Files
@ 3aa4d95db0a9
Branch filter:
Location: DA/protocols/vldb-protocols2.R - annotation
3aa4d95db0a9
2.0 KiB
text/S-plus
new script
3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 3aa4d95db0a9 |
library(dplyr)
library(ggplot2)
library(ggthemes)
theme <- theme_few(base_size = 24) +
theme(axis.title.y=element_text(vjust=0.9),
axis.title.x=element_text(vjust=-0.1),
axis.ticks.x=element_blank(),
text=element_text(family="serif"))
read.table("12.csv", header=T, sep=",", stringsAsFactors=F, na.strings="-1") -> dd3
sysnames <- c("mariadb-default"="MySQL", "mariadb-compress"="MySQL+C", "oracle-default"="DBMS X", "postgres-default" = "PostgreSQL", "db2-default" = "DB2", "monetdb-default"="MonetDB", "hbase-default" = "HBase", "hive-default" = "Hive", "mongodb-default" = "MongoDB")
dd3 %>% filter(network=="unlimited", protocol == "native", timeout != 1, system %in% c("hive-default", "hbase-default", "mariadb-default", "mariadb-compress", "db2-default", "oracle-default", "postgres-default", "monetdb-default", "mongodb-default"), tuple %in% c(1, 1000000)) %>% mutate(system=sysnames[system]) %>% select(system, tuple, time, bytes) %>% group_by(system, tuple) %>% summarise_each(funs(mean)) -> dd4
dd5 <- dd4 %>% filter(tuple==1000000) %>% rename(time_1m=time, bytes_1m=bytes) %>% left_join(dd4 %>% filter(tuple == 1) %>% rename(time_1=time, bytes_1=bytes), by="system") %>% mutate(tuple_time_ms=((time_1m-time_1)/1000000)*1000, tuple_bytes=(bytes_1m-bytes_1)/1000000)
pdf("time-per-tuple.pdf", width=10, height=5)
ggplot(dd5, aes( y=tuple_time_ms, x=reorder(system, -tuple_time_ms), label=round(tuple_time_ms, 4))) + geom_bar(stat="identity", position="dodge", width=.5) + theme + xlab("") + ylab("Time per tuple (ms)") + coord_flip() + guides(fill=FALSE) + geom_text(size=7, hjust=-.2, family="serif") + scale_y_continuous(limits=c(0, .03))
dev.off()
pdf("bytes-per-tuple.pdf", width=10, height=5)
ggplot(dd5, aes( y=tuple_bytes, x=reorder(system, -tuple_bytes), label=round(tuple_bytes))) + geom_bar(stat="identity", position="dodge", width=.5) + theme + xlab("") + ylab("Bytes per tuple") + coord_flip() + guides(fill=FALSE) + geom_text(size=7, hjust=-.2, family="serif") + scale_y_continuous(limits=c(0, 410))
dev.off()
|