| Meno: | Juraj
|
|---|
| Priezvisko: | Macháč
|
|---|
| Názov: | Optimization of execution plans in the FlumeJava model
|
|---|
| Vedúci: | prof. RNDr. Rastislav Královič, PhD.
|
|---|
| Rok: | 2016
|
|---|
| Blok: | INF
|
|---|
| Kµúčové slová: | FlumeJava, MapReduce, Min-cut, Network usage
|
|---|
| Abstrakt: | MapReduce is a paradigm used for processing large sets of data in parallel. FlumeJava,
developed by Google, is a library providing an abstraction of this MapReduce in form of Collection and operations on these Collections. The operations defined by programmer form
an execution plan, and FlumeJava has to transform this plan into pipelines of MapReduces.
There are many options on how to create these pipelines from given execution plan, and
some are better in terms of network usage. This thesis studies the creation of MapReduce
pipelines with optimal network usage when given an execution plan.
|
|---|