tpcc_current (Дополнительные методические материалы), страница 23

This inclu d es d isclosing w hich case w as follow ed for the execu tion of Isolation Test 7.TPC Benchm ark™C - Stand ard Sp ecification, Revision 5.11 - Page 93 of 1308.1.5Scaling and D atabase Population Related Items8.1.5.1The card inality (e.g., the nu m ber of row s) of each table, as it existed at the start of the benchm ark ru n(see Clau se 4.2), m u st be d isclosed .

If the d atabase w as over -scaled and inactive row s of the WAREH OUSE tablew ere d eleted (see Clau se 4.2.2), the card inality of the WAREH OUSE table as initially configu red and the nu m ber ofrow s d eleted m u st be d isclosed .8.1.5.2The d istribu tion of tables and logs across all m ed ia m u st be exp licitly d ep icted for the tested andp riced system s.Disk name: WDC01For each disk WDC01 to WDC0520% of each WAREHOUSE, DISTRICT, CUSTOMER,NEW_ORDER, ORDER, ORDER-LINE, ITEM and STOCKdatabase tables and indexesDisk name: WDC05CPUDisk name: HIST01History 100%Two additional v olumes were usedOperating sy stem root diskOperating sy stem user diskDisk name: LOG01Phy sical log: 100%Logical log: 100%Disk name: LOG02Logical log: 100%(mirrored w/LOG01)TPC Benchm ark™C - Stand ard Sp ecification, Revision 5.11 - Page 94 of 130CPUrootOperatingSy stemrootv olume/usrOperatingSy stem/userf ilesdb03db08db04Sy stempage1 pagev olumehistlogdb01HISTORYf ile100%phy sicallog f ile100%db0210% ofWAREHOUSECUSTOMERDISTRICTNEW_ORDERORDERORDER_LINEtables10%10%10%db09db05db0710%10%db10db0610%10%10%10%itemITEMSTOCKtables100%Comment: Detailed d iagram s for layou t of d atabase files on d isks can w id ely vary, and it is d ifficu lt to p rovid e exactgu id eline su itable for all im p lem entations.

The intent is to p rovid e su fficient d etail to allow ind ep end entreconstru ction of the test d atabase. The tw o figu res below are exam p les of d atabase layou t d escrip tions and are notintend ed to d ep ict or im p ly any op tim al layou t for the TPC -C d atabase.8.1.5.3A statem ent m u st be p rovid ed that d escribes:1. The d ata m od el im p lem ented by the DBMS u sed (e.g., relational, netw ork, hierarchical)2. The d atabase interface (e.g., em bed d ed , call level) and access langu age (e.g., SQL, DL/ 1, COBOL read / w rite)u sed to im p lem ent the TPC-C transactions. If m ore than one interface/ access langu age is u sed to im p lem entTPC-C, each interface/ access langu age m u st be d escribed and a list of w hich interface/ access langu age isu sed w ith w hich transaction typ e m u st be d isclosed .8.1.5.4The m ap p ing of d atabase partitions/ rep lications m u st be exp licitly d escribed .Comment: The intent is to p rovid e su fficient d etail ab ou t p artitioning and rep lication to allow ind ep end entreconstru ction of the test d atabase.An d escrip tion of a d atabase p artitioning schem e is p resented below as an exam p le.

The nom enclatu re of thisexam p le w as ou tlined u sing the CUSTOMER table (in Clau se 8.1.2.1), and has been extend ed to u se the ORDER andORDER_LIN E tables as w ell.C_p art_1C_IDC_D_IDC_W_ID--------- p artition------C_p art_2C_FIRSTO_p art_1O_IDO_D_IDO_W_IDO_C_ID----------- p artition-------OL_p art_1OL_O_IDOL_D_IDOL_W_IDOL_N UMBEROL_I_IDTPC Benchm ark™C - Stand ard Sp ecification, Revision 5.11 - Page 95 of 130C_MIDDLEC_LASTC_STREET_1C_STREET_2C_CITYC_STATEC_ZIPC_PH ON EC_SIN CE----------partition------C_p art_3C_CREDITC_CREDIT_LIMC_DISCOUN TC_BALAN CEC_YTD_PAYMEN TC_PAYMEN T_CN TC_DELIVERY_CN T----------partition------C_p art_4C_DATAO_p art_2O_EN TRY_DO_OL_CN T----------- p artition------O_p art_3O_CARRIER_IDO_ALL_LOCALC_part_1O_part_1OL_part_1----------- p artition------OL_p art_2OL_SUPPLY_W_IDOL_DELIVERY_DOL_QUAN TITYOL_AMOUN T----------- p artition------OL_p art_3OL_DIST_IN FOC_part_3O_part_2C_part_2C_part_4O_part_3OL_part_3OL_part_2One WAREHOUSE Customer/Order/Order_line "cell"8.1.5.5Details of the 60-d ay sp ace com p u tations along w ith p roof that the d atabase is configu red to su stain 8hou rs of grow th for the d ynam ic tables (Ord er, Ord er -Line, and H istory) m u st be d isclosed (see Clau se 4.2.3).8.1.68.1.6.1Performance Metrics and Response Time Related ItemsMeasu red tp m C m u st be rep orted .8.1.6.2N inetieth p ercentile, m axim u m an d average resp onse tim es m u st be rep orted for all transaction typ esas w ell as for the Menu resp onse tim e.8.1.6.3The m inim u m , the average, and the m axim u m keying and think tim es m u st be rep orted for eachtransaction typ e.8.1.6.4typ e.Resp onse Tim e frequ ency d istribu tion cu rves (see Clau se 5.6.1) m u st be rep orted for each transactionTPC Benchm ark™C - Stand ard Sp ecification, Revision 5.11 - Page 96 of 1308.1.6.5The p erform ance cu rve for resp onse tim es versu s throu ghp u t (see Clau se 5.6.2) m u st be rep orted forthe N ew -Ord er transaction .8.1.6.6transaction .Think Tim e frequ ency d istribu tion cu rves (see Clau se 5.6.3) m u st be rep orted for the N ew -Ord er8.1.6.7There is no requ irem ent to rep ort Keying Tim e d istribu tion cu rves.8.1.6.8transaction .A grap h of throu ghp u t versu s elap sed tim e (see Clau se 5.6.4) m u st be rep orted for the N ew -Ord er8.1.6.9The m ethod u sed to d eterm ine that the SUT had reached a stead y state p rior to com m encing them easu rem ent interval (see Clau se 5.5) m u st be d escribed .8.1.6.10A d escrip tion of how the w ork norm ally p erform ed d u ring a su stained test (for exam p lecheckp ointing, w riting red o/ u nd o log record s, etc.), actu ally occu rred d u ring the m easu rem ent interval m u st berep orted .8.1.6.11The start tim e and d u ration in second s of at least the fou r (4) longest checkp oints d u ring theMeasu rem ent Interval m u st be d isclosed (see Clau se 5.5.2.2 (2)).8.1.6.12A statem ent of the d u ration of the m easu rem ent interval for the rep orted Maxim u m Qu alifiedThrou ghp u t (tp m C) m u st be inclu d ed .8.1.6.13The m ethod of regu lation of the transaction m ix (e.g., card d ecks or w eighted rand om d istribu tion)m u st be d escribed .

If w eighted d istribu tion is u sed and the RTE ad ju sts the w eights associated w ith each transactiontyp e, the m axim u m ad ju stm ents to the w eight from the initial valu e m u st be d isclosed .8.1.6.14The p ercentage of the total m ix for each transaction typ e m u st be d isclosed .8.1.6.15d isclosed .The p ercentage of N ew -Ord er transaction s rolled back as a resu lt of invalid item nu m ber m u st be8.1.6.16The average nu m ber of ord er-lines entered p er N ew -Ord er transaction m u st be d isclosed .8.1.6.17The p ercentage of rem ote ord er-lines entered p er N ew -Ord er transaction m u st be d isclosed .8.1.6.18The p ercentage of rem ote Paym ent transaction s m u st be d isclosed .8.1.6.19The p ercentage of cu stom er selections by cu stom er last nam e in the Paym ent and Ord er-Statu stransaction s m u st be d isclosed .8.1.6.20The p ercentage of Delivery transaction s skip p ed d u e to there being few er than necessary ord ers in theN ew -Ord er table m u st be d isclosed .8.1.6.21The nu m ber of checkp oints in the Measu rem ent Interval, the tim e in second s from the start of theMeasu rem ent Interval to the first checkp oint and the Checkp oint Interval m u st be d isclosed .TPC Benchm ark™C - Stand ard Sp ecification, Revision 5.11 - Page 97 of 1308.1.7SUT, D river, and Communication D efinition Related Items8.1.7.1The RTE inp u t p aram eters, cod e fragm ents, fu nctions, etc.

u sed to generate each transaction inp u tfield m u st be d isclosed .Comment: The intent is to d em onstrate the RTE w as configu red to generate transaction inp u t d ata as sp ecified inClau se 2.8.1.7.2The nu m ber of term inal connections lost d u ring the Measu rem ent Interval m u st be d isclosed (seeClau se 6.6.2).8.1.7.3It m u st be d em onstrated that the fu nctionality and p erform ance of the com p onents being em u lated inthe Driver System are equ ivalent to that of the p riced system . Th e resu lts of the test d escribed in Clau se 6.6.3.4 m u stbe d isclosed .8.1.7.4A com p lete fu nctional d iagram of both the benchm ark configu ration and the configu ration of thep rop osed (target) system m u st be d isclosed . A d etailed list of all softw are and har d w are fu nctionality bein gp erform ed on the Driver System , and its interface to the SUT m u st be d isclosed (see Clau se 6.6.3.6).8.1.7.5The netw ork configu ration s of both the tested services and the p rop osed (targ et) services w hich arebeing rep resented and a thorou gh exp lanation of exactly w hich p arts of the p rop osed configu ration are beingrep laced w ith the Driver System m u st be d isclosed (see Clau se 6.6.4).8.1.7.6The band w id th of the netw ork(s) u sed in the tested / p riced configu ration m u st be d isclosed .8.1.7.7If the configu ration requ ires op erator intervention (see Clau se 6.6.6), the m echanism and the frequ encyof this intervention m u st be d isclosed .8.1.8Pricing Related Items8.1.8.1Ru les for rep orting p ricin g inform ation are inclu d ed in the cu rrent revision of the TPC PricingSp ecification, located at w w w .tp c.org.8.1.9Audit Related Items8.1.9.1The au d itor‟ s nam e, ad d ress, p hone nu m ber, and a cop y of the au d itor's attestation letter ind icatingthe au d itor‟ s op inion of com p liance m u st be inclu d ed in the Full Disclosu re Rep ort .8.2Availability of the Full D isclosure ReportThe Full Disclosu re Rep ort m u st be read ily available to the p u blic at a reasonable charge, sim ilar to charges forsim ilar d ocu m ents by that test sp onsor .

The rep ort m u st be m ad e available w hen resu lts are m ad e p u blic. In ord er tou se the p hrase "TPC Benchm ark™C", the Fu ll Disclosu re Rep ort m u st have been su bm itted to the TPC Ad m inistratoras w ell as w ritten p erm ission obtained to d istribu te sam e.8.3Revisions to the Full D isclosure Report8.3.1In ad d ition to the requ irem ents for revising the Full Disclosu re Rep ort fou nd in the cu rrent revision ofthe TPC Pricing Sp ecification , the follow ing com p onents in the p riced configu ration m ay be su bstitu ted if they areno longer ord erable:TPC Benchm ark™C - Stand ard Sp ecification, Revision 5.11 - Page 98 of 130front-end system sd isks, d isk enclosu res, external storage controllersterm inal serversnetw ork ad ap tersrou ters, brid ges, rep eaters, sw itchescables8.3.2Su bstitu tion of the Server or the H ost system , OS, DBMS or TP Monitor is not allow ed u nd er anycircu m stances.TPC Benchm ark™C - Stand ard Sp ecification, Revision 5.11 - Page 99 of 130Clause 9: AUD IT9.1General Rules9.1.1An ind ep end ent au d it of the benchm ark resu lts by an au d itor certified by the TPC is requ ired .