Geography Reference
In-Depth Information
For example, Microsoft Bing Maps uses a tiling scheme where the first level allows
representing the whole world in four tiles (2x2) of 256x256 pixels. The next level represents
the whole world in 16 tiles (4x4) of 256x256 pixels and so on in powers of 4. A comprehensive
study on tiling schemes can be found in [2].
2.1. Simplified model
Given the exponential nature of the scale pyramid, the resource consumption to store map tiles
results often prohibitive for many providers when the cartography covers a wide geographic
area for multiple scales. Consider for example that Google's BigTable, which contains the
high-resolution satellite imagery of the world's surface as shown in Google Maps and Google
Earth, contained approximately 70 terabytes of data in 2006 [6].
Besides the storage of map tiles, many caching systems also maintain metadata associated to
each individual tile, such as the time when it was introduced into the cache, the last access to
that object, or the number of times it has been requested. This information can then be used to
improve the cache management; for example, when the cache is out of space, the LRU ( Least
Recently Used ) replacement policy uses the last access time to discard the least recently used
items first.
However, the space required to store the metadata associated to a given tile may only differ
by two or three orders of magnitude to the one necessary to store the actual map image object.
Therefore, it is not usually feasible to work with the statistics of individual tiles. To alleviate
this problem, a simplified model has been proposed by different researchers. This model
groups the statistics of adjacent tiles into a single object [7]. A grid is defined so all objects
inside the same grid section are combined into a single one. The pyramidal structure of scales
is therefore transformed in some way in a prism-like structure with the same number of items
in all the scales.
3. Web Map Server workload
In order to deal with this complexity some cache management algorithms have been
created. However, the efficiency of the designed algorithms usually depends on the service's
workload. Because of this, prior to diving into the details of the cache management policies, a
workload characterization of the WMS services need to be shown. Lets take some real-life
examples for such characterization: trace files from two different tiled web map services,
Cartociudad 1
and IDEE-Base 2 , provided by the National Geographic Institute (IGN) 3
of
Spain, are presented in this chapter.
Cartociudad is the official cartographic database of the Spanish cities and villages with their
streets and roads networks topologically structured, while IDEE-Base allows viewing the
Numeric Cartographic Base 1:25,000 and 1:200,000 of the IGN.
Available trace files were filtered to contain only valid web map requests according to the
WMS-C recommendation. Traces from Cartociudad comprise a total of 2.369.555 requests
1 http://www.cartociudad.es
2 http://www.idee.es
3 http://www.ign.es/ign/main/index.do?locale=en
Search WWH ::




Custom Search