近幾年來，因為各種社交平台、網路服務的興起，對於一個擁有高擴展性、高可得性以及彈性的資料庫系統的需求日益增加。這些服務其中一個最常被提及的特性，就是其無可預測的負載流量，一個新聞或者事件都有可能造成系統負載的增減，導致系統運行不暢。所以，平衡系統各節點的附載，並且能夠動態的重新規劃節點間資料的分配，是設計一個資料庫系統時必須考量的重要課題。如何以最小的成本，在資料庫系統的節點間，動態的搬移資料以重新平衡系統附載量，便是「資料庫系統即時搬移技術」研究的主題。
許多先前的研究已經專注於如何在不影響系同正常運行的前提下，進行資料庫節點間的資料搬移。但是，我們發現在這些技術中，皆有一個共同的問題 – 資料搬移的程序和原本系統中交易的執行會無可避免地互相影響，導致系統效能與交易吞吐量的下降，以及資料搬移的時長增加。當我們在一台原本已經是高負載的節點上，進行資料搬移，將會對原本已經負載飽和的系統，造成更嚴重的影響。
在這篇論文中，我們提出了一個全新的資料庫系統即時搬移的技術。此技術基於先決式資料庫系統的設計原理，得以在資料搬移進行的過程中，讓搬移源節點與搬移目的節點同時執行相同的交易；因而使得搬移的過程中，完全不會中斷和減緩原本交易執行的速度，進而完全解決了上述先前研究中無可避免的問題。在我們的實驗中顯示，最新技術的資料搬移方法，在資料搬移開始時，也會有 65% 的吞吐量下降，而我們提出的新方法，則幾乎完全沒有對系統的延遲和吞吐量有任何負面影響，有顯著的進步。

In the last decade, the need of scalable, distributed and elastic database management systems are motivated by the increasing popularity of social media platforms and the springing up cloud services. One of the main characteristics of these applications is the unpredictable workload. Therefore, balancing the load of each instance and dynamically arranging the number of instances are important features in the design of modern cloud database management systems, and requires a low cost technique to migrate data between hosts, a feature referred to as live migration.
Many existing techniques were proposed to solve the live migration problem without shutting down the system. But a severe trade-off between the delay of transaction execution and the delay of migration can be identified in these techniques, which is unacceptable especially when the migration process usually starts in an already-hot system.
This paper presents Deterministic Crabbing, a database live migration technique for distributed database systems. Leveraging on the characteristics of deterministic transaction-processing technique, Deterministic Crabbing prevents the execution of transactions from being blocked by the migration process by executing transactions at both source and destination nodes. Our experiments show that the transaction throughput of Deterministic Crabbing was barely affected during the migration period, where the state-of-the-art approach incurs 65% performance drop.

Acknowledgments iv
Abstract v
摘要 vi
Table of Contents vii
List of Figures ix
List of Tables x
Chapter 1 Introduction 1
Chapter 2 Related Works 6
Chapter 3 The Design of Deterministic Crabbing 9
3.1. A Brief Review of Determinism 9
3.2. Initial Phase 11
3.3. Migration Phase 11
3.4. Completion Phase 12
3.5. Migration Cost Analysis 12
Chapter 4 Implementation 14
4.1. System Overview 14
4.2. Deterministic Implementation 16
4.3. Deterministic Crabbing Implementation 17
4.4. Asynchronous Pushing 18
Chapter 5 Related Work 20
5.1. Correctness 20
5.2. Recovery 21
Chapter 6 Evaluation 23
6.1. Microbenchmark 24
6.1.1. Migration time and throughput comparison 25
6.1.2. The effect of long transactions 27
6.1.3. Transaction execution time breakdown 27
6.1.4. Asynchronous Pushing Effect 29
Chapter 7 Conclusion 31
Bibliography 32

[1] 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.
[2] Alexander Thomson and Daniel J Abadi. The case for determinism in database systems. Proc. of the VLDB Endowment, 3(1-2):70-80, 2010.
[3] 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.
[4] Lamport, Leslie. "The part-time parliament." ACM Transactions on Computer Systems (TOCS) 16.2 (1998): 133-169.
[5] Junqueira, Flavio P., Benjamin C. Reed, and Marco Serafini. "Zab: High-performance broadcast for primary-backup systems." Dependable Systems & Networks (DSN), 2011 IEEE/IFIP 41st International Conference on. IEEE, 2011.
[6] Elmore, Aaron J., et al. "Zephyr: live migration in shared nothing databases for elastic cloud platforms." Proceedings of the 2011 ACM SIGMOD International Conference on Management of data. ACM, 2011.
[7] Das, Sudipto, et al. "Albatross: lightweight elasticity in shared storage databases for the cloud using live data migration." Proceedings of the VLDB Endowment 4.8 (2011): 494-505.
[8] Mishima, Takeshi, and Yasuhiro Fujiwara. "Madeus: Database Live Migration Middleware under Heavy Workloads for Cloud Environment." Proceedings of the 2015 ACM SIGMOD International Conference on Management of Data. ACM, 2015.
[9] Barker, Sean, et al. "Cut me some slack: Latency-aware live migration for databases." Proceedings of the 15th international conference on extending database technology. ACM, 2012.
[10] Elmore, Aaron J., et al. "Towards an elastic and autonomic multitenant database." Proc. of NetDB Workshop. sn, 2011.
[11] Elmore, Aaron J., et al. "Squall: Fine-Grained Live Reconfiguration for Partitioned Main Memory Databases." Proceedings of the 2015 ACM SIGMOD International Conference on Management of Data. ACM, 2015.
[12] Mohan, C., et al. "ARIES: a transaction recovery method supporting fine-granularity locking and partial rollbacks using write-ahead logging." ACM Transactions on Database Systems (TODS) 17.1 (1992): 94-162