From upper table 1 test result, for single request, the response time of cluster environment and single node environment is roughly the same, this also is consistent with expected result, although the forwarding of load equalizer 1 has been passed through in the request that also illustrates simultaneously, obvious decline does not appear in overall performance, in a word, for single request, the handling property difference of cluster environment and single node environment is also not obvious.

The contrast experiment 2

The cluster scale of the cluster environment response time to concurrent request is compared, and the cluster environment of using is with contrast experiment 1.The test cluster scale is 1,2,3,4 respectively, concurrent quantity is that the concurrent request processing ability of 100 o'clock is recorded respectively the average response time of overall response time, each request and the number of makeing mistakes, here cluster scale represents the number of working node 2 in cluster environment, for example, only has 1 working node 2 in cluster scale 1 expression cluster environment, the cluster environment of this moment just is equivalent to the single node environment, in cluster scale 4 expression cluster environment, 4 parallel working nodes 2 is arranged.Identical in the request that sends to cluster environment and contrast experiment 1, test result is as shown in the table:

Table 2 overall response time

Table 3 average response time

Table 4 number of makeing mistakes

From above result, we can find out, are 100 o'clock at concurrency, and along with the increase of cluster scale, response total time and average response time are shorter and shorter, and the number of makeing mistakes is also fewer and feweri.Although, at cluster scale be 1 o'clock response total time than cluster scale be 2 o'clock little, make mistakes owing to having occurred 63 in 100 requests, average response time has but been grown a lot.From above test result, we can find out, along with the increase of test environment cluster scale, the disposal ability of concurrent request have also been increased.

The contrast experiment 3

This contrast experiment compares test to single node environment and cluster environment to response time of concurrent request, the cluster environment of using is with contrast experiment 1, measurement scope is that 4 cluster and unit are the response condition of 10,20,30,40,50,60 o'clock at concurrency, record respectively overall response time, the average response time of each request and error situation.Each time the map datum of request be distinguished as 256*256, the longitude and latitude scope is 10, namely maximum poor with minimum longitude and latitude is 10.

Table 5 overall response time

Table 6 average response time

Table 7 number of makeing mistakes

As seen from the above table, along with the rising of concurrency, the difference of the concurrent request processing ability of cluster and unit is more and more obvious.And we can see that the cluster of 4 nodes is than very significantly advantage is arranged on unit on performance.Simultaneously, unit is 60 o'clock at concurrency, occured to make mistakes, and cluster is still very sane, mistake substantially do not occur.

In a word, the above is only preferred embodiment of the present invention, and all equalizations of doing according to the present patent application the scope of the claims change and modify, and all should belong to the covering scope of patent of the present invention.

Claims

1. Web Map Services implementation method based on the cloud framework, it is characterized in that, comprise load equalizer (1), working node (2), described working node (2) comprises WMS service module (21), distributed file system (22), described distributed file system (22) comprises distributed file system interface (221), memory node (222), and concrete steps are as follows:

Step 1: described load equalizer (1) receives the request of Web Map Services;

Step 2: described load equalizer (1) selects a working node (2) as the working node that receives request according to load balancing, and described request is sent to selected working node (2);

Step 3: the WMS service module (21) of selected working node (2) is resolved the URL of described request, according to longitude and latitude scope and the resolution of resolving resulting URL parameter positioning map, calculate the needed Figure List of cutting according to described longitude and latitude scope and resolution;

step 4: described WMS service module (21) is sent to described distributed file system (22) with the described title of cutting figure in Figure List of cutting, described distributed file system (22) cuts accordingly figure according to the name lookup of the described figure of cutting and the figure that cuts that will find returns to described WMS service module (21), described WMS service module (21) is accessed described distributed file system (22) by described distributed file system interface (221), the described figure of cutting stores with document form, described distributed file system (22) is accessed described file by described memory node (222),

Step 5: described WMS service module (21) splices the figure that cuts that finds, adjust the resolution of cutting figure after splicing, according to described longitude and latitude scope, the figure that cuts that adjusts after resolution is carried out cutting and obtains the remote sensing image raster data, and described remote sensing image raster data is sent to described load equalizer (1);

Step 6: described load equalizer (1) returns to the remote sensing image raster data of asking.

2. the Web Map Services implementation method based on the cloud framework according to claim 1, is characterized in that, described load balancing comprises poll, IP-HASH.

3. the Web Map Services implementation method based on the cloud framework according to claim 1, is characterized in that, described working node (2) forms the P2P network.

4. the Web Map Services implementation method based on the cloud framework according to claim 1, it is characterized in that, also comprise the double-layer high-speed cache policy, described double-layer high-speed cache policy uses the double-layer high-speed buffer memory, described double-layer high-speed buffer memory comprises the first buffering area (3) that is positioned at described load equalizer (1) and the second buffering area (4) that is positioned at described working node (2), and described double-layer high-speed cache policy comprises following concrete steps:

In described step 1, described load equalizer (1) detects described remote sensing image raster data in the first buffering area (3), if described the first buffering area (3) has comprised described remote sensing image raster data execution in step six;

In described step 3, described WMS service module (21) judges one by one to described image list, if described the second buffering area has comprised all figure that cuts, execution in step five; If described the second buffering area not yet comprises all figure that cuts, the title of cutting figure that will not yet be loaded on the second buffering area is sent to described distributed file system (22);

In described step 5, described load equalizer (1) puts down in writing the remote sensing image raster data that obtains to described the first buffering area.

5. the Web Map Services implementation method based on the cloud framework according to claim 4, is characterized in that, the file that is cached in described double-layer high-speed buffer memory comprises configuration file and remote sensing image raster data.