Font Size: a A A

Research On The Transaction Processing Issues Of Tenants In Multi-Tenant Environment

Posted on:2016-09-14Degree:MasterType:Thesis
Country:ChinaCandidate:T KangFull Text:PDF
GTID:2308330461984229Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
Now, the well-known cloud computing in the academic and business have made a great success, as a new computing paradigm, it continues to promote the generation of new technologies. SaaS (Software as a Service) has become an important way to provide users in cloud computing services, yet there is an important theoretical basis for achieving the application of multi-tenant SaaS. As scholars study tenants’data storage, and the data storage has been optimized, then the performance of SaaS applications has been some improvement, but there are still other issues such as optimizing related issues of the tenants’ transaction. SaaS applications will have a lot of users, and these users belong to different tenants at the same time, which doomed the high concurrency tenant matters. High availability, scalability, distribution, and so determine the characteristics of the urgent need for the tenant SaaS applications optimized transaction processing to improve application performance. But, in a distributed system environment, the theoretical of the CAP has limit, that is, for the consistency (C), the availability of (A) and fault tolerance partition (P) can only be satisfied two of the three.According to the characteristics of SaaS data customization, a variety of data storage method is proposed, which uses the basic and extended data storage table, tenant transaction processing based on the data storage and data storage, the traditional mode of transaction processing is very different. Transaction processing technology and traditional research also does not apply to the multi tenant environment, firstly, used in traditional concurrency control lock mechanism for transaction control multi tenant environment has become a bottleneck, the traditional lock mechanism to ensure that tenants data consistency is difficult, and will cause the system to low load capacity, tenant transaction scalability, and some did not lock optimization technology in a multi tenant environment research, how to improve the throughput of concurrent transactions need to study a single node multi tenant. Secondly, the combination of distributed technology of multi tenant database, especially the multi tenant data sharing mode, because the distribution of multi tenant mix and the user access to the complex as well as data storage, transaction characteristics of various tenants, keep the data consistent global also is the traditional technology cannot solve the problem.The main work is as follows:1) Put forward the design model of multilevel transaction processing. Data storage model is based on shared multi tenant data storage mode and tenant data is in the cloud and multi node distributed storage. Tenant transaction into the model with a time stamp and enter the queue scheduling model; and then to the transaction; concurrency control of tenant’s transaction execution lastly.2) Put forward the tenant oriented lightweight locking mechanism (MLLL). By increasing the count field to record the information into the memory of tenants and tenants data binding to lock together, at the same time, the tenant transaction ordering it in a multi tenant environment makes the system throughput can be improved, and meet the transaction scalability.3) Put forward the strategy to ensure data consistency by increasing the range of time-stamp to the tenant transaction. By setting the upper limit and lower limit of two time-stamp attributes to the tenant transaction, and adjust the range of concurrent execution of transactions to ensure tenants, in order to obtain high throughput at the same time, the global consistency of the tenants’ data is guaranteed.
Keywords/Search Tags:SaaS Application, Multi-Tenant, Transaction, Concurrency control, No Lock Technology
PDF Full Text Request
Related items