R. G. G. Cattell and K. Skeen, Object Operations Benchmark, ACM Transactions on Database Systems, 17(3), 1992, 1-31.  @ NUS  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... pure executable code ) on the Macintosh. This is because MacOS is not a SASOS, and can therefore not ensure that a dynamic library module is always linked at the same address. 6 Performance Performance figures are shown in Table 1. The table gives execution times of a benchmark program (OO1 [CS92] run as a single process as in [HEV 98] for static and dynamic linking under Irix 6.2 and Mungi. All runs were performed on the same hardware running either Irix or Mungi in single user mode. Lazy loading (for Irix) and lazy initialisation (for Mungi) were turned off to make timings more ....

R. G. G. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17:1--31, 1992.

....results for a secure component based programming system being implemented on the Mungi OS. Section 4.1 examines the overhead of the component model by comparing method invocation and component creation costs with the raw system costs. Section 4. 2 presents results from a subset of the OO1 benchmark [4] to show that the micro benchmark results lead to end to end performance gains. Mungi benchmarks were run on a 100MHz MIPS R4600 based computer with 64MB of RAM developed at UNSW. Irix and Linux figures were obtained by [25] on a 100MHz R4600 base SGI Indy workstation. 4.1. Micro benchmarks As ....

....used by the client. CICAPs are also generated by the constructor. A simple hashing of the system clock was used by the measured component, resulting in a 1 s overhead. Obviously more sophisticated random number generators may result in higher overheads. 4.2. The OO1 Benchmark The OO1 benchmark [4] is designed to simulate typical operations on an object oriented database. The data stored in the database should be accessible to the user only via the defined interface functions, and so is a natural situation for components. A database component is constructed with appropriate access methods, ....

R. G. G. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17:1--31, 1992.

....purchasing decisions partially relying on benchmark results, and database designers measure the performance of their systems by using an appropriate benchmark. There has been muchwork in the area of database benchmarking; e.g. the Wisconsin Benchmark #Boral and DeWitt, 1984#, the OO1 Benchmark #Catell 1 and Skeen, 1992#, and the OO7 Benchmark #Carey et al. 1993#. However, there have been only a few attempts to evaluate the performance of ADBMSs, the most important of which are the BEAST Benchmark #Geppert et al. 1995#, and the ACT 1 Benchmark #Zimmermann et al. 1995#. In this paper, we describe the ....

Catell R. and Skeen J., Object Operations Benchmark, ACM Transactions on Database Systems, 17#1#:1-31, 1992.

....and can therefore not en der to avoid a proliferation of terms we will stick to calling them modules . sure that a dynamic library module is always linked at the same address. 6 Performance Performance figures are shown in Table 1. The table gives execution times of a benchmark program (OO1 [CS92] run as a single process as in [HEV 98] for static and dynamic linking under Irix 6.2 and Mungi. All runs were performed on the same hardware running either Irix or Mungi in single user mode. Lazy loading (for Irix) and lazy initialisation (for Mungi) were turned off to make timings more ....

R. G. G. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17:1--31, 1992.

....not include the validation of the capabilities that are likely to be passed explicitly by the caller. A typical application would therefore see an even greater difference in performance. OO1 As an approximation to a real life application we implemented the Object Operations ( OO1 ) benchmark [44]. OO1 simulates typical operations in a simple object oriented database system, which is an example of the class of applications which we believe to benefit most from the SASOS model. Client code invokes a database system to perform lookup, traverse and insert operations on a database. The ....

....who took the time to write several pages of very detailed and constructive comments. The project was supported by grant no A49330285 under the Australian Research Council s Large Grants scheme. OO1 Implementation Details For our benchmarks we used the small database (20,000 parts) defined in [44]. The lookup operation consists of searching 1000 random parts in the database; the database server is invoked once for each part. The insert operation creates 100 new parts in the database and connects each to 3 random parts. The total number of database server invocations is 400 in this case. ....

R. G. G. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17:1--31, 1992. 33

....bases [17, 18] 2 A lot of individuals have been transformed to general concepts because in our tests we only considered terminological knowledge but did not want to cut all the nice information about di erent wineries and wines. 3 For related work on random structures in the database area see [2, 9, 8]. 15 The generated knowledge bases have the following properties: 80 of the concepts are primitive. There are exactly 10 di erent roles. Each concept de nition is a conjunction containing one or two concept symbols (explicit super concepts) zero or one at least restrictions, ....

R. G. G. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17(1):1-31, Mar. 1992.

.... fields (OLTP and Decision Support) TPC E failed to garner enough support because being an enterprise benchmark, it was only relevant to a relatively small number of companies competing in that space [11] The Simple Database Operations Benchmark [9] Altair Complex Object Benchmark [14] OO1 [6] and OO7 [3, 2] Benchmarks were created to provide useful insight for end users evaluating the performance of Object Oriented Database Management Systems (OODBMS) Most object relational DBMS are build on top of relational database by adding the following four key features [12] inheritance, ....

R.G.G. Cattell, and J. Skeen. Object Operations Benchmark. ACM Transactions on Database Systems, 17(1): 1-31. March 1992.

....PDX (first call) 874 1104 PDX (repeated call) 10 20 22 Opal Irix Linux 133 450 160 Table 8.3: Cross domain call performance (in s) 8.2. 3 Object benchmarks In order to establish that PDX provides a good abstraction and incurs a reasonable performance penalty, a subset of the OO1 benchmark [CS92] was implemented and run on various CHAPTER 8. MUNGI PERFORMANCE 84 operating systems. OO1 is supposed to represent a benchmark that performs typical operations on an object oriented database. The database that is accessed in the benchmark should be inaccessible to the user that is performing ....

R. G. G. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17:1--31, 1992.

....fastest fine grained scheme that is comparable to a coarse grained address translation scheme; however, it still falls short in terms of performance. Based on the results presented in [Whi94] E is about 48 slower than transient C C for the hot traversals of the OO1 database benchmark [Cat91, CS92] 11 This is a fairly significant overhead considering that the overhead of our system is zero for hot traversals and much smaller (less than 5 ) otherwise. Even with generous estimates for performance improvements (say, double the performance) the costs might be reduced to only 11 The hot ....

....vs. file system I O) and precise timing requirements. Finally, we describe the hardware and operating systems used for gathering our results. 5.2. 1 Benchmarks One of the most popular object database benchmarks that has become the de facto standard is the OO1 (Object Operations One) benchmark [CS92] OO1 was also one of the first widely used database benchmarks, followed by others such the HyperModel [ABM 90] and OO7 [CDN93] benchmarks. The OO7 benchmark is designed as a successor to the OO1 benchmark, and supports some advanced data structures and complex operations over these ....

[Article contains additional citation context not shown here]

Rick G. G. Cattell and J. Skeen. Object Operations Benchmark. ACM Transactions on Database Systems, 17(1):1--31, March 1992.

....brief overview of the systems tested. Section 5 describes the benchmark s operations and discusses the experimental results for each operation as it is presented. Finally, Section 6 contains some conclusions and our plans for future work. 2 Related Work 2. 1 The OO1 Benchmark The OO1 Benchmark [CS92] 2 , commonly referred to as the Sun Benchmark, was really the first standard benchmark that attempted to predict DBMS performance for engineering design applications. Its intent was to measure performance for navigation and simple updates, and as such it was primarily intended to capture the ....

....e.g. several relationship operations in the initial benchmark were subsumed by a more general traversal operation in OO1. Because of its early visibility and its simplicity, OO1 has became a de facto standard for OODB benchmarking. As such, it has been run on many of the current OODB products [CS92] The database that the OO1 Benchmark is based upon consists of part objects and connections between them. Each part has five data fields: a part id, a type, an (x,y) coordinate pair, and a build date. Each part has exactly three out going ( to ) connections to other parts plus a variable number ....

R. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17(1), March 1992.

....messages at transaction commit for several pages and piggybacking protocol messages on data messages whenever possible. 4. 2 Performance of the Page Server To directly compare the performance of the two client server architectures for GRAS (GRAS2 and GRAS3) we implemented the the OO1 benchmark [6] for both systems. The results indicate that the performance of the pageserver system is better than that of the query server system, given server and client are separate processes. A client linked together with the query server, though, outperforms the page server by a factor of 3 for read and ....

R. G. G. Cattell and J. Skeen. Object Operations Benchmark. ACM Transactions on Database Systems, 17(1):1--31, March 1992.

....the design of a family of interesting multi user workloads. 1 Introduction The OO7 benchmarking effort is an on going research project that aims to evaluate the performance of OODBMSs. The first result of this effort, which built upon the foundation laid by earlier OODBMS benchmarking efforts [CS92, RKC87, And90, DD88] was the single user OO7 benchmark [CDN93] This benchmark differed from its predecessors in that it This work was funded by Digital Equipment Corporation. This paper initially appeared in the proceedings of OOPSLA 94. 0 attempted to provide a truly comprehensive test of ....

....contains the integer field length and the short character array type. Figure 1 depicts a composite part, its associated document object, and its associated graph of atomic parts. One way to view this is that the union of all atomic parts corresponds to the object graph in the OO1 benchmark [CS92] however, in OO7 this object graph is broken up into semantic units of locality by the composite parts. Thus, the composite parts in OO7 provide an opportunity to test how effective various OODBMS products are at supporting complex objects. 2.1.2 Assembling Complex Designs The design library, ....

R. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17(1), March 1992.

....Therefore, it is well suited to measure the performance of a software engineering database system. In contrast, many benchmarks which have been designed for relational database systems [36] do not meet this requirement. Furthermore, even the OO1 benchmark for object oriented database systems [13] uses a database which contains only one type of relationship (connecting randomly selected objects) Therefore, it is not appropriate to test the effects of different clustering algorithms with respect to different types of queries. The data model underlying the hypermodel benchmark is an ....

R.G.G. Cattell, J. Skeen. Object Operations Benchmark, ACM Transactions on Database Systems , vol. 17-1, ACM Press, pp. 1-31 (1992).

....is the first synthetic web database generator developed. It can also be used to generated a collection of web pages. In some way, a collection of web pages may look similar to the objects in an object oriented database (OODB) and there are a few performance benchmarks designed for OODBs, e.g. 001[8], 007[7] Hypermodel[1] and Sequoia 2000 [9] Nevertheless, these benchmarks do not have have a synthetic database generator that produces databases with different web schema complexities and sizes. 1.2 Objectives In the following, we present the major objectives to be accomplished by the ....

R.G.G. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17(1):1--31, 1992.

....http: www.dbis.informatik.uni frankfurt.de Abstract When the schema of an object database system is modified, the database needs to be changed in such a way that the schema and the database remain consistent with each other. This paper uses the OO1 benchmark [2], appropriately modified, to compare the two most used approaches for transforming the database, namely the immediate and the deferred database transformation [4] 1 Immediate vs. Lazy Database Updates When the schema of an object database system is modified, the database needs to be changed in ....

....a schema manager with operations for schema manipulation and for the definition of user defined conversion functions which override automatically generated default conversion functions. To evaluate the performance of the two alternative database transformations, we opted for the OO1 benchmark [2] mainly because of its simplicity. We used the OO1 lookup query to randomly access objects in the database. In addition, we slightly modified the lookup query to make sure to access more than one object in a page (e.g. scanning a subset of a class extension) We call this a cluster query. We ....

[Article contains additional citation context not shown here]

R. G. G. Cattell and J. Skeen. Object Operations Benchmark. ACM Transactions on Database Systems, 17(1):1--31, March 1992.

....instruction pointer of the current activation Send bytecodes invoke compiled or primitive methods Return bytecodes terminate execution of the current activation, and return control to the calling activation 4. 2 Benchmarks The performance evaluation draws on the OO1 object operations benchmarks [CS92] to compare the alternative execution scenarios. The operations are retrieval oriented and operate on substantial data structures, although the benchmarks themselves are simple, and so easily understood. Their execution patterns include phases of intensive computation so that memory residence is ....

R. G. G. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17(1):1--31, March 1992.

....eight query operations. One of them is an exact match query on Atomic Part Id. Two of them are date range search on Atomic Parts. None of them involve the complex conjunctive and disjunctive queries that we wish to use with ADAMS. 55 6.4. 2 Earlier OODB Benchmarks The 001 benchmark [Cat93, CS92] is intended to represent engineering applications, much like 007 which followed it. As with 007, the complex conjunctive and disjunctive queries that we wish to test are not to be found in 001. The suite of queries in 001 doesn t include range search operations either. The HyperModel benchmark ....

R. G. G. Cattell and J. Skeen. Object Operations Benchmark. ACM Transactions on Database Systems, 17(1):1--31, March 1992.

....performance of a real application, so although the benchmarks make an effort to capture the behavior of applications, there is no proof that they are an adequate substitute for a real application. However, OO7 seems superior to other published synthetic benchmarks for object databases, such as OO1 [14] and HyperModel [3] In a recent evaluation of lessons learned so far from OO7 [11] the authors do not indicate that they have received any complaints that their benchmark is particularly unrealistic or unrepresentative. 3.1 Database Structure The OO7 database is made up of a tree of Assemblies ....

....to have surrogates as well as edge marking. However, when objects are to be evicted from the cache, surrogates can be useful even if fetched objects are swizzled using edge marking. I am aware of three groups that have done swizzling studies, all using databases derived from the OO1 benchmark [14]: White and DeWitt [73] Hosking and Moss [33, 34] and Kemper and Kossmann [40] White and DeWitt compared different implementations of lazy direct swizzling (edge marking) and swizzling at the granularity of pages (ObjectStore [42] They found that swizzling on discovery is superior to ....

R. G. G. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17(1):1--31, March 1992.

....data. Since cpos were originally designed to optmise read operations, the test application emulates the behaviour of OODB operations which traverse a large n ary tree in order to find the target object. The test application is based on the navigation experiments described and used in the OO1[68] and OO7[69] benchmarks. The program defines objects which contain the following attributes (see figure 8.1) ffl Object Identifier this is the oid of the object; ffl Object data attributes two integer values are used to represent the value of the object; ffl Three references to ....

R. Cattell and J. Skeen. Object Operations Benchmark. ACM Transactions on Database Systems, 17(1), March 1992.

....so does this mapping, leading to large application programs. To retrieve an application specific object, several DBMS operations are required. This makes applications slow. Evidence to back the claims in improved performance for engineering applications have been produced by the OO1 Benchmark [Catte92], and more recently by the OO7 Benchmark [Carey93] The major features of OO1 are: Measures response time for single user access. Bill of Materials (BOM) parts and connections (one part, one relationship) Local and remote access. Three operations (look up, traversal, insert) Cold and ....

Cattell, R.G.G. & Skeen, J. Object Operations Benchmark. ACM Transactions on Database Systems. 17 (1):1-31, 1992.

No context found.

R. G. G. Cattell and K. Skeen, Object Operations Benchmark, ACM Transactions on Database Systems, 17(3), 1992, 1-31.

....those from Persistence and Ontos) Given the architecture of a typical object relational system, the performance of such tools would be more a function of a caching layer on top of the DBMS than of the DBMS itself. Also, this functionality is well tested by existing OODBMS benchmarks (e.g. OO1 [CS92] and OO7 [CDN93] Thus, BUCKY does not attempt to test navigational program performance. Finally, it should also be noted that, due to their extensible nature, object relational database systems can be customized to provide specialized solutions designed for specific application domains or ....

R. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17(1), March 1992.

No context found.

R. G. G. Cattell and J. Skeen, Object Operation Benchmark, ACM Transactions on Database Systems, 17 (1), Marsh 1992, pp 1-31

No context found.

Cattell, R.G.G., Skeen, J.: Object Operations Benchmark, ACM Transactions on Database Systems, 17 (1), 1992 (pp. 1-31)

No context found.

R. G. G. Cattell and J. Skeen. Object operations benchmark. ACM Transactions on Database Systems, 17(1):1--31, Mar. 1992.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST at NUS   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST at NUS - Copyright Penn State and NEC. Hosted by the School of Computing, National University of Singapore.