現代的線上交易處理工作負載有兩個顯著的特徵。首先，大多數交易僅對少量的資料做操作，而且可由應用程式開發者預先定義的其執行邏輯並撰寫成資料庫中的程序。其次，存取資料的變化隨著時間的推移，可能不容易對資料做分割。第一個特徵激勵了先決式資料庫系統，相較於傳統分散式資料庫，先決式資料庫系統可避免代價高昂的二階段提交(交易同步協定)，由近年的研究指出，當資料有一個(幾乎)完美的分割方式時，先決式資料庫系統可以在同質電腦組成的叢集上有相當高的單位時間產出，同時確保副本之間的強一致性。然而，現代線上交易處理工作負載的第二個特徵使得達到非常好的資料分割是不可行的，這嚴重傷害了先決式資料庫系統的吞吐量和限制了它的適用性。在本文中，我們提出了先決式資料庫系統分派交易的概念，以應對不可預見的或難以分派的工作負載。相對於尋找好的資料分割，我們採取分派交易本身，並改變交易執行的流程來滿足當前工作負載。通過模擬交易之間的依賴關係作為交易相依性圖，我們允許每台機器在以分散式的方式來決定哪些交易應該在哪台機器上執行，並且搭配交易執行決策的優化，例如哪些交易需要複製在多台機器執行或是改變相對地執行順序，以產生更好的交易分派。我們實施一個新的交易執行引擎，稱為T-Part，並討論我們在實施過程中遇到的挑戰，以及實作上的各種優化與考量。

Chapter 1 Introduction
Chapter 2 Systems and Challenges
2.1. Deterministic Database Systems
2.2. Impacts of Imperfect Data Partitions
2.3. Limitations of Data Partitioning
Chapter 3 Overview of T-Part
3.1. T-Graph Partitioner
3.1.1. Modeling the T-graph
3.1.2. Partitioning the T-graph and Sinking the Transactions
3.2. Push Executor
3.3. Merits
Chapter 4 Implementation
4.1. Defining Weights
4.2. Partitioning T-Graph in Real-Time
4.3. T-Part Architecture and Alternative
4.4. Cache Management
4.5. Handling Failure
4.5.1. Failure Recovery of T-part
4.5.2. Failure Recovery in the Alternative Architecture
Chapter 5 Related Work
5.1. Data Partitioning
5.2. Deterministic Distributed Database
5.3. Content-Aware Request Dispatching
Chapter 6 Conclusion
Bibliography

[1] Cristiana Amza, Alan L Cox, and Willy Zwaenepoel. Conflict-aware scheduling for dynamic content applications. In USENIX Symposium on Internet Technologies and Systems, volume 21, page 22, 2003.
[2] Carlo Curino, Evan Jones, Yang Zhang, and Sam Madden. Schism: a workload-driven approach to database replication and partitioning. Proc. of the VLDB Endowment, 3(1-2):48-57, 2010.
[3] Sameh Elnikety, Steven Dropsho, and Willy Zwaenepoel. Tashkent+: Memory-aware load balancing and update filtering in replicated databases. ACM SIGOPS Operating Systems Review, 41(3):399-412, 2007.
[4] [4] George Karypis and Vipin Kumar. A fast and high quality multilevel scheme for partitioning irregular graphs. SIAM journal on Scientific Computing, 20(1):359-392, 1999.
[5] Vivek S Pai, Mohit Aron, Gaurov Banga, Michael Svendsen, Peter Druschel, Willy Zwaenepoel, and Erich Nahum. Locality-aware request distribution in cluster-based network servers. In ACM Sigplan Notices, volume 33, pages 205-216. ACM, 1998.
[6] Abdul Quamar, K Ashwin Kumar, and Amol Deshpande. Sword: scalable workload-aware data placement for transactional workloads. In Proc. of EDBT, pages 430-441. ACM, 2013.
[7] Isabelle Stanton and Gabriel Kliot. Streaming graph partitioning for large distributed graphs. In Proc. of SIGKDD, pages 1222-1230. ACM, 2012.
[8] Michael Stonebraker, Samuel Madden, Daniel J Abadi, Stavros Harizopoulos, Nabil Hachem, and Pat Helland. The end of an architectural era: (it's time for a complete rewrite). In Proc. of the VLDB Endowment, pages 1150-1160. VLDB Endowment, 2007.
[9] Aubrey L Tatarowicz, Carlo Curino, Evan PC Jones, and Sam Madden. Lookup tables: Fine-grained partitioning for distributed databases. In Proc. of ICDE, pages 102-113. IEEE, 2012.
[10] Alexander Thomson and Daniel J Abadi. The case for determinism in database systems. Proc. of the VLDB Endowment, 3(1-2):70-80, 2010.
[11] Alexander Thomson, Thaddeus Diamond, Shu-Chun Weng, Kun Ren, Philip Shao, and Daniel J Abadi. Calvin: Fast distributed transactions for partitioned database systems. In Proc. of SIGMOD, pages 1-12. ACM, 2012.
[12] Debra VanderMeer, Kaushik Dutta, and Anindya Datta. A cost-based database request distribution technique for online e-commerce applications. Management Information Systems Quarterly, 36(2):479-507, 2012.
[13] Takeshi Yamamuro, Yoshiharu Suga, Naoya Kotani, Toshio Hitaka, and Masashi Yamamuro. Buffer cache de-duplication for query dispatch in replicated databases. In Database Systems for Advanced Applications, pages 352-366. Springer, 2011.
[14] Shengqi Yang, Xifeng Yan, Bo Zong, and Arijit Khan. Towards effective partition management for large graphs. In Proc. of SIGMOD, pages 517-528. ACM, 2012.
[15] Gae-Won You, Seung-Won Hwang, and Navendu Jain. Ursa: Scalable load and power management in cloud storage systems. ACM Transactions on Storage, 9(1):1, 2013.
[16] Xiaolan Zhang, Michael Barrientos, J Bradley Chen, and Margo Seltzer. Hacc: An architecture for cluster-based web servers. In Proc. of the USENIX Windows NT Symposium, volume 3, pages 16-16. USENIX Association, 1999.
[17] Vaide Zuikeviciute and Fernando Pedone. Conflict-aware load-balancing techniques for database replication. In Proc. of ACM SAC, pages 2169-2173. ACM, 2008.