-
Notifications
You must be signed in to change notification settings - Fork 3
/
DBx1000_README
90 lines (70 loc) · 3.59 KB
/
DBx1000_README
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
DBMS BENCHMARK
------------------------------
== General Features ==
dbms is a OLTP database benchmark with the following features.
1. Seven different concurrency control algorithms are supported.
DL_DETECT[1] : deadlock detection
NO_WAIT[1] : no wait two phase locking
WAIT_DIE[1] : wait and die two phase locking
TIMESTAMP[1] : basic T/O
MVCC[1] : multi-version T/O
HSTORE[3] : H-STORE
OCC[2] : optimistic concurrency control
[1] Phlip Bernstein, Nathan Goodman, "Concurrency Control in Distributed Database Systems", Computing Surveys, June 1981
[2] H.T. Kung, John Robinson, "On Optimistic Methods for Concurrency Control", Transactions on Database Systems, June 1981
[3] R. Kallman et al, "H-Store: a High-Performance, Distributed Main Memory Transaction Processing System,", VLDB 2008
2. Two benchmarks are supported.
2.1 YCSB[4]
2.2 TPCC[5]
Only Payment and New Order transactions are modeled.
[4] B. Cooper et al, "Benchmarking Cloud Serving Systems with YCSB", SoCC 201
[5] http://www.tpc.org/tpcc/
== Config File ==
dbms benchmark has the following parameters in the config file. Parameters with a * sign should not be changed.
CORE_CNT : number of cores modeled in the system.
PART_CNT : number of logical partitions in the system
THREAD_CNT : number of threads running at the same time
PAGE_SIZE : memory page size
CL_SIZE : cache line size
WARMUP : number of transactions to run for warmup
WORKLOAD : workload supported (TPCC or YCSB)
THREAD_ALLOC : per thread allocator.
* MEM_PAD : enable memory padding to avoid false sharing.
MEM_ALLIGN : allocated blocks are alligned to MEM_ALLIGN bytes
PRT_LAT_DISTR : print out latency distribution of transactions
CC_ALG : concurrency control algorithm
* ROLL_BACK : roll back the modifications if a transaction aborts.
ENABLE_LATCH : enable latching in btree index
* CENTRAL_INDEX : centralized index structure
* CENTRAL_MANAGER : centralized lock/timestamp manager
INDEX_STRCT : data structure for index.
BTREE_ORDER : fanout of each B-tree node
DL_TIMEOUT_LOOP : the max waiting time in DL_DETECT. after timeout, deadlock will be detected.
TS_TWR : enable Thomas Write Rule (TWR) in TIMESTAMP
HIS_RECYCLE_LEN : in MVCC, history will be recycled if they are too long.
MAX_WRITE_SET : the max size of a write set in OCC.
MAX_ROW_PER_TXN : max number of rows touched per transaction.
QUERY_INTVL : the rate at which database queries come
MAX_TXN_PER_PART : maximum transactions to run per partition.
// for YCSB Benchmark
SYNTH_TABLE_SIZE : table size
ZIPF_THETA : theta in zipfian distribution (rows accessed follow zipfian distribution)
READ_PERC :
WRITE_PERC :
SCAN_PERC : percentage of read/write/scan queries. they should add up to 1.
SCAN_LEN : number of rows touched per scan query.
PART_PER_TXN : number of logical partitions to touch per transaction
PERC_MULTI_PART : percentage of multi-partition transactions
REQ_PER_QUERY : number of queries per transaction
FIRST_PART_LOCAL : with this being true, the first touched partition is always the local partition.
// for TPCC Benchmark
NUM_HW : number of warehouses being modeled.
PERC_PAYMENT : percentage of payment transactions.
DIST_PER_WARE : number of districts in one warehouse
MAXITEMS : number of items modeled.
CUST_PER_DIST : number of customers per district
ORD_PER_DIST : number of orders per district
FIRSTNAME_LEN : length of first name
MIDDLE_LEN : length of middle name
LASTNAME_LEN : length of last name
// !! centralized CC management should be ignored.