,
HYBRID HEURISTICS FOR THE PERMUTATION
FLOW SHOP PROBLEM
M.G.RA VETTI,F.G.NAKAMURA,C.N.MENESES,M.G.C.RESENDE,G.R.MATEUS,
AND P.M.PARDALOS
A BSTRACT.The Flow Shop Problem(FSP)is known to be NP-hard when more than
three machines are considered.Thus,for non-trivial size problem instances,heuristics
are needed to?nd good orderings.We consider the permutation case of this problem.
For this case,denoted by F|prmu|Cmax,the sequence of jobs has to remain the same at
each machine.We propose and test two hybrid heuristics,combining elements from the
standard Greedy Randomized Adaptive Search Procedure(GRASP),Iterated Local Search
(ILS),Path Relinking(PR)and Memetic Algorithm(MA).The results obtained are shown
to be competitive with existing algorithms.
1.I NTRODUCTION
The Flow Shop Problem(FSP)is a scheduling problem in which n jobs have to be pro-cessed by m machines.The problem is to?nd the sequence of jobs for each machine to minimize completion time,also known as makespan(6).This problem is NP-Hard for m>3(8;13).Several papers in the literature address this problem,proposing models, heuristics,and bounds.Dannenbring(9)tested several heuristics.Nawaz,Enscore,and Ham(16)presented a polynomial time algorithm(NEH),?nding interesting results.Until now,NEH is one of the best polynomial time heuristics for this problem.Taillard(26) presented an improvement in the complexity of the NEH algorithm,a heuristic based on tabu search,and a useful characterization of the distribution of the objective function.Tail-lard(27)proposed a series of test problems with strong upper bounds.The running time required by Taillard’s tabu search heuristic was not given and he focused on solution qual-ity.Ben-Daya and Al-Fawzan(7)implemented and tested an improved variant of Taillard’s tabu search,reporting times and comparing their performance with Ogbu and Smith’s sim-ulated annealing algorithm(17).However,they did not match all of Taillard’s results for large instances.St¨u tzle presented and tested an Iterated Local Search(ILS)heuristic ob-taining good results.Ruiz and Maroto(21)compared25methods,from very basic ones, such as Johnson’s algorithm(12),to more sophisticated ones,such as tabu search and sim-ulated annealing.The results of their study concluded that NEH was the best polynomial-time heuristic,while St¨u tzle’s ILS(25)and Reeves’s genetic algorithm(18)were the best metaheuristic-based heuristics.Ruiz et al.(22)proposed a new memetic algorithm for this problem,obtaining improved results when compared with the ILS and tabu search.The same authors followed up with another paper in the same direction(23),proposing and test-ing two genetic algorithms and obtaining strong results.Agarwal,Collak,and Eryarsoy(1) implemented a heuristic improvement procedure based on adaptive learning and applied it to the NEH algorithm,leading to additional improvements.However,for larger instances, their results were of poor quality and their algorithm was computationally intensive.These Date:November5,2006.
Key words and phrases.Heuristics,GRASP,Iterated Local Search,Memetic Algorithm,Flow Shop Problem.
AT&T Labs Research Technical Report TD-6V9MEV..
1
2M.G.RA VETTI,F.G.NAKAMURA,C.N.MENESES,M.G.C.RESENDE,G.R.MATEUS,AND P.M.PARDALOS results seemed to present a few issues,that we discuss in Section5.Ru′?z and St¨u tzle(24) described a simple algorithm with which they obtained good results and presented six new upper bounds.
In(1;7;27)and almost all other papers dealing with the FSP,the problem of interest was a special case called Permutational Flow Shop Problem(PFSP)in which the jobs have identical ordering sequences on all machines.This widespread approach is useful because it allows a simpler implementation,specially for genetic algorithms,and it is known that a PFSP solution is a good approximation of the FSP solution.
In this paper,we consider the PFSP and propose two hybrid approaches using GRASP-ILS-PR and a memetic algorithm.The paper is organized as follows.In Sections2and3, we describe the basic and the memetic algorithm.In Section3we introduce the proposed hybrid algorithms.In Section5,we present the computational experiments and their ana-lyzes.Finally,in Section6we make some concluding remarks.
2.B ASIC ALGORITHMS
As it is well known,a traditional GRASP and ILS have two main phases,a construction phase and a local search.Readers unfamiliar with GRASP and ILS are referred to(10;11;
19)and(14;25),respectively.In our implementation,we also use path-relinking(20). 2.1.Construction https://www.doczj.com/doc/5811897804.html,ually,the construction phase is a greedy algorithm,with some randomness,in which a new solution is obtained at each iteration.In our case,we use the NEH heuristic to construct the?rst solution.The pseudo-code of the NEH algorithm is presented in Algorithm1.
Algorithm1NEH Algorithm
1:Sort the jobs by decreasing sums of processing times.
2:Schedule the?rst two jobs minimizing the partial schedule.
3:for i=3to n do
4:Insert the i-th job in one of the i possible places of the partial solution,minimizing the partial schedule.
5:end for
2.2.GRASP.In a standard GRASP,at each iteration a new solution is constructed,usu-ally blending a random procedure with a greedy heuristic.In the heuristic proposed in this paper,we maintain a pool of good-quality solutions on which to apply a path-relinking pro-cedure.After20iterations without improvement,we reconstruct the pool using a partial NEH algorithm while preserving most of the structure of the current best solution.
2.3.ILS.As in the implementations of ILS,the GRASP-ILS-PR heuristic has a perturba-tion phase which changes the current solution by making two-swap movements at random positions.If an improved solution is found,the current solution is updated;if no improve-ment is found,the current solution is updated with probability p T>0.This is a simulated-annealing-like acceptance criterion(24;25).If n is the number of jobs,m is the number of machines,and p i j is the processing time for job i at machine j,then the temperature T used in the acceptance process is
T=0.5·∑n i=1∑m j=1p i j n·m·10
.
HYBRID HEURISTICS FOR FLOW SHOP SCHEDULING 3
2.4.Local Search.We use a well-known local search that has been previously applied to this and other combinatorial optimization problems.Tabu search,ILS,and genetic algo-rithms have used the same type of local search scheme,e.g.(24).This scheme,which we call LSInsertion is based on the insertion procedure of the NEH algorithm.For every job,we remove the job from its original position and attempt to insert it in one of the n possible positions.If an improvement is obtained,then the process is repeated.This pro-cedure works by choosing the positions at random and terminates after a complete search without improvement.
2.5.Path-relinking.Path-relinking (PR)is used as a form of intensi?cation of the search procedure.PR is applied from the incumbent solution to a solution randomly selected from the pool (20).Beginning at a random position and considering only swaps,three possible moves are analyzed and the best move is selected.At each step,the new solutions are analyzed and the best one is saved.In Figure 1an example of the procedure is shown.A similar approach and some variations can be found in (20).
Guiding Sol
Initial Sol
2571346
6435172253461725137462371546
6
457132S
S
2713546F IGURE 1.Example of the path-relinking procedure.At each step,a random position is chosen and from this point,in a cyclic order,three possible changes are analyzed.The letter ‘S’indicates which solution is selected for the next step.This solution corresponds to the best move analyzed.Only ?ve steps are performed if the guiding solution is not reached ?rst.
The PR procedure terminates when the whole path between the solutions is analyzed or after testing a small number of solutions in the path,returning the best solution found in the process.In our implementation,we only allow ?ve steps,and at each step only three possible solutions are analyzed.
4M.G.RA VETTI,F.G.NAKAMURA,C.N.MENESES,M.G.C.RESENDE,G.R.MATEUS,AND P.M.PARDALOS The main characteristic of the PR is that at each new step the solutions maintain most of the original structure.When the procedure begins from the best solution,the goal is to preserve most of this successful structure with some in?uence of the selected pool solution.
https://www.doczj.com/doc/5811897804.html,bining these ideas results in the hybrid GRASP-ILS-PR heuris-tic whose pseudo-code is shown in Algorithm2.In line1of the pseudo-code,the jobs are sorted in decreasing order of the sums of their processing times and the order is saved in the array JobSorted.In line2,the NEH heuristic is called to produce the initial solution that is inserted in the pool in line
3.In line4,the local search is applied on the initial solution and the local optimum can itself be inserted into the pool in line5.NonImproves, the counter of iterations without incumbent solution improvement,is initialized in line7. The loop in lines8to38constitutes the main iterations of the heuristic.From line9 to line11the current solution CurrSol is perturbed and the local search is applied to the perturbed solution.Lines13to21are only executed every PRF iterations,where PRF is the path-relinking frequency parameter.In line13the path-relinking procedure is applied from the incumbent solution(BestSolution)to a solution selected at random from the pool(PoolSolution).If certain criteria are met,the path-relinking solution(SolPR)is inserted into the pool in line1
4.In lines15to21and23to29the current solution can be updated when the current solution is improved or with a positive probability otherwise.The pool of solutions is replaced after a certain number of non-improving iterations(NImp). This pool of solutions contains a set of good,or elite,solutions found during the process. As we can see in Algorithm2,after each local search and after the path-relinking,the algorithm tries to introduce the new solution into the pool.For this insertion,not only the quality of the solution is considered but also the similarity with the other solutions in the pool.This idea is not new and the reader can refer to(3)for example.In our case we do not allow the insertion of solutions that are too similar to the ones already in the pool.A solution is inserted if its objective value is the best so far or if its objective value is better than the worst objective value in the pool and the solution is different enough from the solutions in the pool.
The similarity between two solutions is computed by counting the number of different jobs for a certain position.The importance of a diversi?ed pool is discussed in the next section.
3.M EMETIC A LGORITHM
When working with the PFSP we only deal with a permutation vector,which has an easy representation.This is one of the main advantages of working with the PFSP.The idea of the post-optimization is to use the information gathered by the GRASP-ILS-PR heuristic to search for new solutions near solutions in the pool.For this reason,we use a memetic algorithm(MA)based on the experience of Ruiz and Maroto(21)and Ruiz,Maroto,and Alcaraz(23).The MA works using the pool of solutions produced by the GRASP-ILS-PR as the initial population in addition to a number of random solutions to allow it to search other regions of the feasible solution space.
As the MA is used along with GRASP-ILS-PR,we need a simple structure and only three operators.A mutation procedure and a path crossover,and a cold restart are the only operators used.These operators are described next.
3.1.Mutation Operator(M).This operator works as a perturbation approach of GRASP-ILS-PR.Its main goal is to allow the algorithm to search beyond the neighborhood de?ned
HYBRID HEURISTICS FOR FLOW SHOP SCHEDULING5 Algorithm2GRASP-ILS-PR Algorithm.Solution,CurrSol and BestSolution, represent different solutions and their corresponding makespan values.PoolSolution is a random solution chosen from the pool.
1:JobSorted←Sort the jobs by decreasing sums of processing times;
2:Solution←NEH(JobsSorted);
3:PoolInsertion(Solution);
4:Solution←LSInsertion(Solution);
5:PoolInsertion(Solution,Proximity);
6:CurrSol←BestSolution;
7:NonImproves←0;
8:for i←1,...,MaxIterations do
9:Solution←Perturbation(CurrSol);
10:Solution←LSInsertion(Solution);
11:PoolInsertion(Solution,Proximity);
12:if i mod PRF=0then
13:SolPR←PathRelinking(BestSolution,PoolSolution);
14:PoolInsertion(SolPR,Proximity);
15:if CurrSol>SolPR then
16:CurrSol←SolPR;
17:else
18:if RND(0,1) 19:CurrSol←SolPR; 20:end if 21:end if 22:end if 23:if CurrSol>Solution then 24:CurrSol←Solution; 25:else 26:if RND(0,1) 27:CurrSol←Solution; 28:end if 29:end if 30:if BestSolution is improved then 31:NonImproves←0; 32:else 33:NonImproves←NonImproves+1; 34:end if 35:if NonImproves=NImp then 36:RestartPOOL(); 37:end if 38:end for by the local search.This operator makes two swap moves at random positions.It is applied only to the best or second best solution,randomly chosen in the population. 3.2.Path Crossover(PX).This procedure is similar to path-relinking and it is used as a crossover by Ahuja et al.(2)for solving quadratic assignment problems.After the selection of two parents,the crossover consists of the construction of a path between those parents. 6M.G.RA VETTI,F.G.NAKAMURA,C.N.MENESES,M.G.C.RESENDE,G.R.MATEUS,AND P.M.PARDALOS 2571346 2713546 P1 P2 2 2 5 5 7 7 1 1 3 3 4 4 6 6 O1 O2 X S 2 2 2 2 5 5 5 5 3 3 3 3 7 7 7 7 1 1 1 1 4 4 4 4 6 6 6 6 P1 P3 O3 O4 6 6 6 6 4 4 4 4 2 2 2 2 1 1 1 1 5 5 5 5 3 3 3 3 7 7 7 7 P3 P4 O5 O6 X S Return the Best Offspring 2546 1 237546 1 O2 O3 73 S X = O6 F IGURE2.PX–S indicates the offspring with better objective function and X the discarded offspring.The?rst position to check is randomly chosen. This operator begins the procedure from a random position and continues until it reaches this position again.The parent’s alleles(jobs)are compared.If they are the same,then the offspring inherits that allele.If the alleles differ,two distinct swap moves can be done,one on each parent.The algorithm always selects the move which results in a better objective function value.The offspring resulting from this crossover is the best solution generated during the construction of the path.Figure2shows an example of the PX operator. We also tested a few other crossover operators,such as multi-parent and similar block order crossovers,but these did not produce results as good as PX or were excessively expensive to compute. 3.3.Cold Restart.Cold restart is used in many genetic algorithms with the idea of cre-ating a more diversi?ed population.This is useful when the algorithm reaches a state of stagnation,that is,when no improvement occurs after a certain number of generations. Cold restart changes the population by modifying,replacing,and maintaining part of the solutions.The three best solutions are left unchanged.Approximately half of the popula-tion results from applying the mutation operator on one of the three best solutions.About one fourth of the population is generated by applying the iterated greedy algorithm of Ruiz and St¨u tzle(24)to the incumbent solution.The remaining elements of the population are replaced by randomly generated solutions. 4.H YBRID ALGORITHMS The hybrid algorithms combine the GRASP-ILS-PR and MA heuristics.This combina-tion can be done in several ways,but in this paper we use two approaches called Hybrid A and B. HYBRID HEURISTICS FOR FLOW SHOP SCHEDULING7 Initially,the solution of the polynomial-time heuristic NEH is used as the initial pool solution.If this solution can be locally improved,then the improved solution is also in-serted into the pool.In both cases the pool of elite solutions is used by the path-relinking component of the hybrid heuristics. The difference between both hybrid approaches lays in when the MA heuristic is called. In the?rst algorithm,we use the MA as a post-optimization procedure to intensify the search around the best solutions found during GRASP-ILS-PR phase.After running GRASP-ILS-PR for a?xed number of iterations,the MA uses the pool of elite solutions produced in the GRASP-ILS-PR phase as its initial population.Path-relinking also plays a role in this hybrid heuristic in that it contributes to the pool as well as makes use of elements of the pool for intensi?cation. In the second approach,the GRASP-ILS-PR algorithm calls the MA algorithm after the algorithm reaches a state of stagnation for a?xed number of iterations.In this case,the MA algorithm uses the pool of elite solutions as the initial population for a certain number of generations.After that,the GRASP-ILS-PR continues to run with the pool provided by the MA output. The algorithm structures are shown in Figure3,for now on we will refer to these cases as Hybrid A and Hybrid B,respectively.The diversity of solutions in the pool is critical in both the GRASP-ILS-PR and MA phases.It allows different promising regions of the solution space to be explored. F IGURE3.Hybrid algorithm structures. 5.C OMPUTATIONAL R ESULTS To analyze the performance of the hybrid heuristics,we designed several experiments using benchmark instances proposed by Taillard(26)as our testbed.These experiments are composed of two parts.In the former,we de?ne tests to calibrate certain parameters of our algorithm and in the latter,we analyze the behavior of the algorithm and its performance. All the algorithms tested in this paper were implemented by the authors in the C pro-gramming language and compiled with GNU GCC version3.2.3,using compiler options -O6-funroll-all-loops-fomit-frame-pointer-march=pentium4.CPU times were 8M.G.RA VETTI,F.G.NAKAMURA,C.N.MENESES,M.G.C.RESENDE,G.R.MATEUS,AND P.M.PARDALOS computed using the function getrusage().The experiments for the performance analy-sis were run on a Dell PowerEdge2600computer with dual3.2GHz1Mb cache XEON III processors and6Gb of memory RAM under a Red Hat Linux3.2.3-53.The algo-rithm used for random-number generation is an implementation of the Mersenne Twister algorithm described in(15). 5.1.Calibration.We?rst conduct a set of experiments to engineer the algorithm,i.e. set parameters and load balance the various components of the algorithm.Though we conducted extensive computational testing with numerous crossover operators,initial pool solution heuristics,local search algorithms,path-relinking schemes,population and pool sizes,and other parameters,we limit this discussion to the calibration of the memetic algorithm when using PX as a crossover operator,the results of adding path-relinking to the GRASP-ILS-PR heuristic,and the relationship between both components of the hybrid heuristic. We determine the parameter PRF(frequency of path-relinking),the crossover operator to be used in the MA,and the load balance between the GRASP-ILS-PR and MA compo-nents. 5.1.1.Frequency of path-relinking(PRF).The GRASP-ILS-PR heuristic works with a pool of15solutions.Since path-relinking is not activated at each GRASP-ILS-PR iteration, the main parameter to be analyzed is the frequency in which it is called.This is parameter PRF in Algorithm2. To test the effect of using path-relinking,we compared?ve versions of the hybrid al-gorithm which differed only with respect to the PRF parameter.This included a variant in which PRF=0,i.e.the pure ILS proposed by St¨u tzle(25).To perform these tests,we allowed the algorithm to run for200CPU seconds,regardless of instance size.The30cho-sen instances have50,100,and200jobs,with20stages,and they proved to be the most dif?cult of this benchmark.The use of time as a stopping criterion was considered because the path-relinking strategy directly affects CPU time and it is this tradeoff of quality and time that we want to understand. For each instance tested,each variant was run?ve times using different random number generator seeds and the average percentage increase over the best known solution(API) was calculated.These results are summarized in Table1.API is computed considering the best upper bound known to the date.The formula for API is given by API= 1R∑i=1 X?UB ×100 , where R is the number of repetitions,X the solution of the heuristic,and UB the best known upper bound. The results in Table1indicate that GRASP-ILS-PR with path-relinking outperforms the pure ILS.Furthermore,on average the use of path-relinking frequency parameter PRF=2 results in a better performance than any other frequency tested. 5.1.2.Crossover comparison.The memetic algorithm component of the hybrid heuristic has several parameters that need to be set.We conducted extensive experimentation to set these parameters.These experiments resulted in the following parameter choices.The size of the population was set to20solutions of which?ve are randomly generated at each new generation.The probability of applying the crossover is40%,the mutation1%, and after applying the operators,local search is carried out with a probability of5%.To avoid super?uous computations,all the solutions have a?ag that indicates if local search HYBRID HEURISTICS FOR FLOW SHOP SCHEDULING9 T ABLE1.Average percentage increase over the best known solution for GRASP-ILS-PR varying the path-relinking frequency parameter PRF. Jobs×Machines PRF=2PRF=5PRF=10PRF=20Pure ILS 1.36104 1.78701 1.70909 1.49610 1.76623 2.11123 2.00864 1.77106 1.76026 2.12203 2.52129 2.39495 2.01044 2.48283 2.42790 1.57358 1.57358 1.68099 1.83673 1.53598 50×20 1.54528 1.53974 1.47328 1.71144 1.62836 1.74763 1.81275 1.97558 1.94844 1.78019 1.57085 1.29015 1.67881 2.04049 1.73819 2.53048 2.96938 2.86643 2.62260 2.92062 1.84344 1.68314 1.58696 1.83810 1.46407 1.29546 1.09371 1.12025 1.22113 1.39634 Average 1.810 1.815 1.787 1.896 1.878 2.51532 2.56369 2.52499 2.42180 2.73460 2.17911 2.30197 2.17911 2.17265 2.16618 2.09217 2.01882 2.03795 2.21336 2.17509 1.80252 1.82166 1.80252 1.88228 1.80252 100×20 2.30915 2.35350 2.05892 2.22363 1.95439 2.22816 2.31301 2.20930 2.16216 2.22187 2.11232 2.23676 2.08679 2.27186 2.34844 2.78706 2.75269 2.83706 2.56522 2.59334 2.82709 2.78247 2.76335 2.75697 2.80478 1.82157 1.82468 1.75008 1.94591 2.04228 Average 2.267 2.297 2.225 2.2616 2.284 1.74185 1.82939 1.76686 1.82582 1.81510 2.36722 2.39579 2.37079 2.40650 2.35116 2.37578 2.54346 2.53993 2.45521 2.45698 1.98512 2.09846 2.21179 2.23481 2.25961 200×20 1.39977 1.64491 1.72129 1.72484 1.68043 2.25042 2.22540 2.05917 2.17535 2.17535 1.92118 1.95109 2.05489 1.88600 1.90887 2.23222 2.25163 2.28163 2.34869 2.33104 2.10686 2.20515 2.32666 2.46247 2.34989 2.56741 2.52851 2.61339 2.77783 2.70179 Average 2.095 2.167 2.195 2.230 2.203 Overall Average 2.057 2.093 2.069 2.129 2.122 Overall Standard Deviation0.4170.4440.4190.3720.411 was applied to the solution.After completing a generation,local search is applied to the best solution with a probability of10%if its?ag indicates that local search has not been previously applied.We next compare the performance of the memetic algorithm when using two crossovers. After comparing several crossovers,Ruiz,Maroto,and Alcaraz(23)choose the Simi-lar Block Order Crossover(SBOX),which presented a better performance than the other operators.The SBOX transfers to the offspring similar blocks of alleles from its parents. Two sequences of alleles are considered a similar block if there are at least two consecutive identical alleles in the same position in both parents.After the recognition of the similar blocks,a crossover point is randomly chosen and the remaining alleles are copied from their parents.By using the crossover point it is possible to generate two offsprings after each crossover.We compare the performance of SBOX with the PX operator described in Section3.2. As seen in Table2,we were unable to reproduce the results obtained by Ruiz,Maroto, and Alcaraz(23).The difference between the performances could be attributed factors such as the stochastic nature of the heuristics,differences in coding,data structure design, compilers,hardware discrepancies,or even computer con?guration.Furthermore,Ruiz, Maroto,and Alcaraz report elapsed time instead of CPU time. 10M.G.RA VETTI,F.G.NAKAMURA,C.N.MENESES,M.G.C.RESENDE,G.R.MATEUS,AND P.M.PARDALOS To evaluate these operators,once again we use the30instances from Taillard’s bench-mark with50,100,and200jobs and20stages.The experiment consisted of allowing each algorithm to perform?ve runs with different seeds where both algorithms are identical with the exception of the crossover used.Table2shows the results,where for each operator we present the API and the number of times in which the best solution was found using the crossover. T https://www.doczj.com/doc/5811897804.html,parison between crossover operators listing API,the average percentage increase over the best known solution,and the number of times in which each variant obtained the best solution in the comparison. SBOX crossover PX crossover Jobs×Machines API times best found API times best found 50×20 2.288904 2.251866 100×20 2.605786 2.576156 200×20 2.380823 2.416227 We use the‘times best found’metric because sometimes the average value does not show all the information needed to make the correct choice.In this particular case,we can see that for the200-job instances the variant with the SBOX crossover presents a lower average value but the algorithm using PX crossover obtains better results more times.In these tests,the use of PX as crossover has a better average performance than SBOX,?nding best solutions in most cases.All other parameters used in the algorithm were?xed in the comparison. 5.1.3.Load balancing GRASP-ILS-PR and MA for Hybrid A.When combining compo-nents GRASP-ILS-PR and MA into a hybrid heuristic we need to determine what portion of the total running time will be allocated to each component.We call this choice load balancing.We conducted experiments to determine a good load balancing.Once again, we consider the same instances as before and use the200CPU seconds as the stopping criterion.The experiment consisted of allowing each variant(with distinct load balance)to perform?ve runs with different seeds.The possibilities of using the pure GRASP-ILS-PR or the pure MA are also considered in the experiment. Table3and Figure4present the results.The?gure shows the con?dence intervals for the different con?gurations of the hybrid algorithm.The variants are represented by their respective load balances.For example,variant95/5has a load balance with95%of the CPU time used by the GRASP-ILS-PR and5%by the MA.As can be seen in the table and ?gure,the95%-5%load balance variant has the best average performance. 5.2.Performance Analysis.We next investigate the algorithm’s dependence on the ini-tial random number generator seed and in its running time.The tables below show the performance of the algorithm after ten runs for each instance.In these experiments,we use number of iterations as the stopping criterion,so that the experiments can be more accurately reproduced. Hybrid B does not need a load balancing calibration,but after performing similar ex-perimentation two important parameters were set,the algorithm calls the GA only after HYBRID HEURISTICS FOR FLOW SHOP SCHEDULING11 T ABLE3.Average relative percentage deviation for different GRASP-ILS-PR and MA load balances.G/M indicates G%of running time allocated to GRASP- ILS-PR and M%of running time allocated to MA. Jobs×Machines GRASP-ILS-PR95/580/2050/5020/805/95Memetic 1.39221 1.36104 1.53247 1.48052 1.75065 1.71429 1.91169 2.20302 2.11123 2.23542 1.87905 2.32721 2.23002 2.12203 2.58720 2.52129 2.52678 2.48833 2.55974 2.60368 2.50481 1.76155 1.57358 1.79914 1.69710 1.79914 2.05693 2.10526 50×20 1.78898 1.54528 1.78898 1.78344 1.91083 2.07699 2.06037 2.05156 1.74763 2.05156 2.05156 2.05156 2.10583 2.09498 1.76518 1.57085 1.79757 1.74359 1.84615 1.81916 2.35358 2.56841 2.53048 2.56299 2.56299 2.91520 3.00732 3.07234 1.85413 1.84344 1.85413 1.85413 1.91825 1.82207 2.23350 1.46005 1.29546 1.39634 1.41757 1.40165 1.08840 2.05999 Average 1.94323 1.81003 1.95454 1.89583 2.04804 2.05247 2.25186 2.47662 2.51532 2.75395 2.72170 2.73460 2.88617 3.02806 2.16618 2.17911 2.26964 2.54122 2.64791 2.75461 3.01326 2.09217 2.09217 2.14958 2.21017 2.16233 2.58970 2.49721 1.81528 1.80252 1.80890 1.83442 1.73233 2.04498 1.88547 100×20 2.27431 2.30915 2.36300 2.50871 2.41368 2.22997 2.44219 2.23130 2.22816 2.22816 2.32558 2.70585 2.69956 2.66813 2.11232 2.11232 2.09317 2.17294 2.29419 2.53350 2.59413 2.85268 2.78706 2.85893 2.99328 3.06827 3.16825 2.74957 2.82709 2.82709 2.83665 2.89721 2.87809 3.04064 3.05976 1.82157 1.82157 1.89307 2.02362 2.15107 1.83712 2.11999 Average 2.26695 2.26745 2.32551 2.42289 2.47883 2.57845 2.60578 1.81331 1.74185 1.81331 1.82760 1.91157 1.83117 1.95266 2.50647 2.36722 2.49755 2.55467 2.46006 2.41899 2.53861 2.52405 2.37578 2.53993 2.54699 2.52758 2.46404 2.60524 2.27554 1.98512 2.35346 2.52701 2.44200 2.43492 2.62617 200×20 1.57030 1.39977 1.66800 1.65024 1.78169 1.64491 1.89182 2.37912 2.25042 2.38806 2.39878 2.29154 2.32192 2.47386 2.06193 1.92118 2.08128 2.03730 2.00739 1.79803 2.17628 2.25693 2.23222 2.24634 2.38574 2.45280 2.54632 2.54632 2.18728 2.10686 2.28020 2.20872 2.34811 2.44282 2.41780 2.68411 2.56741 2.73362 2.78490 2.86270 2.70179 2.93343 Average 2.22590 2.09479 2.26018 2.29220 2.30854 2.26049 2.41622 Overall Average 2.14536 2.05742 2.18007 2.20364 2.27847 2.29714 2.42462 a stagnation of20iterations,and it is allowed to run20generations after returning the improved pool of elite solutions. 5.2.1.Time to target.Many local search based combinatorial optimization heuristics,in-cluding GRASP,GRASP with path-relinking,and memetic algorithms have running time to the optimal solution that are distributed according to a shifted exponential distribution (4).To study the random variable time to target solution value we ran Hybrid A and Hy-brid B using one hundred different seeds for one instance with100jobs and20stages1. The seeds were randomly generated and are distinct from the ones used in the calibration process. Hybrid A uses95%of the processing time allowed with GRASP-ILS-PR and the last 5%with the MA.In this test the maximum CPU time allowed to each run is10,000sec-onds.For the Hybrid B algorithm,the MA is called after a period of20iterations without improving the incumbent solution,and after20generations the MA algorithm returns the pool to the GRASP-ILS-PR algorithm. A target value of2.5%above the best known solution was chosen for this experiment. For this particular case,the best known solution has a value of6314,therefor the target 1The instance used is this experiment is one from Taillard’s benchmark,called ta085. 12M.G.RA VETTI,F.G.NAKAMURA,C.N.MENESES,M.G.C.RESENDE,G.R.MATEUS,AND P.M.PARDALOS 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 0/100 5/9520/8050/5080/2095/5 100/0A v e r a g e p e r c e n t a g e i n c r e a s e o v e r b e s t k n o w n s o l u t i o n % time GRASP-ILS / % time MA F IGURE 4.95%con?dence intervals for hybrid algorithms. was set to 6471.The algorithms stop after obtaining a solution less or equal to 6471or after 10,000seconds.Time-to-target plots (ttt-plots)are produced for these values using (5).Figure 5shows the ttt-plots obtained. Of interest is that within 80seconds,almost 80%of the cases for both algorithms reached the target and within 1000seconds more than 90%.These results show that the choice of the seed is not a critical aspect for obtaining a reasonable performance.By this comparison it is possible to conclude Hybrid A presents a performance slightly better than Hybrid B. 5.2.2.Performance.In this section,we present the results obtained by using our algorithm in the 120instances from Taillard’s Benchmark.The algorithm performs a ?xed number of iterations for both the GRASP-ILS-PR and the memetic algorithm components. Some authors use a relation between the number of jobs and the number of machines to set a time for testing the performance of the algorithms.The problem with this approach is that for this particular benchmark,the dif?culty of the instances does not seem to in-crease with an increase in the number of jobs.We resume this discussion in next section.In Tables 4and 5,the average values of our experiments are presented,for both hybrid algorithms.Not only are these results competitive,but it is possible to see that the stan-dard deviation (StdDev)is low,indicating a robust method.When comparing the hybrid approaches,it is possible to observe,the performance of Hybrid A is slightly better than Hybrid B,which is consistent with the time to target experiment. 5.2.3.Remarks regarding the instances.It is interesting to analyze the dif?culty of our algorithms for some instances.It is quite obvious and expected that the increase in the HYBRID HEURISTICS FOR FLOW SHOP SCHEDULING 13 F IGURE 5.CPU time (seconds)needed to reach a solution no further than 2.5%of the best known upper bound.The time is on a logarithmic scale. 0.2 0.4 0.6 0.8 1 1 10 100 1000 10000 100000 c u m u l a t i v e p r o b a b i l i t y time to target solution (2.5%) Exponencial distribution comparison Hybrid A Hybrid B T ABLE 4.Average relative percentage increase for Taillard instances.Column CPU 1indicates the average time (in CPU seconds)spent to ?nd the best solution.Column CPU 2indicates the average time (in CPU seconds)taken by Hybrid A. Hybrid A Jobs ×Machines API Best Worst StdDev CPU 1CPU 220×50.0790.000 1.1380.2340.28 5.0620×100.3280.000 1.1940.327 5.4418.7420×200.2240.0000.8260.2338.0028.6450×50.0030.0000.0710.01415.18125.2450×100.7130.000 2.4070.602206.101186.9250×20 1.2220.324 3.7910.7831743.904647.40100×50.0120.0000.2850.044129.16680.38100×100.2560.000 1.2630.270322.321422.72100×20 1.4300.702 2.1830.3881905.284350.28200×100.2700.0090.7540.164498.121433.04200×20 1.6720.782 2.6150.3422523.444218.04500×200.9810.570 1.459 0.213 2493.02 3894.82 Average 0.604 number of machines increases the effort needed by our algorithms to ?nd a good solution.In part,this could be explained by considering that the makespan for an speci?c permuta-tion can be calculated by using a function depending on the number of jobs and machines, 14M.G.RA VETTI,F.G.NAKAMURA,C.N.MENESES,M.G.C.RESENDE,G.R.MATEUS,AND P.M.PARDALOS T ABLE5.Average relative percentage increase for Taillard instances. Column CPU1indicates the average time(in CPU seconds)spent to ?nd the best solution.Column CPU2indicates the average time(in CPU seconds)taken by Hybrid B. Hybrid B Jobs×Machines API Best Worst StdDev CPU1CPU2 20×50.0580.0000.8100.1670.439 4.584 20×100.2750.000 2.0110.382 3.47214.813 20×200.2310.0000.6530.186 4.40428.169 50×50.0190.0000.3530.0718.02693.952 50×100.7000.000 1.6050.485260.7491179.363 50×20 1.3900.297 3.7870.6031479.5744108.018 100×50.0060.0000.0950.020153.866687.559 100×100.2580.0170.8580.256263.2911175.743 100×20 1.8570.798 3.6240.5501096.5543575.116 200×100.2570.0090.7540.184287.0791044.240 200×20 1.8630.835 2.7660.4311105.3032129.254 500×200.9020.471 1.3880.2033149.8584816.021 Average0.651 and the time taken on these evaluations explains the need of extra time to reach the same result quality. Nevertheless,it is quite remarkable to notice that the worst performance of our algo-rithms happens for the group of instances with100jobs and20stages(100×20)and for 200×20.And these results are consistent with previous results no matter the method used. The main question to answer is why for500×20the algorithm has a better performance. In Taillard’s webpage we can?nd that for500×20there are seven out of ten instances with results proved to be optimal solutions,for200×20there four out of ten optimal solu-tions and for the case100×20only one optimal result has been found so far and none for 50×20. Next we discuss some new facts that need review.In2006,Agarwal et al.,(1)discuss an approach to solve the PFSP.In that paper,the authors claim to have found a better upper bound for one instance of Taillard’s benchmark by using an adaptive-learning approach (over well known polynomial heuristics).The instance is ta040(50jobs and5stages). In the paper,they found a permutation with2774as a makespan value,but they did not present the actual permutation.The problem is that in Taillard’s website2the optimal value for this instance is2782.Something similar occurs with instance ta068(100jobs and5stages)for which we point out two problems.First,the authors incorrectly use an upper bound of5034,while the correct upper bound,and supposedly also optimal value is5094.While this could be only a typo,we point out a second problem.The solutions found for two methods presented in that paper have5082as a makespan value,which is less than the optimal solution value.Furthermore,the upper bounds for most of the larger instances are exchanged with the corresponding lower bounds.For those cases,the results presented by the authors tend to be better than they actually are.It is important to review these questions to avoid future dif?culties for other researchers,especially because these instances are referenced in almost every paper dealing with FSP.For future publications,it seems to be more appropriate to present the actual permutation when a new upper bound or a signi?cant result is reached. 2http://ina2.eivd.ch/Collaborateurs/etd/problemes.dir/problemes.html HYBRID HEURISTICS FOR FLOW SHOP SCHEDULING15 6.C ONCLUDING R EMARKS In this paper,we discussed the Permutational Flow Shop Problem.This problem has been extensively studied and there are several algorithms published with excellent perfor-mance.We presented two implementations of hybrid algorithms based on GRASP,ILS, PR,and a memetic algorithm.In one approach,we use the GRASP-ILS-PR algorithm?rst maintaining a pool of good and diversi?ed solutions in order to apply path-relinking and to construct a population for the memetic algorithm.In the second approach,we use the memetic algorithm to improve the pool of elite solutions every time the GRASP-ILS-PR algorithm reaches a state of stagnation for a speci?c number of iterations.The memetic al-gorithm uses a crossover based on path-relinking(PX)that was shown to be very effective. The best performance is obtained using Hybrid A,when we allow the GRASP-ILS-PR to perform a large search almost95%of the time and then the memetic algorithm try to obtain a better solution using the output of the previous one. The results are competitive,showing that for the hardest instances of Taillard’s bench-mark,i.e.the ones with100jobs and20stages,the heuristics can?nd a solution no worse than2.5%of the best known upper bound in less than80seconds with80%probability. We also discussed some characteristics about the dif?culty of the benchmark and some new results presented in a recent paper that need to be reviewed. This paper is an important step in the direction of a more ambitious project in which we consider a real-world scheduling problem.The next step in our research is the con-sideration of sequence dependent setups and due dates in some or all stages.We are also considering a multi-objective approach,trying to obtain a?exible algorithm that could face different practical circumstances.We believe that in these more complicated cases the use of these kinds of hybrid approaches will be most advantageous. A CKNOWLEDGEMENT This research was supported by CNPq,NSF,and U.S.Air Force Grants. R EFERENCES [1]A.Agarwal,S.Colak,and E.Eryarsoy.Improvement heuristic for the?ow-shop scheduling problem:An adaptive-learning approach.European Journal of Opera-tional Research,169(3):801–815,2006. [2]R.K.Ahuja,J.B.Orlin,and A.Tiwari.A greedy genetic algorithm for the quadratic assigment https://www.doczj.com/doc/5811897804.html,puters&Operations Research,27(10):917–934,2000. [3]R.M.Aiex,S.Binato,and M.G.C.Resende.Parallel GRASP with path-relinking for job shop scheduling.Parallel Computing,29:393–430,2003. [4]R.M.Aiex,M.G.C.Resende,and C.C.Ribeiro.Probability distribution of solution time in GRASP:An experimental investigation.Journal of Heuristics,8:343–373, 2002. [5]R.M.Aiex,M.G.C.Resende,and C.C.Ribeiro.TTTPLOTS:A perl program to create time-to-target plots.Technical report,AT&T Labs,Technical Report TD-6HT7EL, 2005.To appear in Optimization Letters. [6]K.R.Baker.Requirements planning.In Handbooks in OR&MS,editors S C Graves et al.1993. [7]M.Ben-Daya and M.Al-Fawzan.A tabu search approach for the?ow shop schedul- ing problem.European Journal of Operational Research,109(1):88–95,1998. [8]https://www.doczj.com/doc/5811897804.html,puter and Job-Shop Scheduling Theory.Wiley,New York,1976. 16M.G.RA VETTI,F.G.NAKAMURA,C.N.MENESES,M.G.C.RESENDE,G.R.MATEUS,AND P.M.PARDALOS [9]D.G.Dannenbring.An evaluation of?ow shop sequencing heuristics.Management Science,23(11):1174–1182,1977. [10]T.A.Feo and M.G.C.Resende.Greedy randomized adaptive search procedures.Jour- nal of Global Optimization,6(2):109–133,1995. [11]P.Festa and M.G.C.Resende.GRASP:An annotated bibliography.In C.C.Ribeiro and P.Hansen,editors,Essays and surveys in metaheuristics,pages325–367.Kluwer Academic Publishers,2002. [12]S.M.Johnson.Optimal two-and three-stage production schedules with setup times included.Nav.Res.Log.Quarterly,1:61–68,1954. [13]A.H.G.Rinnooy Kan.Machine scheduling problems:classi?cation,complexity and computations.The Hague:Nijhoff,1976. [14]H.Lourenc?o,O.Martin,and T.St¨u tzle.Iterated local search.In F.Glover and G.Kochenberger,editors,Handbook of metaheuristics.Kluwer,2003. [15]M.Matsumoto and T.Nishimura.Mersenne twister:a623-dimensionally equidis- tributed uniform pseudo-random number generator.ACM Transactions on Modeling and Computer Simulation,8:3–30,1998. [16]M.Nawaz,E.E.Enscore,and I.Ham.A heuristic algorithm for the m-machine,n-job ?ow-shop sequencing problem.Omega,11(1):91–95,1983. [17]F.A.Ogbu and D.K.Smith.The application of the simulated annealing to the solution of the n/m/cmax?ow shop https://www.doczj.com/doc/5811897804.html,puters&Operations Research,17(3):243–253,1990. [18]C.R.Reeves.A genetic algorithm for?owshop https://www.doczj.com/doc/5811897804.html,puters&Operations Research,22(1):5–13,1995. [19]M.G.C.Resende and C.C.Ribeiro.Greedy randomized adaptive search procedures. In F.Glover and G.Kochenberger,editors,Handbook of Metaheuristics,pages219–249.Kluwer Academic Publishers,2003. [20]M.G.C.Resende and C.C.Ribeiro.GRASP with path-relinking:Recents advances and aplications.Metaheuristics:Progress as Real Problem Solvers.Ibakari and Nonobe and Yaguira(eds).Springer,pages29–63,2005. [21]R.Ruiz and C.Maroto.A comprehensive review and evaluation of permutation ?owshop heuristics.European Journal of Operational Research,165(2):479–494, 2005. [22]R.Ruiz,C.Maroto,and J.Alcaraz.New genetic algorithms for the permutational ?owshop scheduling problem.MIC2003:The Fifth Metaheuristics International Conference,pages63.1–63.8,2003. [23]R.Ruiz,C.Maroto,and J.Alcaraz.Two new robust genetic algorithms for the?ow- shop scheduling problem.OMEGA,The International Journal of Management Sci-ence,34:461–476,2006. [24]R.Ruiz and T.St¨u tzle.A simple and effective iterated greedy algorithm for the per- mutation?owshop scheduling problem.European Journal of Operational Research, Available on-line,2006. [25]T.St¨u tzle.Applying iterated local search to the permutation?ow shop problem. Technical report,TU Darmstadt,AIDA-98-04,FG Intellektik,1998. [26]′E.Taillard.Some ef?cient heuristic methods for the?ow shop sequencing problem. European Journal of Operational Research,47(1):65–74,1990. [27]′E.Taillard.Benchmarks for basic scheduling problems.European Journal of Oper- ational Research,64(2):278–285,1993. HYBRID HEURISTICS FOR FLOW SHOP SCHEDULING17 (Mart′?n G.Ravetti)D EPARTAMENTO DE C I?E NCIA DA C OMPUTAC??AO,U NIVERSIDADE F EDERAL DE M I-NAS G ERAIS,B ELO H ORIZONTE,MG,B RAZIL. E-mail address:martin@dcc.ufmg.br (Fabiola G.Nakamura)D EPARTAMENTO DE C I?E NCIA DA C OMPUTAC??AO,U NIVERSIDADE F EDERAL DE M INAS G ERAIS,B ELO H ORIZONTE,MG,B RAZIL. E-mail address:fgnaka@dcc.ufmg.br (Claudio N.Meneses)D EPARTMENT OF I NDUSTRIAL AND S YSTEMS E NGINEERING,U NIVERSITY OF F LORIDA, G AINESVILLE,FL,USA. E-mail address:claudio@https://www.doczj.com/doc/5811897804.html, (Mauricio G.C.Resende)A LGORITHMS AND O PTIMIZATION R ESEARCH D EPARTMENT,AT&T L ABS R ESEARCH,180P ARK A VENUE,R OOM C241,F LORHAM P ARK,NJ07932USA. E-mail address:mgcr@https://www.doczj.com/doc/5811897804.html, (Geraldo R.Mateus)D EPARTAMENTO DE C I?E NCIA DA C OMPUTAC??AO,U NIVERSIDADE F EDERAL DE M I-NAS G ERAIS,B ELO H ORIZONTE,MG,B RAZIL. E-mail address:mateus@dcc.ufmg.br (Panos M.Pardalos)D EPARTMENT OF I NDUSTRIAL AND S YSTEMS E NGINEERING,U NIVERSITY OF F LORIDA, G AINESVILLE,FL,USA. E-mail address:pardalos@https://www.doczj.com/doc/5811897804.html, Web性能测试方案 1测试目的 此处阐述本次性能测试的目的，包括必要性分析与扩展性描述。 性能测试最主要的目的是检验当前系统所处的性能水平，验证其性能是否能满足未来应用的需求，并进一步找出系统设计上的瓶颈，以期改善系统性能，达到用户的要求。 2测试范围 此处主要描述本次性能测试的技术及业务背景，以及性能测试的特点。 编写此方案的目的是为云应用产品提供web性能测试的方法，因此方案内容主要包括测试环境、测试工具、测试策略、测试指标与测试执行等。 2.1测试背景 以云采业务为例，要满足用户在互联网集中采购的要求，实际业务中通过云采平台询报价、下单的频率较高，因此云采平台的性能直接决定了业务处理的效率，并能够支撑业务并发的压力。 例如：支撑100家企业用户的集中访问，以及业务处理要求。 2.2性能度量指标 响应时间（TTLB） 即“time to last byte”，指的是从客户端发起的一个请求开始，到客户端接收到从服务器端返回的响应结束，这个过程所耗费的时间，响应时间的单位一般为“秒”或者“毫秒”。响应时间＝网络响应时间+应用程序响应时间。 响应时间标准： 事务能力TPS（transaction per second） 服务器每秒处理的事务数； 一个事务是指一个客户机向服务器发送请求然后服务器做出反应的过程。 客户机在发送请求时开始计时，收到服务器响应后结束计时，一次来计算使用的时间和完成的事务个数。它是衡量系统处理能力的重要指标。 并发用户数 同一时刻与服务器进行交互的在线用户数量。 吞吐率（Throughput） 单位时间内网络上传输的数据量，也可指单位时间内处理的客户端请求数量，是衡量网络性能的重要指标。 吞吐率=吞吐量/传输时间 资源利用率 这里主要指CPU利用率（CPU utilization），内存占用率。 3测试内容 此处对性能测试整体计划进行描述，包括测试内容以及关注的性能指标。Web性能测试内容包含：压力测试、负载测试、前端连接测试。 3.1负载测试 负载测试是为了测量Web系统在某一负载级别上的性能，以保证Web系统在需求范围内能正常工作。负载级别可以是某个时刻同时访问Web系统的用户数量，也可以是在线数据处理的数量。例如：Web应用系统能允许多少个用户同时在线？如果超过了这个数量，会出现什么现象？Web应用系统能否处理大 第3章面向对象程序设计基础 【1】什么是Java程序使用的类？什么是类库？ [解答]：Java程序的基本单位是类。对象是对事物的抽象,而类是对对象的抽象和归纳，找出事物的共性,把具有共同性质的事物归结为一类,得出一个抽象的概念——类。类是具有相同属性和方法的一组对象的集合，类是在对象之上的抽象，对象则是类的具体化，一旦建立类之后，就可用它来建立许多你需要的对象。 Java的类库是系统提供的已实现的标准类的集合，是Java编程的API(Application Program Interface)，它可以帮助开发者方便、快捷地开发Java程序。# 【2】如何定义方法？在面向对象程序设计中方法有什么作用？ [解答]：方法的定义由两部分组成：方法声明和方法体。方法声明的基本格式如下： 返回类型方法名（形式参数） { … //方法体内容 } 方法声明包括方法名、返回类型和形式参数，方法的小括号是方法的标志；方法体为实现方法行为的Java语句。 在面向对象程序设计中，方法所起的作用是完成对类和对象属性操作。 【3】简述构造方法的功能和特点。下面的程序片段是某学生为student类编写的构造方法，请指出其中的错误。 void Student(int no,String name) { studentNo=no; studentName=name; return no; } [解答]：构造方法是一个特殊的方法，主要用于初始化新创建的对象。构造方法的方法名要求与类名相同，用户不能直接调用，只能通过new运算符调用，而且构造方法是不返回任何数据类型，甚至也不返回void数据类型，即不能在构造方法前加void。 以上的代码段出错于：①构造方法Student()前不能加void ②不能用return语句 【4】定义一个表示学生的student类，包括的域有学号、姓名、性别、年龄，包括的方法有获得学号、姓名、性别、年龄及修改年龄。编写Java程序创建student类的对象及测试其方法的功能。 [解答]：程序代码段如下： class student{ private String id; private String name; private String sex; private int age; 1 配置本地端口镜像 2 1.2.1 配置任务简介 本地端口镜像的配置需要在同一台设备上进行。 首先创建一个本地镜像组，然后为该镜像组配置源端口和目的端口。 表1-1 本地端口镜像配置任务简介 ●一个端口只能加入到一个镜像组。 ●源端口不能再被用作本镜像组或其它镜像组的出端口或目的端口。 3 1.2.2 创建本地镜像组 表1-2 创建本地镜像组 配置源端口目的端口后，本地镜像组才能生效。 4 1.2.3 配置源端口 可以在系统视图下为指定镜像组配置一个或多个源端口，也可以在端口视图下将当前端口配置为指定镜像组的源端口，二者的配置效果相同。 1. 在系统视图下配置源端口 表1-3 在系统视图下配置源端口 2. 在端口视图下配置源端口 表1-4 在端口视图下配置源端口 一个镜像组内可以配置多个源端口。 5 1.2.4 配置源CPU 表1-5 配置源CPU 一个镜像组内可以配置多个源CPU。 6 1.2.5 配置目的端口 可以在系统视图下为指定镜像组配置目的端口，也可以在端口视图下将当前端口配置为指定镜像组的目的端口，二者的配置效果相同。 1. 在系统视图下配置目的端口 表1-6 在系统视图下配置目的端口 2. 在端口视图下配置目的端口 表1-7 在端口视图下配置目的端口 ●一个镜像组内只能配置一个目的端口。 ●请不要在目的端口上使能STP、MSTP和RSTP，否则会影响镜像功能的正常使 用。 ●目的端口收到的报文包括复制自源端口的报文和来自其它端口的正常转发报文。 为了保证数据监测设备只对源端口的报文进行分析，请将目的端口只用于端口镜 像，不作其它用途。 ●镜像组的目的端口不能配置为已经接入RRPP环的端口。 7 1.3 配置二层远程端口镜像 8 1.3.1 配置任务简介 二层远程端口镜像的配置需要分别在源设备和目的设备上进行。 ●一个端口只能加入到一个镜像组。 ●源端口不能再被用作本镜像组或其它镜像组的出端口或目的端口。 ●如果用户在设备上启用了GVRP（GARP VLAN Registration Protocol，GARP VLAN注册协议）功能，GVRP可能将远程镜像VLAN注册到不希望的端口上， 此时在目的端口就会收到很多不必要的报文。有关GVRP的详细介绍，请参见“配 置指导/03-接入/GVRP配置”。 《Java Web开发实战经典》期末总结 第一部分Web开发前奏 第一章Java Web开发简介 通过本章的的学习，使我了解了web的发展过程。 第二章HTML、Javascript简介 通过本章的的学习，使我了解了JavaScript的基本语法，主要事件，主要对象的作用。可以使用JavaScript完成表单的交互程序开发。 第三章XML简介 通过本章的的学习，使我了解了XML语言的基础语法，还可以清楚区分HTML和XML。并掌握了java中DOM和SAX解析XML文件的操作。学习了DOM解析XML文件的操作和DOM4J 解析XML文件的操作。使用XML可以进行数据存储结构的规范化。 第四章Tomcat服务器的安装及配置 通过本章的的学习，使我了解了Tomcat服务器的主要作用和Tomcat服务器的安装及配置。知道了Tomcat安装目录下主要目录的作用。理解JSP页面的执行流程，和编写第一个交互式程序。 第二部分Web基础开发 第五章JSP基础语法 通过本章的的学习，使我了解了Jsp中注释语句的使用和Jsp中Scriptlet的使用及使用的区别，掌握了page指令的作用和两种包含语句及两种包含语句的区别，懂得了跳转指令的操作。基本可以使用Jsp基础语法结合JDBC完成登录程序的开发。 第六章JSP内置对象 通过本章的的学习，使我了解了Jsp中的9个内置对象及对应的操作接口，掌握了jsp 中的4种属性范围及属性操作，掌握了request，response,session,application,pagecontext这些常用内置对象的使用。 第七章JavaBean 通过本章的的学习，使我了解了JavaBean的基本定义格式和web目录的标准结构，掌握了jsp中对于javabean支持的3种标签，即 以太网端口聚合+RSTP配置 拓扑图 功能要求： 通过在网络中配置RSTP功能，实现消除网络环路的目的， 当RSTP的根桥DOWN掉后，可以通过非根桥正常通信，达到根桥和备用根桥的切换，某个链路DOWN后，可以通过将某个阻塞端口恢复为根端口或转发端口，以实现正常的数据通信， 当聚合链路中的某个链路DOWN掉后，不会影响正常的通信 配置过程： S5700-LSW1 [Huawei]DIS CU # sysname Huawei # vlan batch 10 20 # stp mode rstp # cluster enable ntdp enable ndp enable # drop illegal-mac alarm # diffserv domain default # drop-profile default # aaa authentication-scheme default authorization-scheme default accounting-scheme default domain default domain default_admin local-user admin password simple admin local-user admin service-type http # interface Vlanif1 # interface MEth0/0/1 # interface GigabitEthernet0/0/1 port link-type trunk port trunk allow-pass vlan 10 20 # interface GigabitEthernet0/0/2 port link-type trunk port trunk allow-pass vlan 10 20 # interface GigabitEthernet0/0/3 port link-type access port default vlan 10 stp disable # interface GigabitEthernet0/0/4 port link-type access port default vlan 20 stp disable # interface GigabitEthernet0/0/5 # interface GigabitEthernet0/0/6 # interface GigabitEthernet0/0/7 # interface GigabitEthernet0/0/8 # interface GigabitEthernet0/0/9 品质体系框架图 图中各缩写词含义如下： QC：Quality Control 品质控制 QA：Quality Assurance 品质保证 QE：Quality Engineering 品质工程 IQC：Incoming Quality Control 来料品质控制 LQC：Line Quality Control 生产线品质控制 IPQC：In Process Quality Control 制程品质控制 FQC：Final Quality Control 最终品质控制 SQA：Source (Supplier) Quality Assurance 供应商品质控制 DCC：Document Control Center 文控中心 PQA：Process Quality Assurance 制程品质保证 FQA：Final Quality Assurance 最终品质保证 DAS：Defects Analysis System 缺陷分析系统 FA：Failure Analysis 坏品分析 CPI：Continuous Process Improvement 连续工序改善 CS：Customer Service 客户服务 TRAINNING：培训 一供应商品质保证（SQA） 1.SQA概念 SQA即供应商品质保证，识通过在供应商处设立专人进行抽样检验，并定期对供应商进行审核、评价而从最源头实施品质保证的一种方法。是以预防为主思想的体现。 2.SQA组织结构 3.主要职责 1）对从来料品质控制（IQC）/生产及其他渠道所获取的信息进行分析、综合，把结果反馈给供应商，并要求改善。 2）耕具派驻检验远提供的品质情报对供应商品质进行跟踪。 3）定期对供应商进行审核，及时发现品质隐患。 4）根据实际不定期给供应商导入先进的品质管理手法及检验手段，推动其品质保证能力的提升。 5）根据公司的生产反馈情况、派驻人员检验结果、对投宿反应速度及态度对供应商进行排序，为公司对供应商的取舍提供依据。 4.供应商品质管理的主要办法 1）派驻检验员 把IQC移至供应商，使得及早发现问题，便于供应商及时返工，降低供应商的品质成本，便于本公司快速反应，对本公司的品质保证有利。同时可以根据本公司的实际使用情况及IQC的检验情况，专门严加检查问题项目，针对性强。 2）定期审核 通过组织各方面的专家对供应商进行审核，有利于全面把握供应商的综合能力，及时发现薄弱环节并要求改善，从而从体系上保证供货品质定期排序，此结果亦为供应商进行排序提供依据。 一般审核项目包含以下几个方面 A．品质。 B．生产支持。 C．技术能力及新产品导入。 D．一般事务. 具体内容请看“供应商调查确认表”. 3）定期排序 排序的主要目的是评估供应商的品质及综合能力，以及为是否保留、更换供应商提供决策依据.排序主要依据以下几个方面的内容： A.SQA批通过率：一般要求不低于95%。 B.IQC批合格率：一般要求不低于95%。 目录 1前言 (2) 1.1 文档目的 (2) 1.2 适用对象 (2) 2性能测试目的 (2) 3性能测试所处的位置及相关人员 (3) 3.1 性能测试所处的位置及其基本流程 (3) 3.2 性能测试工作内容 (4) 3.3 性能测试涉及的人员角色 (5) 4性能测试实施规范 (5) 4.1 确定性能测试需求 (5) 4.1.1 分析应用系统，剥离出需测试的性能点 (5) 4.1.2 分析需求点制定单元测试用例 (6) 4.1.3 性能测试需求评审 (6) 4.1.4 性能测试需求归档 (6) 4.2 性能测试具体实施规范 (6) 4.2.1 性能测试起始时间 (6) 4.2.2 制定和编写性能测试计划、方案以及测试用例 (7) 4.2.3 测试环境搭建 (7) 4.2.4 验证测试环境 (8) 4.2.5 编写测试用例脚本 (8) 4.2.6 调试测试用例脚本 (8) 4.2.7 预测试 (9) 4.2.8 正式测试 (9) 4.2.9 测试数据分析 (9) 4.2.10 调整系统环境和修改程序 (10) 4.2.11 回归测试 (10) 4.2.12 测试评估报告 (10) 4.2.13 测试分析报告 (10) 5测试脚本和测试用例管理 (11) 6性能测试归档管理 (11) 7性能测试工作总结 (11) 8附录：............................................................................................. 错误！未定义书签。 1前言 1.1 文档目的 本文档的目的在于明确性能测试流程规范，以便于相关人员的使用，保证性能测试脚本的可用性和可维护性，提高测试工作的自动化程度，增加测试的可靠性、重用性和客观性。 1.2 适用对象 本文档适用于部门内测试组成员、项目相关人员、QA及高级经理阅读。 2性能测试目的 性能测试到底能做些什么，能解决哪些问题呢？系统开发人员，维护人员及测试人员在工作中都可能遇到如下的问题 1.硬件选型，我们的系统快上线了，我们应该购置什么样硬件配置的电脑作为 服务器呢？ 2.我们的系统刚上线，正处在试运行阶段，用户要求提供符合当初提出性能要 求的报告才能验收通过，我们该如何做？ 3.我们的系统已经运行了一段时间，为了保证系统在运行过程中一直能够提供 给用户良好的体验（良好的性能），我们该怎么办？ 4.明年这个系统的用户数将会大幅度增加，到时我们的系统是否还能支持这么 多的用户访问，是否通过调整软件可以实现，是增加硬件还是软件，哪种方式最有效？ 5.我们的系统存在问题，达不到预期的性能要求，这是什么原因引起的，我们 应该进行怎样的调整？ 6.在测试或者系统试点试运行阶段我们的系统一直表现得很好，但产品正式上 线后，在用户实际环境下，总是会出现这样那样莫名其妙的问题，例如系统运行一段时间后变慢，某些应用自动退出，出现应用挂死现象，导致用户对我们的产品不满意，这些问题是否能避免，提早发现？ 7.系统即将上线，应该如何部署效果会更好呢？ 并发性能测试的目的注要体现在三个方面：以真实的业务为依据，选择有代表性的、关键的业务操作设计测试案例，以评价系统的当前性能；当扩展应用程序的功能或者新的应用程序将要被部署时，负载测试会帮助确定系统是否还能够处理期望的用户负载，以预测系统的未来性能；通过模拟成百上千个用户，重复执行和运行测试，可以确认性能瓶颈并优化和调整应用，目的在于寻找到瓶颈问题。 Java开发领域的大牛有哪些 本篇文章小编和大家分享一下Java开发领域的大牛有哪些，对Java开发感兴趣或者是对Java开发大牛感兴趣的小伙伴就随小编一起来了解一下吧。 1、James Gosling 在他12岁的时候，他已能设计电子游戏机，帮忙邻居修理收割机。大学时期在天文系担任程式开发工读生，1977年获得了加拿大卡尔加里大学计算机科学学士学位，1983年获得了美国卡内基梅隆大学计算机科学博士学位，博士论文的题目是："The Algebraic Manipulation of Constraints"。毕业后到IBM工作，设计IBM第一代工作站NeWS系统，但不受重视。后来转至Sun公司。1990年，与Patrick Naughton和Mike Sheridan等人合作“绿色计划”，后来发展一套语言叫做“Oak”，后改名为Java。1994年底，James Gosling在硅谷召开的“技术、教育和设计大会”上展示Java程式。2000年，Java成为世界上最流行的电脑语言。 2009年四月，Sun被甲骨文公司并购。高斯林于2010年四月时宣布从甲骨文公司离职。 2011年3月29日，高斯林在个人博客上宣布将加入Google[3]。 2011年8月30日，仅仅加入Google数月之后的高斯林就在个人博客上宣布离开Google，加盟一家从事海洋机器人研究的创业公司Liquid Robotics[1]，担任首席软件架构师。 在2011年5月建立的Scala公司Typesafe Inc.，高斯林被聘请为公司顾问。 2、Joshua Bloch ——Collections 框架 早在1996年，适逢Java刚刚崭露头角。先是1月份发布JDK1.0，然后是5月底在旧金山召开首届JavaOne大会，年末又是JDK1.1紧跟其后。正是在Java 技术如火如荼、大展拳脚的背景之下，JoshuaBloch来到了Sun。2004 年7月初，就在J2SE5.0发布在即，Jusha Bloch刚刚荣获Sun“杰出工程师(Distinguished Engineer)”的称号之时，他突然离开Sun而去了正值发展态势迅猛的Google，成为了Google的Java首席架构师。 在Bloch来到Sun之后，很快他被从安全组调入核心平台组，从事底层API 设计。至此以后，每逢JDK的重大版本发布，总能在其中见到Joshua的“妙笔”。JDK1.1中的java.math、1.4中的assertions，还有大家所熟识的Collections Framework皆是Joshua一手打造。其中的Collections Framework还获得了当年的 目录 第1章以太网端口配置 ............................................................................................................ 1-1 1.1 以太网端口简介.................................................................................................................. 1-1 1.2以太网端口配置步骤.......................................................................................................... 1-1 1.2.1 配置以太网端口描述................................................................................................ 1-1 1.2.2 配置以太网接口状态变化上报抑制时间................................................................... 1-1 1.2.3 以太网端口专有参数配置......................................................................................... 1-2 1.3 以太网端口显示和调试....................................................................................................... 1-4 1.4 以太网端口配置示例 .......................................................................................................... 1-6 1.5 以太网端口排错.................................................................................................................. 1-7第2章以太网端口聚合配置..................................................................................................... 2-1 2.1 以太网端口聚合简介 .......................................................................................................... 2-1 2.2以太网端口聚合配置步骤 .................................................................................................. 2-1 2.3 以太网端口聚合显示和调试................................................................................................ 2-2 2.4 以太网端口聚合配置示例 ................................................................................................... 2-2 2.5 以太网端口聚合排错 .......................................................................................................... 2-3第3章以太网端口镜像配置..................................................................................................... 3-1 3.1 以太网端口镜像简介 .......................................................................................................... 3-1 3.2 以太网端口镜像配置步骤 ................................................................................................... 3-1 3.3 以太网端口镜像显示和调试................................................................................................ 3-2 3.4 以太网端口镜像配置示例 ................................................................................................... 3-2 3.5以太网端口镜像排错.......................................................................................................... 3-4 华为交换机端口镜像配置举例 配置实例 文章出处：https://www.doczj.com/doc/5811897804.html, 端口镜像是将指定端口的报文复制到镜像目的端口，镜像目的端口会接入数据监测设备，用户利用这些设备分析目的端口接收到的报文，进行网络监控和故障排除。本文介绍一个在华为交换机上通过配置端口镜像实现对数据监测的应用案例，详细的组网结构及配置步骤请查看以下内容。 某公司内部通过交换机实现各部门之间的互连，网络环境描述如下： 1）研发部通过端口Ethernet 1/0/1接入Switch C；λ 2）市场部通过端口Ethernet 1/0/2接入Switch C；λ 3）数据监测设备连接在Switch C的Ethernet 1/0/3端口上。λ 网络管理员希望通过数据监测设备对研发部和市场部收发的报文进行监控。 使用本地端口镜像功能实现该需求，在Switch C上进行如下配置： 1）端口Ethernet 1/0/1和Ethernet 1/0/2为镜像源端口；λ 2）连接数据监测设备的端口Ethernet 1/0/3为镜像目的端口。λ 配置步骤 配置Switch C： # 创建本地镜像组。 为了帮助对java编程感兴趣的同学更好、更快的提高编程技术，武汉北大青鸟光谷校区专业老师在此推荐几本学习编程非常有用的书籍，以供大家参考。 入门类 1.《java从入门到精通》（第3版） 【内容介绍】 本书从初学者角度出发，通过通俗易懂的语言、丰富多彩的实例，详细介绍了使用Java 语言进行程序开发应该掌握的各方面技术。全书共分28章，包括：初识Java，熟悉Eclipse 开发工具，Java 语言基础，流程控制，字符串，数组，类和对象，包装类，数字处理类，接口、继承与多态，类的高级特性，异常处理，Swing程序设计，集合类，I/O输入输出，反射，枚举类型与泛型，多线程，网络通信，数据库操作，Swing表格组件，Swing树组件，Swing其他高级组件，高级布局管理器，高级事件处理，AWT绘图与音频播放，打印技术和企业进销存管理系统等。所有知识都结合具体实例进行介绍，涉及的程序代码给出了详细的注释，可以使读者轻松领会Java程序开发的精髓，快速提高开发技能。 【推荐理由】 本书内容详尽，实例丰富，非常适合作为编程初学者的学习用书，也适合作为开发人员的查阅、参考资料。 2.《java核心技术卷一》 【内容介绍】 本书是Java技术经典参考书，多年畅销不衰，第7版在保留以前版本风格的基础上，涵盖Java2开发平台标准版J2SE5.0的基础知识，主要内容包括面各对象程序设计、反射与代理、接口与内部类、事件监听器模型、使用Swing UI工具箱进行图形用户界面设计，异常处理、流输入/输出和对象序列化、泛型程序设计等。本书内容翔实、深入浅出，附有大量程序实例，极具实用价值，是Java初学者和Java程序员的必备参考书。 【推荐理由】 本书出版以来，深受广大读者的喜爱，它全面介绍了Java SE的基础知识，不但具有使用价值，而且易于阅读，易理解，它必成为想进军Java和基础薄弱同学的不二之选。 3.《java编程语言》 【内容介绍】 本书以Java语言创建者的独特视角向读者展现了为什么以及如何使用Java语言，以便读者理解Java语言的基本设计目标以及它在现实世界中的应用。本书介绍的是Java语言的 配置以太网单板的内部端口 当网元通过以太网板内部端口（即VCTRUNK）将以太网业务传输到SDH侧时，需配置VCTRUNK端口的各种属性，以便配合对端网元的以太网单板，实现以太网业务在SDH网络中的传输。 前提条件 用户具有“网元操作员”及以上的网管用户权限。 已创建以太网单板。 注意事项 注意：错误的配置绑定通道，可能会导致业务中断。 操作步骤 1.在网元管理器中选择以太网单板，在功能树中选择“配置 > 以太网接口管理 > 以太 网接口”。 2.选择“内部端口”。 3.配置内部端口的TAG属性。 a.选择“TAG属性”选项卡。 b.配置内部端口的TAG属性。 c.单击“应用”。 4.配置内部端口的网络属性。 a.选择“网络属性”选项卡。 b.配置内部端口的网络属性。 图1支持QinQ功能的以太网单板的内部端口属性 图2支持MPLS功能的以太网单板的内部端口属性 c.单击“应用”。 5.配置内部端口使用的封装映射协议。 a.选择“封装/映射”选项卡。 b.配置内部端口使用的封装协议及各参数。 说明：传输线路两端的以太网单板的VCTURNK的“映射协议”和协议参数应保 持一致。 c.单击“应用”。 6.配置内部端口的LCAS功能。 a.选择“LCAS”选项卡。 b.设置“LCAS使能”以及LCAS其他参数。 说明：传输线路两端的以太网单板的VCTURNK的“LCAS使能”和LCAS协议参 数应保持一致。 c.单击“应用”。 7.设置端口的绑定通道。 a.选择“绑定通道”选项卡，单击“配置”，出现“绑定通道配置”对话框。 b.在“可配置端口”中选择VCTRUNK端口作为配置端口，在“可选绑定通道”中 选择承载层时隙。单击。 c.单击“确定”，单击“是”。出现“操作结果”对话框，提示操作成功。 ERP性能测试总结分享 1分享 (3) 1.1测试环境搭建 (3) 1.2并发量计算及场景设计 (3) 1.3测试框架搭建 (4) 1.4测试脚本开发/调试 (5) 1.5场景调试/执行 (5) 1.6性能监控分析 (6) 1.7结果报告 (7) 2展望 (8) 2.1业务调研及场景确定 (8) 2.2场景监控与分析 (8) 1分享 1.1 测试环境搭建 在我们进行性能测试之前，通常需要搭建一个供测试用的环境，使用这个环境来录制脚本，根据在这个环境下执行测试的结果，得出最终的测试结论。 有些时候，测试环境就是生产环境，例如：一个新的项目上线前进行的性能测试，通常就是在未来的生产环境下进行的。在这种情况下，可以排除测试环境与生产环境差异带来影响，测试结果相对比较准确。 反之，如果测试环境与生产环境不是同一环境，这个时候，为了保证测试结果的准确性，需要对生产环境进行调研。在搭建测试环境时，尽量保证搭建的测试环境和生成环境保持一致（环境主体框架相同，服务器硬件配置相近，数据库数据相近等）。 另外，最好输出一个测试环境搭建方案，召集各方参加评审确认。同时，在测试方案、测试报告中，对测试环境进行必要的阐述。 1.2 并发量计算及场景设计 首先，在确定场景及并发量之前，需要对业务进行调研，调研的对象最好是业务部门，也可以通过数据库中心查询数据，进行辅助。 场景选取一般包括：登陆场景、操作频繁的核心业务场景、涉及重要信息（如：资金等）业务场景、有提出明确测试需求的业务场景、组合场景等。 每个场景的并发量，需要根据业务调研的结果进行计算。可以采用并发量计算公式：C=nL / T 进行计算（其中C是平均的并发用户数，n是平均每天访问用户数，L是一天内用户从登录到退出的平均时间（操作平均时间），T是考察时间长度（一天内多长时间有用户使用系统））。 每个场景的思考时间，也可以通过业务调研获得。 另外，也可以采用模拟生产业务场景TPS（每秒通过事务数）的方式，来确定场景。相比上一种方式，模拟生产业务场景TPS，能更加准确模拟生产压力。本次ERP性能测试采用的就是这种方式：首先，通过调研确定业务高峰时段，各核心业务TPS量及产生业务单据量。然后，通过调整组合场景中，各单场景的Vusr（虚拟用户数）和Thinktime（思考时间），使每个场景的TPS接近业务调研所得到的TPS量,每个场景相同时间（即高峰时间段长度）通过事务数接近调研业务单据量，从而确定一个，可以模拟生成环境压力的基准场景。最后，通 java开发实战经典答案 【篇一：java开发实战经典】 ass=txt>首先需要为读者说明的是，《java开发实战经典》，如图1所示，不只是一本纯粹讲解基础的书，里面更多的内容是为java ee的开发而服务的，所以读者在学习本书的时候应该重点培养的是思想，按照程序解释问题的思想。在本书的序言的视频之中已经为读者介绍了本书的整体结果，实际上每个读者必须首先搞清楚java se和java ee的关系，如图2所示。 从图2中可以清楚的发现，如果要想进行java ee的开发，java se 是其中必要的组成部分，这也就是为什么在学习java ee之前要读者一定要有扎实的java se基础了。 在本书的序章中也为读者讲解过，java se核心的部分就是以下四个： 1、面向对象：主要是要求读者建立好一个面向对象的概念，并可以灵活的运用这些概念进行程序的分析，可以说在整个java se部分核心就是面向对象的开发，里面涉及的概念较多，但是所有的概念最终都是为接口和抽象类服务的，而所有java开发中涉及到的各个设计模式，实际上也是针对于接口和抽象类的应用。 2、类集框架：大部分的读者都应该听过或学习过数据结构，虽然从大学就开始学习《数据结构与算法》，但是实际上来讲，学习完后真正可以进行数据结构编写的人并不多，而且大部分的数据结构都是以c语言为基础讲解的，而类集就相当于是java实现的数据结构，包括树、链表、队列等等，如果只是进行应用层次开发的读者没有必要深入了解其内部的操作，但是必须清楚的掌握其核心的接口和各个操作类。 3、io操作：对于大部分的自学者而言，io操作也许是最难过的一个坎了，如果没有很好的理解抽象类和接口的概念，那么要想充分的理解io操作是很难的，在整个io操作之中，完整的体现了java 语言多态性的设计思想，而且在学习io的时候必须始终把握住一个原则：“根据使用的子类不同，输入输出的位置也不同”。 4、java数据库操作（jdbc）：jdbc首先并不能算是一门技术，更多的应该算是一种服务—— java提供的数据库访问服务，里面提供了大量的操作接口，各个数据库生产商根据这些接口实现自己的数据库操作，从面向对象的设计上来看，用户完全没有必要考虑其 一以太网端口配置命令 1.1．1 display interface 【命令】 display interface[ interface_type | interface_type interface_num | interface_name ] 【视图】 所有视图 【参数】 interface_type：端口类型。 interface_num：端口号。 interface_name：端口名，表示方法为interface_name=interface_type interface_num。 参数的具体说明请参见interface命令中的参数说明。 【描述】 display interface命令用来显示端口的配置信息。 在显示端口信息时，如果不指定端口类型和端口号，则显示交换机上所 有的端口信息；如果仅指定端口类型，则显示该类型端口的所有端口信 息；如果同时指定端口类型和端口号，则显示指定的端口信息。 【举例】 # 显示以太网端口Ethernet0/1的配置信息。 端口镜像典型配置举例 1.5.1 本地端口镜像配置举例 1. 组网需求 某公司内部通过交换机实现各部门之间的互连，网络环境描述如下： ●研发部通过端口GigabitEthernet 1/0/1接入Switch C； ●市场部通过端口GigabitEthernet 1/0/2接入Switch C； ●数据监测设备连接在Switch C的GigabitEthernet 1/0/3端口上。 网络管理员希望通过数据监测设备对研发部和市场部收发的报文进行监控。 使用本地端口镜像功能实现该需求，在Switch C上进行如下配置： ●端口GigabitEthernet 1/0/1和GigabitEthernet 1/0/2为镜像源端口； ●连接数据监测设备的端口GigabitEthernet 1/0/3为镜像目的端口。 2. 组网图 图1-3 配置本地端口镜像组网图 3. 配置步骤 配置Switch C： # 创建本地镜像组。Web性能测试方案
java第三章课后习题解答
h3c端口镜像配置及实例
《JAVA WEB开发实战经典》总结
以太网端口聚合+RSTP配置案例
品质体系框架图.
性能测试流程规范
Java开发领域的大牛有哪些
以太网端口
华为交换机端口镜像配置举例
JAVA编程不得不看的几本经典书籍
配置以太网单板的内部端口
金蝶ERP性能测试经验分享
java开发实战经典答案
以太网端口配置命令
端口镜像典型配置举例
相关主题
文本预览