Evaluating partitioning and bucketing strategies for Hive-based Big Data Warehousing systems

Journal of Big Data

Table 9 SSB execution times (in seconds): partitioning by “Od_Year” and bucketing by “Orderkey” (star schema with partitions and buckets (SS-PB)).

	SF = 30		SF = 100		SF = 300		SF = 30		SF = 100		SF = 300
	HIVE						PRESTO
	SS	SS-PB	SS	SS-PB	SS	SS-PB	SS	SS-PB	SS	SS-PB	SS	SS-PB
Q1.1	25	16	31	21	44	26	5	2	13	4	36	7
Q1.2	24	23	29	32	42	42	5	6	13	13	34	44
Q1.3	24	18	29	21	43	25	4	2	13	4	35	8
Q2.1	32	33	47	60	531	682	8	11	19	26	59	98
Q2.2	31	32	46	53	531	677	7	10	18	23	51	76
Q2.3	30	30	44	52	531	670	7	9	17	22	49	74
Q3.1	35	31	59	56	651	667	8	10	29	29	81	100
Q3.2	30	28	45	50	677	634	6	7	17	19	51	63
Q3.3	33	33	219	78	665	648	5	6	15	16	43	53
Q3.4	34	31	222	228	675	674	6	7	15	19	43	59
Q4.1	38	39	86	102	226	253	13	17	43	50	119	164
Q4.2	49	35	70	63	141	91	9	7	26	18	69	48
Q4.3	34	28	54	50	116	77	8	6	23	14	63	38
Total	420	378	982	865	4874	5166	92	100	262	256	733	835
Diff		− 10%		− 12%		6%		8%		− 2%		14%

Italic values indicate the fastest processing time by query, workload, tool and data model