{"title":"中间件的四分部件","authors":"Ashish Singhai, A. Sane, R. Campbell","doi":"10.1109/ICDCS.1998.679502","DOIUrl":null,"url":null,"abstract":"We make two observations about communications middleware: first, most middleware are similar, the differences are in their interfaces and optimizations; second, neither a fixed set of abstractions nor a fixed implementation of a set of abstractions is likely to be sufficient and well-performing for all applications. Based on these observations, we present Quarterware, a customizable middleware architecture. It abstracts basic middleware functionality, and admits application specific specializations and extensions. We demonstrate its flexibility by deriving implementations for core facilities of CORBA, RMI, and MPI. The performance results show that the derived implementations equal or exceed the performance of corresponding native versions. These results suggest that customizing middleware on a per-application basis is an effective approach for building robust, high-performance applications.","PeriodicalId":289230,"journal":{"name":"Proceedings. 18th International Conference on Distributed Computing Systems (Cat. No.98CB36183)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1998-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"75","resultStr":"{\"title\":\"Quarterware for middleware\",\"authors\":\"Ashish Singhai, A. Sane, R. Campbell\",\"doi\":\"10.1109/ICDCS.1998.679502\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"We make two observations about communications middleware: first, most middleware are similar, the differences are in their interfaces and optimizations; second, neither a fixed set of abstractions nor a fixed implementation of a set of abstractions is likely to be sufficient and well-performing for all applications. Based on these observations, we present Quarterware, a customizable middleware architecture. It abstracts basic middleware functionality, and admits application specific specializations and extensions. We demonstrate its flexibility by deriving implementations for core facilities of CORBA, RMI, and MPI. The performance results show that the derived implementations equal or exceed the performance of corresponding native versions. These results suggest that customizing middleware on a per-application basis is an effective approach for building robust, high-performance applications.\",\"PeriodicalId\":289230,\"journal\":{\"name\":\"Proceedings. 18th International Conference on Distributed Computing Systems (Cat. No.98CB36183)\",\"volume\":\"1 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"1998-05-26\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"75\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Proceedings. 18th International Conference on Distributed Computing Systems (Cat. No.98CB36183)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/ICDCS.1998.679502\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings. 18th International Conference on Distributed Computing Systems (Cat. No.98CB36183)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICDCS.1998.679502","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
We make two observations about communications middleware: first, most middleware are similar, the differences are in their interfaces and optimizations; second, neither a fixed set of abstractions nor a fixed implementation of a set of abstractions is likely to be sufficient and well-performing for all applications. Based on these observations, we present Quarterware, a customizable middleware architecture. It abstracts basic middleware functionality, and admits application specific specializations and extensions. We demonstrate its flexibility by deriving implementations for core facilities of CORBA, RMI, and MPI. The performance results show that the derived implementations equal or exceed the performance of corresponding native versions. These results suggest that customizing middleware on a per-application basis is an effective approach for building robust, high-performance applications.