web action Average response time (second) Count Total time (second) Overall 1.625 210,572 342179.5 Select shipping address 0.595 2,623 1560.685 Add shipping address 0.336 1,890 635.04 Display category 2.946 95,236 280565.256 Log off 0.34 4,924 1674.16 Log on 0.589 2,654 1563.206 Go to log on page 0.465 2,654 1234.11 Add item to cart 1.694 6,514 11034.716 View billing address 1.558 5,380 8382.04 Process order 2.022 595 1203.09 View order summary 3.404 1,179 4013.316 Add payment info 1.795 595 1068.025 Display product 0.135 31,714 4281.39 View shipping method 0.807 5,376 4338.432 Display top category 0.484 31,769 15376.196 Visit store front 0.182 12,377 2252.614 Register user 0.758 2,546 1929.868 Go to register page 0.43 2,546 1094.78 Response times in the test with 100 users and caching of product display page
As we can see from Table | -6, the response time for the product display page is now only 0.135 seconds, down considerably from 35.338. Because of this dramatic drop, the total number of pages visited in 3 hours has increased from 35,459 to 210,572, marking a six-fold increase in throughput. This is typical of many implementations of DynaCache.
Figure 9-4 Total time for URL actions in the test with 100 users and caching of product display page