diff --git a/vldb-protocols2.R b/vldb-protocols2.R index b2c86a746ccb0b7bd6f044b795b843c6511ddb74..00dc74b780ee29b7b637dc630f0e666824f0a056 100644 --- a/vldb-protocols2.R +++ b/vldb-protocols2.R @@ -285,3 +285,32 @@ dd5 %>% select(compression, stringmethod, adj_time, adj_bytes) -> dd6 print(xtable::xtable(dd6 %>% filter(compression=="snappy")%>% full_join(dd6 %>% filter(compression =="none"), by="stringmethod") %>% select(stringmethod, time_uncomp=adj_time.y, bytes_uncom=adj_bytes.y, time_snappy=adj_time.x, bytes_snappy=adj_bytes.x) %>% mutate(ratio=round(bytes_uncom/bytes_snappy, 2), time_uncomp=round(time_uncomp, 2), time_snappy=round(time_snappy, 2), bytes_uncom=round(bytes_uncom/1024/1024, 1)) %>% select(stringmethod, time_uncomp, time_snappy, bytes_uncom, ratio)), include.rownames=FALSE) + + + +sysnames <- c("mariadb-default"="MySQL", "mariadb-compress"="MySQL+C", "postgres-default" = "PostgreSQL", "db2-default" = "DB2", "monetdb-prot10"="MonetDB++", "monetdb-prot10-snappy"="MonetDB++Sy", "netcat-csv"= "netcat") + + +read.table("51.csv", header=T, sep=",", stringsAsFactors=F, na.strings="-1") -> dd3 + +ntuples <- 10000000 + +dd3 %>% filter(system %in% c(names(sysnames)), timeout != 1, tuple %in% c(1,ntuples)) %>% mutate(system=sysnames[system]) %>% select(system, network, tuple, time, bytes) %>% group_by(system, network, tuple) %>% summarise_each(funs(mean,sd,se=sd(.)/sqrt(n()))) -> dd4 + + + dd5 <- as.data.frame(dd4 %>% filter(tuple==ntuples) %>% rename(time_1m=time_mean, bytes_1m=bytes_mean) %>% left_join(dd4 %>% filter(tuple == 1) %>% rename(time_1=time_mean, bytes_1=bytes_mean), by=c("system", "network"))) %>% mutate(tuple_time_ms=((time_1m-time_1)/ntuples)*1000, tuple_bytes=(bytes_1m-bytes_1)/ntuples, adj_time= time_1m-time_1, adj_bytes=bytes_1m-bytes_1) %>% select(system, network, adj_time, adj_bytes) + + + +pdf("10m-final-time-unlimited.pdf", width=10, height=3) + +ggplot(dd5 %>% filter(network=="unlimited", system!="netcat"), aes(y=adj_time, x=reorder(system, -adj_time), label=round(adj_time, 1))) + geom_hline(yintercept=1.634335, linetype="dashed") + geom_bar(stat="identity", position="dodge", width=.5) + theme + xlab("") + ylab("Wall clock time (s)") + coord_flip() + geom_text(size=7, hjust=-.2, family="serif") + scale_y_continuous(limits=c(0, 90)) + +dev.off() + +pdf("10m-final-time-slownet.pdf", width=10, height=3) + +ggplot(dd5 %>% filter(network=="10mbitethhd", system!="netcat"), aes(y=adj_time, x=reorder(system, -adj_time), label=round(adj_time, 1))) + geom_hline(yintercept=1014.811853, linetype="dashed") + geom_bar(stat="identity", position="dodge", width=.5) + theme + xlab("") + ylab("Wall clock time (s)") + coord_flip() + geom_text(size=7, hjust=-.2, family="serif") + scale_y_continuous(limits=c(0, 2000)) + +dev.off() +