Errata

Sentence:
"Values belonging to the same row are stored closely together"
should be:
"Values belonging to the same column are stored closely together"

Note from the Author or Editor:
Column-Oriented Data Layout, p. 10. “Values belonging to the same row are stored closely together” should read as “Values belonging to the same column are stored closely together:”

[MOLINA08] is referenced, but is not in the list of sources at the end of the book.

Note from the Author or Editor:
Some of the references appear to be not sorted alphabetically. Even though all of them are present, they are hard to find in a printed version. All references that were present in the wrong order are added below in alphabetical order:

[ALHOUMAILY10] Yousef J. Al-Houmaily. 2010. Atomic commit protocols, their integration, and their optimisations in distributed database systems. Int. J. Intell. Inf. Database Syst. 4, 4 (September 2010), 373-412.

[BAUDET19] "State Machine Replication in the Libra Blockchain". Mathieu Baudet, Avery Ching, Andrey Chursin, George Danezis, François Garillot, Zekun Li, Dahlia Malkhi, Oded Naor, Dmitri Perelman, Alberto Sonnino. 2019

[BUCHMAN18] "The latest gossip on BFT consensus". Ethan Buchman, Jae Kwon, Zarko Milosevic. 2018.

[CESATI05] "Understanding the Linux Kernel, 3rd Edition". Marco Cesati, Daniel P. Bovet. 2005.

[CLOCK06] "Implementing a Better Cache Replacement Algorithm in Apache Derby". Gokul Soundararajan. 2006.

[CORMODE11] Cormode, Graham and S. Muthukrishnan. “Approximating Data with the Count-Min Data Structure.” (2011).

[DECHEV10] Damian Dechev, Peter Pirkelbauer, and Bjarne Stroustrup. 2010. Understanding and Effectively Preventing the ABA Problem in Descriptor-Based Lock-Free Designs. In Proceedings of the 2010 13th IEEE International Symposium on Objec/Component/Service-Oriented Real-Time Distributed Computing (ISORC '10). IEEE Computer Society, Washington, DC, USA, 185-192.

[DOWNEY12] "Be Careful with Sloppy Quorums". Jim Downey. 2012.

[FOWLER11] "The LMAX Architecture". Martin Fowler. 2011

[FREILING11] Felix C. Freiling, Rachid Guerraoui, and Petr Kuznetsov. 2011. The failure detector abstraction. ACM Comput. Surv. 43, 2, Article 9 (February 2011), 40 pages.

[GIAMPAOLO98] Dominic Giampaolo. 1998. Practical File System Design with the be File System (1st ed.). Morgan Kaufmann Publishers Inc., San Francisco, CA, USA.

[GILAD17] Yossi Gilad, Rotem Hemo, Silvio Micali, Georgios Vlachos, and Nickolai Zeldovich. 2017. Algorand: Scaling Byzantine Agreements for Cryptocurrencies. In Proceedings of the 26th Symposium on Operating Systems Principles, Shanghai , China, October 28-31, 2017. 51–68.

[GRAY05] Gray, Jim; van Ingen, Catharine (December 2005). "Empirical Measurements of Disk Failure Rates and Error Rates" (PDF). Microsoft Research Technical Report MSR-TR-2005-166. Retrieved 4 March 2013.

[HYPARVIEW07] "HyParView: a membership protocol for reliable gossip-based broadcast". Joao Leitao, Jose Pereira, and Luis Rodrigues. 2007.

[KERRISK10] "The Linux Programming Interface". Michael Kerrisk. 2010. No Starch Press.

[KINGSBURY18_2] "Strong consistency models". Kyle Kingsbury. 2018.

[KLEPPMANN15] "Please stop calling databases CP or AP". Martin Kleppmann. 2015.

[KOOPMAN15] "Selection of Cyclic Redundancy Code and Checksum Algorithms to Ensure Critical Data Integrity". Philip Koopman, Kevin R. Driscoll, Brendan Hall. 2015.

[KORDAFSHARI05] M. S. Kordafshari, M. Gholipour, M. Mosakhani, A. T. Haghighat, and M. Dehghan. 2005. Modified bully election algorithm in distributed systems. 2005

[MERKLE87] Ralph C. Merkle. 1987. A Digital Signature Based on a Conventional Encryption Function. In A Conference on the Theory and Applications of Cryptographic Techniques on Advances in Cryptology (CRYPTO '87), Carl Pomerance (Ed.). Springer-Verlag, London, UK, UK, 369-378.

[MOLINA92] H. Garcia-Molina and K. Salem. 1992. Main Memory Database Systems: An Overview. IEEE Trans. on Knowl. and Data Eng. 4, 6 (December 1992), 509-516.

[NICHOLS66] The Past Participle of "Overflow": "Overflowed" or "Overflown". Ann Eljenholm Nichols. American Speech Vol. 41, No. 1 (Feb., 1966), pp. 52-55

[STONE98] "Performance of checksums and CRCs over real data". J. Stone, M. Greenwald, C. Partridge and J. Hughes, in IEEE/ACM Transactions on Networking, vol. 6, no. 5, pp. 529-543, Oct. 1998.

[SYSTEMR81] Donald D. Chamberlin, Morton M. Astrahan, Michael W. Blasgen, James N. Gray, W. Frank King, Bruce G. Lindsay, Raymond Lorie, James W. Mehl, Thomas G. Price, Franco Putzolu, Patricia Griffiths Selinger, Mario Schkolnick, Donald R. Slutz, Irving L. Traiger, Bradford W. Wade, and Robert A. Yost. 1981. A history and evaluation of System R. Commun. ACM 24, 10 (October 1981), 632-646.

[VINOSKI08] S. Vinoski, "Convenience Over Correctness" in IEEE Internet Computing, vol. 12, no. 04, pp. 89-92, 2008.

[ZHAO15] "Fast Paxos Made Easy: Theory and Implementation". Wenbing Zhao. 2015.

Figure 1-6 (b) Use of Primary Index as an indirection layer by Secondary Index is inconsistent with Secondary Index Offsets in Figure 1-6 (a).

Assuming that both primary and secondary indices are zero-based, in Figure 1-6 (b) Entry 1 of secondary index points to Entry 0 of primary index. This is inconsistent with Figure 1-6 (a), where Entry 1 of secondary index points to Data Record 3, which maps back to Entry 3 of primary index.

Similarly, in Figure 1-6 (b) Entry 2 of secondary index points to Entry 3 of primary index. This is inconsistent with Figure 1-6 (a), where Entry 2 of secondary index points to Data Record 0, which maps back to Entry 0 of primary index.

Thus, it appears that in Figure 1-6 (b), primary index indirection for secondary index entries 1 and 2 have been interchanged.

Note from the Author or Editor:
Assuming primary and secondary index nodes in example b) are 0-indexed, node 1 of secondary index should point to node 3 of primary index, and node 2 of secondary index should point to node 0 of primary index. In other words, entries 1 and 2 in a secondary index were interchanged.

In B-Tree node splitting, the split point should only be put into the newly created sibling node for leaf node splits, not nonleaf node splits, yet this paragraph says "All elements after the split point (including split point in the case of nonleaf node split) are transferred to the newly created sibling node, ..." I believe that this should be changed to "All elements after the split point (including the split point in the case of leaf node split) are transferred to the newly created sibling node, ..." Namely, "nonleaf node split" --> "leaf node split".

Note from the Author or Editor:
B-Tree Node Splits, p. 39. Sentence on elements that should be transferred during split contains a typo, and should read as (correction is highlighted): “All elements after the split point (including split point in the case of _leaf node_ split) are transferred to the newly created sibling node.”

In Figure 2-12, the key 10 is a separation key in both the parent node and the depicted child node. The key range that corresponds to the child node is k >= 10. However, there is a child node pointer (the leftmost pointer in the child node) whose key range is k < 10, which would violate the k >= 10 constraint defined by the parent node for this subtree. The subtree pointed to by the leftmost pointer in the child node would necessarily have to be empty, which wouldn't be a well-defined state for a B-Tree.

It may be more well-defined (or at least less confusing) to change the separation key 10 in the parent to 8. That way, the child could still have a separation key 10 with a corresponding k < 10 pointer to a nonempty subtree (could contain the key 9) without violating any key range constraints defined by the parent node's separation keys.

Note from the Author or Editor:
This is right: during the child split, midpoint is promoted, so having the same item repeated twice on two nonleaf levels is indeed incorrect.

In step c of figure 13-9, 'Primary at Account1' should not appear here

Note from the Author or Editor:
Distributed Transactions with Percolator, p. 274, Figure 13-9. In section c) of the figure, “Primary at Account 1” should not be visible.

Alternatively, under assumption that an immutable store is used, “Primary” mark should be added to TS3 under Account 1. Both implementations would be correct given the resolution algorithm.