Rajan Suri

Locking performance in centralized databases

An analytic model is used to study the performance of dynamic locking. The analysis uses only the steady-state average values of the variables. The solution to the model is given by a cubic, which has exactly one valid root for the range of parametric values that is of interest. The models predictions agree well with simulation results for transactions that require up to twenty locks. The model separates data contention from resource contention, thus facilitating an analysis of their separate effects and their interaction. It shows that systems with a particular form of nonuniform access, or with shared locks, are equivalent to systems with uniform access and only exclusive locks. Blocking due to conflicts is found to impose an upper bound on transaction throughput; this fact leads to a rule of thumb on how much data contention should be permitted in a system. Throughput can exceed this bound if a transaction is restarted whenever it encounters a conflict, provided restart costs and resource contention are low. It can also be exceeded by making transactions predeclare their locks. Raising the multiprogramming level to increase throughput also raises the number of restarts per completion. Transactions should minimize their lock requests, because data contention is proportional to the square of the number of requests. The choice of how much data to lock at a time depends on which part of a general granularity curve the system sees.

Modelling flexible manufacturing systems using mean-value analysis

Abstract Flexible manufacturing systems (FMS) represent an important new development in automated manufacturing of parts with mid-volume demand. In the design and operation of these complex systems, it is useful to have tools that predict their performance under various conditions. This paper describes one such tool called MVAQ, a computer program based on mean-value analysis of queues. Part production rates, machine utilization and average work-in-process sizes are all easily obtained using MVAQ. The reader is advised on when MVAQ should be used for modelling FMSs and how the program can be used. A simple design example is given to illustrate the use of MVAQ. A brief tutorial on the theory behind MVAQ is also included.

Decision support requirements in flexible manufacturing

Abstract Flexible manufacturing systems are being installed by many organizations in an effort to improve productivity. Because efficient operation of these systems is such a complex task, their capacity is often underutilized. The concept of computer-based decision support systems promises to remedy the situation. This paper defines the components of a decision support system which will enable the user to get the maximum benefit from a flexible manufacturing system. The structure of this decision support system parallels the organizational activities involved in running the flexible manufacturing system. This structure should be implemented using appropriate hardware and software components.

A control perspective on recent trends in manufacturing systems

Many notions from control theory are relevant here, although their specific realization is quite different from more traditional application areas. The standard control theory techniques do not apply: we have not yet seen a manufacturing system that can be usefully represented by a linear system with quadratic objectives. This is not surprising; standard techniques have been developed for what have been standard problems. Manufacturing systems can be an important area for the future of control; new standard techniques will be developed. Some central issues in manufacturing systems include complexig, hierarchy, discipline, capacir):, uncertaing, and feedback. Important notions of control theory include state and control variables, the objective function, the dynamics or plant model, and constraints. It would be premature to try to identify these with all the issues outlined in this paper; it would even go against the purpose of the paper, which is to stimulate such modeling activity. In this section, we describe the relevance of these notions to the manufacturing context for readers whose primary background is in control and systems.

Robustness of queuing network formulas

Analyuc models of queuing networks have been observed to give good results for systems which do not conform to the assumptions of classical queuing theory This has been explained to some extent by the recently proposed concept of operational analysis However, the use of operational analysis for prediction of system performance revolves a restrictive assumption, that of homogeneous service tnnes (HST), and practical systems often deviate from HST behavior. It is shown that the main system performance measures are surpnsmgly insensitive to violations of the HST assumption. This further explains the robustness of analyacal models for predicting performance of queuing networks Some issues regarding the operational versus stochasuc approaches are resolved, since it is clatmed that the present analysis, which is carried out m the operational framework, cannot be meaningfully carried out m the stochastic framework. Categones and Subject Descnptors: D.4.8 [Operating Systems] Performance--modeling andpredtctwn; operatwnal analysis;, stochastic analysts General TermsExperimentation, Measurement, Performance Additional

A mean value performance model for locking in databases: the no-waiting case

A new performance model for dynamic locking is proposed. It is based on a flow diagram and uses only the steady state average values of the variables. It is general enough to handle nonuniform access, shared locks, static locking, multiple transaction classes, and transactions of indeterminate length. The analysis is restricted to the case in which all conflicts are resolved by restarts. It has been shown elsewhere that, under certain conditions, this pure restart policy is as good as, if not better than, a policy that uses both blocking and restarts. The analysis is straightforward, and the computational complexity of the solution, given some nonrestrictive approximations, does not depend on the input parameters. The solution is also well defined and well behaved. The models predictions agree well with simulation results. The model shows that data contention can cause the throughput to thrash, and gives a limit on the workload that will prevent this. It also shows that systems with a particular kind of nonuniform access and systems in which transactions share locks are equivalent to systems in which there is uniform access and only exclusive locking. Static locking has higher throughput, but longer response time, than dynamic locking. Replacing updates by queries in a multiprogramming mix may degrade performance if the queries are longer than the updates.

Time-optimal control of parts-routing in a manufacturing system with failure-prone machines

We consider the time-optimal control of a dynamic system with jump parameters. The motivating example is that of a parts-manufacturing system in which machines fail and are reputed according to known Markov processes. It is desired to obtain a feedback control of parts-routing to machines, which minimizes the expected completion time of a given production target. Using a Dynamic Programming approach, we derive optimality conditions. These are used for solution of a simple example. It is seen that closed form solutions would be very hard to obtain for large problems, so alternative approaches are also discussed.

Optimal multiprogramming

SummaryThree heuristics for controlling the multiprogramming load to maximize system work capacity are studied. Each allows the highest load possible subject to a given constraint. The knee criterion constrains the memory policy so that program resident sets average near the knees of their inter page fault lifetime curves. The L=S criterion constrains the memory policy or load so that the system inter page fault lifetime L is at least as large as page swap time S. The 50 % criterion constrains the load so that the paging device is busy approximately half the time. Numerical evaluations of queueing networks show that the knee criterion, which is the most difficult to implement, is the most robust, while the easily implemented 50 % criterion is the least robust. These evaluations also circumscribe the conditions under which the criteria are expected to work reliably. Examples from practical systems further validate the criteria. Stability problems are examined.

A Concept of Monotonicity and Its Characterization for Closed Queueing Networks

We define a qualitative property, called monotonicity, for a closed queueing network. This property is defined both for a station in the network and for the network as a whole. Several results in this paper relate monotonic behavior to the known parameters of a network and allow us to state a priori, for many practical systems, whether the monotonicity property will apply for a network or a station. Monotonicity thus ensures that the system is well-behaved with respect to a number of parameters. We present diverse examples that demonstrate how knowledge of this property can be useful in network design and performance evaluation.

Infinitesimal perturbation analysis of discrete event dynamic systems: A general theory

We give a rigorous extension of the perturbation analysis approach to more general Discrete Event Dynamic Systems (DEDS). For a fairly general class of DEDS, and for certain parameters and performance measures of such DEDS, an infinitesimal perturbation analysis algorithm is stated. It is then proved that this algorithm gives exact values for the gradients of performance w.r.t. the parameters, by observing only one sample-path of the DEDS. This enables very efficient calculation of these gradients: a fact that can be used for design/operation of manufacturing systems, communication networks, and many other real-world systems.