Program guide > (deprecated) Partitioning facility > Partitioning facility programming > Data partitioning patterns


Deprecated feature: The partitioning facility (WPF) feature is deprecated. You can configure partitioning with WebSphere eXtreme Scale.


Fixed partition set

Use any hashing scheme that maps the keys to a fixed set of partitions. For example, you could use a normal hashing algorithm that takes the key and converts it to an integer. This type of hash function typically returns a 32 bit or 64 bit integer, which in general produces too large a space of partitions. Instead, apply a modulo to the generated hash to limit the total number of partitions. For example, you could hash the key and then use the 128 as the modulo (number of partitions). The application should create 128 partitions, the first one named 0 and the last one named 127, so (128 in total). This is a good compromise as this pattern can potentially scale to 128 cluster members before we have more servers than partitions.

Another approach is to use ranges. This is still a form of hashing. The partitions could be named A-F, G-N, O-R, S-Z. The routing POJO would then examine the key from the request and return a string from one of these ranges.


Parent topic:

Data partitioning patterns


Related concepts

Data partitioning patterns