数据库基准测试4：HammerDB测试脚本运用（for Oracle）
0.前面要说的话
之前章节简单介绍了一下hammerdbcli的使用方法，但通过命令行的方式不适合大批量数据基准测试（灌了3000仓库的tpcc数据在虚拟 机上耗时超过了1天多），实际hammerdb已经内置了多种数据库类型的基准测试脚本，以下以oracle为例简单说明如何使用这些脚本。
1.基准测试脚本
HammerDB自带的DB2 TPC-C测试脚本的路径是/opt/HammerDB-5.0/scripts/tcl/oracle/tprocc,主要内容有以下，为方便修改，复制一个/tprocc目录为/tprocc_aix，方便后续修改。
[root@xxx ~]# cd /opt/HammerDB-5.0/scripts/tcl/oracle/tprocc
[root@xxx tprocc]# ll
total 28
-rw-r--r-- 1 root root 690 Jul 30 15:20 ora_tprocc_buildschema.tcl
-rw-r--r-- 1 root root 361 Jul 30 15:20 ora_tprocc_checkschema.tcl
-rw-r--r-- 1 root root 360 Jul 30 15:20 ora_tprocc_deleteschema.tcl
-rw-r--r-- 1 root root 1025 Jul 30 15:20 ora_tprocc.ps1
-rw-r--r-- 1 root root 116 Jul 30 15:20 ora_tprocc_result.tcl
-rw-r--r-- 1 root root 673 Jul 30 15:20 ora_tprocc_run.tcl
-rw-r--r-- 1 root root 965 Jul 30 15:20 ora_tprocc.sh
首先看ora_tprocc.sh，这个脚本实际是将buildschema、checkschema、run、deleteschema、result几个过程放在一起执行，对于大测试而言，此脚本最好拆成多个执行。
2.测试脚本的修改
2.1.buildschema的修改
修改ora_tprocc_buildschema.tcl.tcl
参考以下中文注释进行修改，下同
#!/bin/tclsh
# maintainer: Pooja Jain
puts "SETTING CONFIGURATION"
dbset db ora
dbset bm TPC-C
diset connection system_user hammerdb
diset connection system_password my_passwd
diset connection instance 192.168.1.62:1521/testdb
diset tpcc count_ware 3000
diset tpcc num_vu 32
diset tpcc tpcc_user hammerdb
diset tpcc tpcc_pass my_passwd
diset tpcc tpcc_def_tab users
diset tpcc tpcc_def_temp temp
diset tpcc partition true
diset tpcc hash_clusters false
diset tpcc tpcc_ol_tab users
diset tpcc rampup 2
diset tpcc duration 5
puts "SCHEMA BUILD STARTED"
buildschema
puts "SCHEMA BUILD COMPLETED"
2.2.checkschema的修改
修改ora_tprocc_checkschema.tcl
#!/bin/tclsh
# maintainer: Pooja Jain
dbset db ora
dbset bm TPC-C
puts "SETTING CONFIGURATION"
dbset db ora
dbset bm TPC-C
diset connection system_user hammerdb
diset connection system_password my_passwd
diset connection instance 192.168.1.62:1521/testdb
diset tpcc tpcc_user hammerdb
diset tpcc tpcc_pass my_passwd
puts "CHECK SCHEMA STARTED"
checkschema
puts "CHECK SCHEMA COMPLETED"
2.3.run的修改
修改ora_tprocc_run.tcl
#!/bin/tclsh
# maintainer: Pooja Jain
set tmpdir $::env(TMP)
puts "SETTING CONFIGURATION"
dbset db ora
dbset bm TPC-C
diset connection system_user hammerdb
diset connection system_password my_passwd
diset connection instance 192.168.1.62:1521/testdb
diset tpcc tpcc_user hammerdb
diset tpcc tpcc_pass my_passwd
diset tpcc ora_driver timed
diset tpcc total_iterations 10000000
diset tpcc rampup 2
diset tpcc duration 5
diset tpcc ora_timeprofile true
diset tpcc allwarehouse true
diset tpcc checkpoint false
loadscript
puts "TEST STARTED"
vuset vu 32
vucreate
tcstart
tcstatus
set jobid [ vurun ]
vudestroy
tcstop
puts "TEST COMPLETE"
set of [ open $tmpdir/ora_tprocc w ]
puts $of $jobid
close $of
2.4.delete及result的修改
- ora_tprocc_deleteschema.tcl脚本不建议执行，因为有多次执行的需要；
- ora_tprocc_result.tcl脚本不需要修改
- ora_tprocc.sh也不建议执行，手工单步执行更好一点
2.5.拆分执行脚本的修改
事实最好将ora_tprocc.sh脚本折成三个，独立进行：
- 01.build.sh
export TMP=`pwd`/TMP
mkdir -p $TMP
home_dir=/opt/HammerDB-5.0 # 增加home_dir参数，方便在'tprocc_1'当前目录下执行脚本
cur_dir=tprocc_aix # 将默认的'tprocc'目录修改为'tprocc_1'目录，以下对应做修改
echo "BUILD HAMMERDB SCHEMA"
echo "+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-"
${home_dir}/hammerdbcli auto ./scripts/tcl/oracle/${cur_dir}/ora_tprocc_buildschema.tcl
echo "+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-"
echo "CHECK HAMMERDB SCHEMA"
echo "+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-"
${home_dir}/hammerdbcli auto ./scripts/tcl/oracle/${cur_dir}/ora_tprocc_checkschema.tcl
echo "+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-"
- 02.check.sh
export TMP=`pwd`/TMP
mkdir -p $TMP
home_dir=/opt/HammerDB-5.0 # 增加home_dir参数，方便在'tprocc_1'当前目录下执行脚本
cur_dir=tprocc_aix # 将默认的'tprocc'目录修改为'tprocc_1'目录，以下对应做修改
echo "CHECK HAMMERDB SCHEMA"
echo "+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-"
${home_dir}/hammerdbcli auto ./scripts/tcl/oracle/${cur_dir}/ora_tprocc_checkschema.tcl
echo "+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-"
- 03.run.sh
export TMP=`pwd`/TMP
mkdir -p $TMP
home_dir=/opt/HammerDB-5.0 # 增加home_dir参数，方便在'tprocc_1'当前目录下执行脚本
cur_dir=tprocc_aix # 将默认的'tprocc'目录修改为'tprocc_1'目录，以下对应做修改
echo "RUN HAMMERDB TEST"
echo "+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-"
${home_dir}/hammerdbcli auto ./scripts/tcl/oracle/${cur_dir}/ora_tprocc_run.tcl
echo "+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-"
echo "HAMMERDB RESULT"
${home_dir}/hammerdbcli auto ./scripts/tcl/oracle/${cur_dir}/ora_tprocc_result.tcl
3.执行基准测试
对HammerDB的脚本处理完成之后，按以下步骤执行基准测试：
- 执行01.build.sh，加载测试数据，这个过程十分漫长，建议使用nohup后台执行；
- 执行02.check.sh，检查数据是否有问题，有概率在上一步骤加载测试数据量不足，导致check失败；
- 执行03.rub.sh，按
print dict中tpc-c的基准测试，会启动虚拟用户执行共1000万个事务操作，在操作完成输出以下测试结果：
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-
HAMMERDB RESULT
HammerDB CLI v5.0
Copyright © HammerDB Ltd hosted by tpc.org 2019-2025
Type "help" for a list of commands
Usage: hammerdbcli [ tcl|python [ auto [ script_to_autoload.[ tcl|py ] ] ] ]
Initialized Jobs on-disk database /opt/HammerDB-5.0/scripts/tcl/oracle/tprocc_aix/TMP/hammer.DB using existing tables (13,287,424 bytes)
TRANSACTION RESPONSE TIMES
{
"NEWORD": {
"elapsed_ms": "421900.5",
"calls": "259294",
"min_ms": "3.573",
"avg_ms": "33.823",
"max_ms": "3022.898",
"total_ms": "8770332.075",
"p99_ms": "373.239",
"p95_ms": "117.07",
"p50_ms": "14.628",
"sd": "108677.826",
"ratio_pct": "64.962"
},
"PAYMENT": {
"elapsed_ms": "421900.5",
"calls": "259706",
"min_ms": "1.421",
"avg_ms": "11.035",
"max_ms": "2981.316",
"total_ms": "2865928.233",
"p99_ms": "153.627",
"p95_ms": "16.947",
"p50_ms": "4.347",
"sd": "62516.826",
"ratio_pct": "21.228"
},
"DELIVERY": {
"elapsed_ms": "421900.5",
"calls": "25778",
"min_ms": "7.055",
"avg_ms": "38.295",
"max_ms": "3041.013",
"total_ms": "987188.343",
"p99_ms": "376.191",
"p95_ms": "131.793",
"p50_ms": "17.336",
"sd": "122269.254",
"ratio_pct": "7.312"
},
"SLEV": {
"elapsed_ms": "421900.5",
"calls": "26132",
"min_ms": "2.391",
"avg_ms": "24.579",
"max_ms": "2950.456",
"total_ms": "642316.905",
"p99_ms": "268.744",
"p95_ms": "69.19",
"p50_ms": "8.601",
"sd": "60001.272",
"ratio_pct": "4.758"
},
"OSTAT": {
"elapsed_ms": "421900.5",
"calls": "25842",
"min_ms": "1.183",
"avg_ms": "8.329",
"max_ms": "508.925",
"total_ms": "215238.474",
"p99_ms": "113.531",
"p95_ms": "23.129",
"p50_ms": "3.876",
"sd": "21466.349",
"ratio_pct": "1.594"
}
}
TRANSACTION COUNT
{"Oracle tpm": {
"2025-11-04 11:39:53": "0",
"2025-11-04 11:40:03": "27468",
"2025-11-04 11:40:13": "23274",
"2025-11-04 11:40:23": "20046",
"2025-11-04 11:40:33": "18816",
"2025-11-04 11:40:43": "20790",
"2025-11-04 11:40:53": "25176",
"2025-11-04 11:41:03": "29466",
"2025-11-04 11:41:13": "33660",
"2025-11-04 11:41:23": "32826",
"2025-11-04 11:41:33": "38628",
"2025-11-04 11:41:43": "38796",
"2025-11-04 11:41:53": "39192",
"2025-11-04 11:42:03": "40164",
"2025-11-04 11:42:13": "46362",
"2025-11-04 11:42:23": "34242",
"2025-11-04 11:42:33": "39486",
"2025-11-04 11:42:43": "47214",
"2025-11-04 11:42:53": "40920",
"2025-11-04 11:43:03": "32532",
"2025-11-04 11:43:14": "43950",
"2025-11-04 11:43:24": "43980",
"2025-11-04 11:43:34": "37062",
"2025-11-04 11:43:44": "44136",
"2025-11-04 11:43:54": "46986",
"2025-11-04 11:44:04": "48348",
"2025-11-04 11:44:14": "42900",
"2025-11-04 11:44:24": "36558",
"2025-11-04 11:44:34": "34560",
"2025-11-04 11:44:44": "36732",
"2025-11-04 11:44:54": "41094",
"2025-11-04 11:45:04": "30978",
"2025-11-04 11:45:14": "39696",
"2025-11-04 11:45:24": "36684",
"2025-11-04 11:45:34": "43596",
"2025-11-04 11:45:44": "37032",
"2025-11-04 11:45:54": "47124",
"2025-11-04 11:46:04": "39810",
"2025-11-04 11:46:14": "40494",
"2025-11-04 11:46:24": "43566",
"2025-11-04 11:46:34": "51930",
"2025-11-04 11:46:44": "53448",
"2025-11-04 11:46:54": "60054"
}}
HAMMERDB RESULT
[
"6909758995B703E273837333",
"2025-11-04 11:39:53",
"16 Active Virtual Users configured",
"TEST RESULT : System achieved 20006 NOPM from 41926 Oracle TPM"
]
4.一些问题
- oracle归档日志满
当压入3000个仓库的数据时，总的数据量约200GB，在buildschema时，如果开启了归档日志，需要确认归档日志磁盘的空间可用，因为在build的过程即使日志满并不会报错（输出的日志一直卡住不动），事后要杀掉HammerDB产生的session也很麻烦。
- 创建索引及存储过程失败
在build阶段的最后，会创建对应的索引及存储过程，但有概率失败，这是可以手动执行相关的创建作业，如以下：
-- 逐个执行索引创建语句（按顺序执行，确保无报错）
CREATE UNIQUE INDEX CUSTOMER_I1 ON CUSTOMER ( C_W_ID, C_D_ID, C_ID) INITRANS 4 PCTFREE 10;
CREATE UNIQUE INDEX CUSTOMER_I2 ON CUSTOMER ( C_LAST, C_W_ID, C_D_ID, C_FIRST, C_ID) INITRANS 4 PCTFREE 10;
CREATE UNIQUE INDEX DISTRICT_I1 ON DISTRICT ( D_W_ID, D_ID) INITRANS 4 PCTFREE 10;
CREATE UNIQUE INDEX ITEM_I1 ON ITEM (I_ID) INITRANS 4 PCTFREE 10;
CREATE UNIQUE INDEX ORDERS_I1 ON ORDERS (O_W_ID, O_D_ID, O_ID) INITRANS 4 PCTFREE 10 LOCAL;
CREATE UNIQUE INDEX ORDERS_I2 ON ORDERS (O_W_ID, O_D_ID, O_C_ID, O_ID) INITRANS 4 PCTFREE 10 LOCAL;
CREATE UNIQUE INDEX STOCK_I1 ON STOCK (S_I_ID, S_W_ID) INITRANS 4 PCTFREE 10;
CREATE UNIQUE INDEX WAREHOUSE_I1 ON WAREHOUSE (W_ID) INITRANS 4 PCTFREE 10;
-- 直接执行存储过程创建语句（复制报错中的完整代码）
CREATE OR REPLACE PROCEDURE DMPMON.NEWORD (
no_w_id BINARY_INTEGER,
no_max_w_id BINARY_INTEGER,
no_d_id BINARY_INTEGER,
no_c_id BINARY_INTEGER,
no_o_ol_cnt BINARY_INTEGER,
no_c_discount OUT NUMBER,
no_c_last OUT VARCHAR2,
no_c_credit OUT VARCHAR2,
no_d_tax OUT NUMBER,
no_w_tax OUT NUMBER,
no_d_next_o_id OUT BINARY_INTEGER,
timestamp IN DATE
) IS
order_amount NUMBER;
no_o_all_local BINARY_INTEGER;
loop_counter BINARY_INTEGER;
not_serializable EXCEPTION;
PRAGMA EXCEPTION_INIT(not_serializable,-8177);
deadlock EXCEPTION;
PRAGMA EXCEPTION_INIT(deadlock,-60);
snapshot_too_old EXCEPTION;
PRAGMA EXCEPTION_INIT(snapshot_too_old,-1555);
integrity_viol EXCEPTION;
PRAGMA EXCEPTION_INIT(integrity_viol,-1);
TYPE intarray IS TABLE OF INTEGER index by binary_integer;
TYPE numarray IS TABLE OF NUMBER index by binary_integer;
TYPE distarray IS TABLE OF VARCHAR(24) index by binary_integer;
o_id_array intarray;
w_id_array intarray;
o_quantity_array intarray;
s_quantity_array intarray;
ol_line_number_array intarray;
amount_array numarray;
district_info distarray;
BEGIN
SELECT c_discount, c_last, c_credit, w_tax
INTO no_c_discount, no_c_last, no_c_credit, no_w_tax
FROM customer, warehouse
WHERE warehouse.w_id = no_w_id
AND customer.c_w_id = no_w_id
AND customer.c_d_id = no_d_id
AND customer.c_id = no_c_id;
no_o_all_local := 1;
FOR loop_counter IN 1 .. no_o_ol_cnt LOOP
o_id_array(loop_counter) := round(DBMS_RANDOM.value(low => 1, high => 100000));
IF ( DBMS_RANDOM.value >= 0.01 ) THEN
w_id_array(loop_counter) := no_w_id;
ELSE
no_o_all_local := 0;
w_id_array(loop_counter) := 1 + mod(no_w_id + round(DBMS_RANDOM.value(low => 0, high => no_max_w_id-1)),no_max_w_id);
END IF;
o_quantity_array(loop_counter) := round(DBMS_RANDOM.value(low => 1, high => 10));
ol_line_number_array(loop_counter) := loop_counter;
END LOOP;
UPDATE district
SET d_next_o_id = d_next_o_id + 1
WHERE d_id = no_d_id AND d_w_id = no_w_id
RETURNING d_next_o_id - 1, d_tax INTO no_d_next_o_id, no_d_tax;
INSERT INTO ORDERS (o_id, o_d_id, o_w_id, o_c_id, o_entry_d, o_ol_cnt, o_all_local)
VALUES (no_d_next_o_id, no_d_id, no_w_id, no_c_id, timestamp, no_o_ol_cnt, no_o_all_local);
INSERT INTO NEW_ORDER (no_o_id, no_d_id, no_w_id)
VALUES (no_d_next_o_id, no_d_id, no_w_id);
IF no_d_id = 1 THEN
FORALL i IN 1 .. no_o_ol_cnt
UPDATE stock_item
SET s_quantity = (CASE WHEN s_quantity < ( o_quantity_array(i) + 10 ) THEN s_quantity + 91 ELSE s_quantity END) - o_quantity_array(i)
WHERE i_id = o_id_array(i) AND s_w_id = w_id_array(i)
RETURNING s_dist_01, s_quantity, i_price * o_quantity_array(i)
BULK COLLECT INTO district_info, s_quantity_array, amount_array;
ELSIF no_d_id = 2 THEN
FORALL i IN 1 .. no_o_ol_cnt
UPDATE stock_item
SET s_quantity = (CASE WHEN s_quantity < ( o_quantity_array(i) + 10 ) THEN s_quantity + 91 ELSE s_quantity END) - o_quantity_array(i)
WHERE i_id = o_id_array(i) AND s_w_id = w_id_array(i)
RETURNING s_dist_02, s_quantity, i_price * o_quantity_array(i)
BULK COLLECT INTO district_info, s_quantity_array,amount_array;
ELSIF no_d_id = 3 THEN
FORALL i IN 1 .. no_o_ol_cnt
UPDATE stock_item
SET s_quantity = (CASE WHEN s_quantity < ( o_quantity_array(i) + 10 ) THEN s_quantity + 91 ELSE s_quantity END) - o_quantity_array(i)
WHERE i_id = o_id_array(i) AND s_w_id = w_id_array(i)
RETURNING s_dist_03, s_quantity, i_price * o_quantity_array(i)
BULK COLLECT INTO district_info, s_quantity_array,amount_array;
ELSIF no_d_id = 4 THEN
FORALL i IN 1 .. no_o_ol_cnt
UPDATE stock_item
SET s_quantity = (CASE WHEN s_quantity < ( o_quantity_array(i) + 10 ) THEN s_quantity + 91 ELSE s_quantity END) - o_quantity_array(i)
WHERE i_id = o_id_array(i) AND s_w_id = w_id_array(i)
RETURNING s_dist_04, s_quantity, i_price * o_quantity_array(i)
BULK COLLECT INTO district_info, s_quantity_array,amount_array;
ELSIF no_d_id = 5 THEN
FORALL i IN 1 .. no_o_ol_cnt
UPDATE stock_item
SET s_quantity = (CASE WHEN s_quantity < ( o_quantity_array(i) + 10 ) THEN s_quantity + 91 ELSE s_quantity END) - o_quantity_array(i)
WHERE i_id = o_id_array(i) AND s_w_id = w_id_array(i)
RETURNING s_dist_05, s_quantity, i_price * o_quantity_array(i)
BULK COLLECT INTO district_info, s_quantity_array,amount_array;
ELSIF no_d_id = 6 THEN
FORALL i IN 1 .. no_o_ol_cnt
UPDATE stock_item
SET s_quantity = (CASE WHEN s_quantity < ( o_quantity_array(i) + 10 ) THEN s_quantity + 91 ELSE s_quantity END) - o_quantity_array(i)
WHERE i_id = o_id_array(i) AND s_w_id = w_id_array(i)
RETURNING s_dist_06, s_quantity, i_price * o_quantity_array(i)
BULK COLLECT INTO district_info, s_quantity_array,amount_array;
ELSIF no_d_id = 7 THEN
FORALL i IN 1 .. no_o_ol_cnt
UPDATE stock_item
SET s_quantity = (CASE WHEN s_quantity < ( o_quantity_array(i) + 10 ) THEN s_quantity + 91 ELSE s_quantity END) - o_quantity_array(i)
WHERE i_id = o_id_array(i) AND s_w_id = w_id_array(i)
RETURNING s_dist_07, s_quantity, i_price * o_quantity_array(i)
BULK COLLECT INTO district_info, s_quantity_array,amount_array;
ELSIF no_d_id = 8 THEN
FORALL i IN 1 .. no_o_ol_cnt
UPDATE stock_item
SET s_quantity = (CASE WHEN s_quantity < ( o_quantity_array(i) + 10 ) THEN s_quantity + 91 ELSE s_quantity END) - o_quantity_array(i)
WHERE i_id = o_id_array(i) AND s_w_id = w_id_array(i)
RETURNING s_dist_08, s_quantity, i_price * o_quantity_array(i)
BULK COLLECT INTO district_info, s_quantity_array,amount_array;
ELSIF no_d_id = 9 THEN
FORALL i IN 1 .. no_o_ol_cnt
UPDATE stock_item
SET s_quantity = (CASE WHEN s_quantity < ( o_quantity_array(i) + 10 ) THEN s_quantity + 91 ELSE s_quantity END) - o_quantity_array(i)
WHERE i_id = o_id_array(i) AND s_w_id = w_id_array(i)
RETURNING s_dist_09, s_quantity, i_price * o_quantity_array(i)
BULK COLLECT INTO district_info, s_quantity_array,amount_array;
ELSIF no_d_id = 10 THEN
FORALL i IN 1 .. no_o_ol_cnt
UPDATE stock_item
SET s_quantity = (CASE WHEN s_quantity < ( o_quantity_array(i) + 10 ) THEN s_quantity + 91 ELSE s_quantity END) - o_quantity_array(i)
WHERE i_id = o_id_array(i) AND s_w_id = w_id_array(i)
RETURNING s_dist_10, s_quantity, i_price * o_quantity_array(i)
BULK COLLECT INTO district_info, s_quantity_array,amount_array;
END IF;
order_amount := 0;
FOR loop_counter IN 1 .. no_o_ol_cnt LOOP
order_amount := order_amount + ( amount_array(loop_counter) );
END LOOP;
order_amount := order_amount * ( 1 + no_w_tax + no_d_tax ) * ( 1 - no_c_discount );
FORALL i IN 1 .. no_o_ol_cnt
INSERT INTO order_line (ol_o_id, ol_d_id, ol_w_id, ol_number, ol_i_id, ol_supply_w_id, ol_quantity, ol_amount, ol_dist_info)
VALUES (no_d_next_o_id, no_d_id, no_w_id, ol_line_number_array(i), o_id_array(i), w_id_array(i), o_quantity_array(i), amount_array(i), district_info(i));
IF DBMS_RANDOM.value < 0.01 THEN
dbms_output.put_line('Rolling back');
ROLLBACK;
ELSE
COMMIT;
END IF;
EXCEPTION
WHEN not_serializable OR deadlock OR snapshot_too_old OR integrity_viol THEN
ROLLBACK;
END;
- 表空间的问题
有一个很奇怪的问题，即使用设置了新的表空间diset tpcc tpcc_def_tab tpcc，但HammerDB并不认这个设定，始终将数据灌到USERS表空间下，所以要注意对此表空间的datafile的容量做好提前确认工作。
- NOPM值缺失
如果执行压测时，发现只有tpm的值但没有NOPM的值，此时则需要检查所建的存储过程是否有缺失或无效。
5.总结
通过修改HammerDB自带的脚本还是很方便进行基准测试的，修改参数也很方便，至于基准测试结果的解读，放到下次说明。
浙公网安备 33010602011771号