Advanced Databases Flashcards

Question

What is the role of the System Catalogue in a DBMS?

Answer 1

Contains metadata about stored data & schemas ## Footnote Metadata includes details like file names, sizes, storage details, and constraints.

Answer 2

* Names & sizes of files * Storage details of files * Names & data types of data items * Mappings between schemas * Constraints * Statistical information ## Footnote This information is crucial for the management and optimization of the database.

Answer 3

Handles schema definitions by processing and translating them into a format the DBMS can understand ## Footnote It also stores schema descriptions in the system catalogue.

Answer 4

* Parses & validates queries * Compiles queries to an internal form called 'query plan' ## Footnote This ensures that queries are correctly formatted before execution.

Answer 5

Improves the query plan for efficient execution ## Footnote It optimizes the sequence of operations and eliminates redundant steps.

Answer 6

False ## Footnote The Query Optimiser generates executable code based on the optimised query plan without direct interaction with physical storage.

Answer 7

A tool that processes embedded SQL statements in a host programming language (like C, Java, or COBOL) before the actual compilation of the program ## Footnote It extracts DML commands and sends them to the DML compiler.

Answer 8

Processes DML commands and compiles them into executable code ## Footnote This conversion allows commands to be executed by the runtime database processor.

Answer 9

Executes queries and manages access to data stored in a database during program execution ## Footnote It interprets high-level database operations (like SQL queries), optimizes them, and coordinates the retrieval or modification of data by interacting with storage and memory subsystems in real time.

Answer 10

Manages the physical storage of data & controls access to it ## Footnote It interacts with the physical storage using basic OS services.

Answer 11

* Loading utility * Backup utility * Recovery utility * File reorganization utility * Performance monitoring ## Footnote These utilities assist with data management and system performance.

Answer 12

Recovery ## Footnote This utility is crucial for maintaining data integrity after failures.

Answer 13

Volatile ## Footnote Cache is temporary storage that loses its contents when power is turned off.

Answer 14

Very fast ## Footnote Cache operates at speeds that are significantly faster than main memory and secondary storage.

Answer 15

Very expensive ## Footnote The high performance of Cache comes at a high cost.

Answer 16

10^9 - 10^10 bytes ## Footnote This indicates the typical storage capacity of RAM.

Answer 17

10^-8 s (20-30 cycles) ## Footnote This is the time it takes to access data in Main Memory.

Answer 18

Non-volatile ## Footnote Secondary storage retains data even when the power is off.

Answer 19

Slow ## Footnote Compared to Cache and Main Memory, Secondary Storage is much slower.

Answer 20

10^11 - 10^12 bytes ## Footnote This indicates the typical storage capacity of hard drives.

Answer 21

10^-3 s (10^6 cycles) ## Footnote This reflects the time it takes to access data on hard disk drives.

Answer 22

Non-volatile ## Footnote Tertiary storage also retains data without power, like Secondary Storage.

Answer 23

Very Slow ## Footnote Tertiary storage is the slowest type of storage compared to Cache and Main Memory.

Answer 24

10^13 - 10^17 bytes ## Footnote This indicates a very large storage capacity typical of archival systems.

Answer 25

10^1 - 10^2 s ## Footnote This shows the long wait times typically associated with Tertiary Storage.

Answer 26

Secondary storage medium for databases ## Footnote HDDs serve as the main storage solution for many database applications.

Answer 27

Data stored in concentric circles ## Footnote Each track can contain multiple sectors.

Answer 28

The same track on all surfaces of the disk ## Footnote This concept helps in understanding how data is organized across multiple platters.

Answer 29

Tracks closer to the disc edge are longer than those closer to the axis ## Footnote This technique improves data storage density.

Answer 30

* Gap * Sync * Address mark * ECC ## Footnote These components help in organizing and error-checking data on the disk.

Answer 31

Time taken to access data on a hard disk

Answer 32

For writing, must verify the data

Answer 33

Process * Read block * Modify in memory * Write block * Verify block

Answer 34

Time taken for head assembly to move to a given track ## Footnote Seek time is a significant factor in overall disk access time.

Answer 35

Time taken for the desired sector to rotate under the read/write head ## Footnote This is a component of the total access time for HDDs.

Answer 36

Time to transfer requested data from the disk to memory ## Footnote Transfer time is calculated as block size divided by transfer rate.

Answer 37

Reading data blocks in a contiguous manner ## Footnote This method minimizes seek time and rotational delay.

Answer 38

How data blocks are located & accessed on storage devices ## Footnote It is essential for efficient data retrieval.

Answer 39

* Cylinder-head-sector * Logical Block Addressing ## Footnote Each method has its advantages and disadvantages in data retrieval.

Answer 40

* Large blocks reduce access costs but increase irrelevant data * Small blocks increase access costs but reduce irrelevant data ## Footnote Choosing the right block size is crucial for optimizing performance.

Answer 41

SSD doesn't have moving parts ## Footnote This fundamental difference leads to various performance characteristics.

Answer 42

More expensive than HDD ## Footnote The technology and performance of SSDs contribute to their higher cost.

Answer 43

Higher I/O performance ## Footnote SSDs typically offer faster read and write speeds compared to HDDs.

Answer 44

Writes are slower than reads ## Footnote This is a common characteristic of SSD technology.

Answer 45

Deals with how data is moved between secondary storage (HDD/SSD) and main memory

Answer 46

Main memory is much smaller than secondary storage, but faster, so managing data improves performance

Answer 47

A region of memory used by a database system to cache frequently accessed data pages, reducing the need to read from disk and improving performance.

Answer 48

The size of a database block plus some metadata

Answer 49

How many users are currently accessing the block in that frame

Answer 50

Set to 1 if the copy of the block in the buffer has been changed and not written back to disk; 0 otherwise

Answer 51

Used for LRU replacement (OPTIONAL)

Answer 52

Used for FIFO replacement (OPTIONAL)

Answer 53

Used for Clock replacement (OPTIONAL)

Answer 54

The system checks if it's already in the buffer pool

Answer 55

The block is read into the empty frame and pin count is set to 1

Answer 56

A frame is selected for replacement using a replacement strategy

Answer 57

Its pin count must be 0

Answer 58

Selects the frame with the oldest access time for replacement

Answer 59

Selects the frame with the oldest loading time for replacement

Answer 60

An approximation of LRU where each buffer is checked in turn and marked for replacement if it hasn't been accessed in a full cycle

Answer 61

Used to read in one block, process it, read another block into the same buffer, causing potential delays

Answer 62

Single buffer time = n(P + R)

Answer 63

* P = time to process a block * R = time to read a block * n = number of blocks

Answer 64

Uses a pair of buffers to optimize data processing and reading

Answer 65

While reading a block into buffer A, process block from buffer B; after reading into A, process A and read next block into B

Answer 66

Double buffer time = R + nP

Answer 67

* P = time to process a block * R = time to read a block * n = number of blocks

Answer 68

Data referenced every five minutes should be memory resident (stored in main memory)

Answer 69

Between the cost of keeping a block on disk and the cost of keeping it in RAM

Answer 70

Jim Gray & Franco Putzolu in 1987

Answer 71

Occurs when the cost of keeping a block on disk equals the cost of keeping it in RAM

Answer 72

X = ($D x P) / (I x $M) ## Footnote X = time interval in seconds $D = cost of a disk unit P = number of pages in 1MB of RAM I = number of I/Os that a disk unit can perform per second $M = cost of 1MB of RAM

Answer 73

* X = time interval in seconds * $D = cost of a disk unit * P = number of pages in 1MB of RAM * I = number of I/Os that a disk unit can perform per second * $M = cost of 1MB of RAM

Answer 74

Changes in technology & cost: 1997: 5 minutes; 2007: 1.5 hours; 2016: 13.5 minutes; Current: DRAM-SSD: 5 minutes, SSD-HDD: 1.5 days

Answer 75

Energy costs of DRAM are much greater than SSDs, and HDDs have greater energy costs than both

Answer 76

The basic units of information that a DBMS stores

Answer 77

Using short integer of 2 bytes

Answer 78

IEEE 754 floating point standard

Answer 79

* ASCII * UTF-8

Answer 80

Using 1 byte per value; multiple values can be packed per byte

Answer 81

As days since a given date or using ISO8601 formats such as YYYYMMDD or YYYYDDD

Answer 82

As seconds since midnight or in ISO8601 formats such as HHMMSS or HHMMSSFF

Answer 83

In different ways

Answer 84

Fixed length: integers & characters; Variable length: strings & bit arrays

Answer 85

Collections of related data items, aka fields

Answer 86

An employee record might consist of a name field, a salary field, and an employment start date field

Answer 87

* Fixed length * Variable length

Answer 88

An employee record with a 2-byte integer employee number, a 10-character name, and a 2-byte department code

Answer 89

An employee record containing a number of fields, a code identifying the field (e.g. as an employee number), a type (e.g. integer), a string length and value, and a code identifying another field (e.g. name)

Answer 90

The data at the beginning of a record that provides descriptive information about the record

Answer 91

* File ID (DB ID, or Relation ID) * Block ID * Record directory * Pointer to free space * Type of block * Pointer to other blocks * Timestamps

Answer 92

* Separating records in a block * Spanned vs unspanned * Sequencing * Indirection

Answer 93

* Use fixed length records * Use a special marker to indicate record end * Give record lengths (or offsets)

Answer 94

* Unspanned: each record must fit within a single block * Spanned: records may be split between blocks

Answer 95

Ordering records in file (and block) by some key value to optimise disk I/O

Answer 96

Record ID directly specifies the physical location of the record

Answer 97

* Quick access to memory/disk location * Low overhead due to no extra data structures or lookups

Answer 98

* Inflexible if record needs to be moved * Extensive updates required for address changes

Answer 99

Uses pointers to refer to records

Answer 100

* Easier to move & rearrange records * More efficient insertion & deletion operations

Answer 101

* Extra overhead due to pointers * Requires extra step to dereference pointers

Answer 102

Handling how data is located & accessed in secondary storage & main memory

Answer 103

Records the mapping between DB addresses & memory addresses

Answer 104

Technique that translates DB addresses to virtual memory addresses

Answer 105

* Automatic * On-demand swizzling * No swizzling

Answer 106

The reverse operation of swizzling

Answer 107

* Immediate Space Reclaim * Marking as Deleted

Answer 108

Handling dangling pointers

Answer 109

Leaves a MARK in the map or old location; physical space is never reused

Answer 110

* Physical tombstones * Logical tombstones

Answer 111

The way data is stored, indexed, and retrieved efficiently from the database.

Answer 112

Improves query performance & enables efficient data retrieval.

Answer 113

Tuples (records) that match specific criteria.

Answer 114

In files, which are collections of blocks.

Answer 115

Records corresponding to tuples in the relation.

Answer 116

Tuples of a relation sorted by their primary key.

Answer 117

Tuples are distributed among blocks in the order of their primary key.

Answer 118

To allow for later insertions.

Answer 119

Query performance vs update cost

Answer 120

A sequence of blocks holding keys & pointers to all records in the data file.

Answer 121

One key/pointer pair for every record in the data file.

Answer 122

Fewer disk accesses need to be made.

Answer 123

Binary search for keys as they are sorted.

Answer 124

One key/pointer pair for every block in the data file. Uses sequential search with the block to find specific record.

Answer 125

The data file must be sorted by search key.

Answer 126

When precise, quick access to individual records is needed.

Answer 127

Uses less space (and better for insertions). Also good for low cardinality datasets (duplicate values).

Answer 128

A sparse index of the first index.

Answer 129

They can be smaller than record pointers.

Answer 130

Pair for each key/pointer & record. Just point at first instance of key.

Answer 131

Index contains first key from each block. Index contains first new key from each block (may mean block gets excluded).

Answer 132

Update the entry in the index after deleting the record.

Answer 133

Unlike primary indexes, it doesn't determine placement of records in the data file.

Answer 134

Record pointers, not block pointers.

Answer 135

Inserts are expensive, and/or lose sequentiality & balance.

Answer 136

Offset = (Key Position Difference) x Block Size.

Answer 137

Consistent performance ## Footnote B+ trees maintain predictable performance regardless of the data distribution.

Answer 138

Sequentiality ## Footnote B+ trees may not maintain the sequential order of data access, but they provide faster search times.

Answer 139

Main memory ## Footnote Keeping the root node in memory allows for quick access to the tree structure.

Answer 140

Contain actual data records stored in the order of the primary key ## Footnote This allows for efficient data retrieval without a sequential scan.

Answer 141

Pointers to the data records ## Footnote This allows access to data in the sequence of the secondary key.

Answer 142

Fixed size ## Footnote Fixed size helps in maintaining a balanced structure and efficient space usage.

Answer 143

To ensure nodes aren't too empty ## Footnote This helps in efficient space utilization.

Answer 144

All leaves are the same distance from the root ## Footnote This characteristic maintains the balance of the tree.

Answer 145

lp = ⌊(b - p)/r⌋ ## Footnote Where b is the block size, p is the block pointer size, and r is the record length.

Answer 146

Around 0.6 ## Footnote This allows for some level of expansion in the tree structure.

Answer 147

ls = ⌊(b - p)/(k + p)⌋ ## Footnote Where b is the block size, p is the block pointer size, and k is the key length.

Answer 148

Space available in leaf (insert 32) ## Footnote This is a straightforward insertion where no restructuring is needed.

Answer 149

Leaf overflow (insert 7) ## Footnote This requires splitting the leaf node and redistributing keys.

Answer 150

Non-leaf overflow (insert 160)

Answer 151

Coalesce with sibling ## Footnote This involves merging with a neighboring node when a node underflows.

Answer 152

New root (insert 45)

Answer 153

Coalesce with sibling ## Footnote This involves merging with a neighboring node when a node underflows.

Answer 154

Redistribution ## Footnote Borrowing a key from a sibling is simpler than merging nodes.

Answer 155

* Better performance * Consume more space * Harder concurrency control ## Footnote These features make B+ trees suitable for certain types of database applications.

Answer 156

* When to reorganize * How full to load pages of new index ## Footnote These uncertainties can lead to inefficient indexing practices.

Answer 157

A function that takes a key as input and computes an integer value to determine appropriate storage location ## Footnote This integer value is used to index into a hash table.

Answer 158

Storage locations that hold records; they point to linked lists of records ## Footnote Keys are sorted in buckets to improve efficiency.

Answer 159

Instances where different keys hash to the same storage location due to more possible keys than available buckets

Answer 160

A hash table where data is stored entirely in memory ## Footnote The hash function computes an integer value from a key to determine a bucket.

Answer 161

A hash table where data is stored on disk, allowing for larger datasets ## Footnote Requires disk I/O to access the bucket array.

Answer 162

1. **Direct Mapping**: hash function gives location of bucket on disk, making access fast and predictable 2. **Use a Directory**: hash function gives index into a directory, which points to the actual disk blocks - more flexible

Answer 163

Between 50% and 80% ## Footnote Lower than 50% indicates wasted space, while higher than 80% indicates significant overflows.

Answer 164

1. Extensible hashing 2. Linear hashing

Answer 165

A method that adapts to database growth by adjusting the directory size and splitting buckets as needed

Answer 166

1. Using a variable number of bits from the hash function's output 2. Using a directory (array of pointers to buckets)

Answer 167

Blocks can be merged and the directory can be reduced if possible, without merging blocks directly

Answer 168

Structures used when inserting records to handle cases when buckets are full

Answer 169

1. Handles growing files without excessive wasted space 2. Avoids reorganizations, reducing overhead

Answer 170

1. Directory introduces indirection, leading to extra memory access 2. Directory size doubles occasionally, increasing disk accesses and decreasing performance

Answer 171

A method that improves on extensible hashing by incrementally expanding the hash file

Answer 172

The algorithm looks at a certain number of the smallest bits from the hash result, called i, to decide where to start storing the data.

Answer 173

If h(k)[i] <= m, look at bucket h(k)[i]; otherwise, look at bucket h(k)[i] - 2^(i-1)

Answer 174

When the utilization U exceeds a certain threshold ## Footnote U = #used slots / total #slots

Answer 175

1. Can handle growing files with less wasted space 2. No full reorganizations required 3. No indirection unlike extensible hashing

Answer 176

Still has overflow chains

Answer 177

Optimal for single-value lookups as it provides direct access to data based on a key

Answer 178

Hashing does not preserve the order of data

Answer 179

B-trees store data in sorted order, allowing for efficient range retrievals

Answer 180

Simple and good for scans as the index is a sequential file; inserts can be expensive

Answer 181

To efficiently retrieve data based on multiple search keys.

Answer 182

* Partial match queries * Range queries * Nearest-neighbour queries

Answer 183

Get all matching records using an index on one attribute and check values of the other attribute.

Answer 184

Requires scanning twice.

Answer 185

Use secondary indexes on each attribute to get two sets of record pointers and take the intersection of sets.

Answer 186

The comparison could be quite expensive.

Answer 187

Use a secondary index on one attribute to select a suitable index on the other attribute.

Answer 188

Like a hash table.

Answer 189

Partitions the space with a grid divided into stripes.

Answer 190

A pointer to a bucket that contains record pointers.

Answer 191

Good for multiple-key search.

Answer 192

* Space management overhead * Need partitioning ranges that evenly split keys.

Answer 193

Divides a large dataset into smaller subsets (partitions) using a hash function — often to improve performance

Answer 194

Each node contains an attribute-value pair and 2 pointers.

Answer 195

Splits the k-dimensional space along a hyperplane.

Answer 196

Keep disk accesses to a minimum.

Answer 197

* Region quad-trees * Point quad-trees

Answer 198

Data that consists of k-dimensional data regions.

Answer 199

Typically defined by top-right and bottom-left coordinates.

Answer 200

Domain of each attribute mapped onto an n-bit integer.

Answer 201

It converts multi-dimensional data into a single value that can be indexed using conventional methods.

Answer 202

Small differences in one dimension might not appear as 'close' in the Z-value.

Answer 203

Attribute values using bit-vectors.

Answer 204

Combines bit-vectors with bitwise operators (&, |).

Answer 205

Efficient answering of partial-match queries.

Answer 206

Requires fixed record numbers.

Answer 207

A mathematical system with operands and operators

Answer 208

Variables/values from which new values can be constructed

Answer 209

Symbols denoting procedures that construct new values from given values

Answer 210

One or more relations

Answer 211

Always a relation

Answer 212

In a cascaded manner

Answer 213

A subset of the Cartesian product of sets

Answer 214

In a table with k columns

Answer 215

Values from domain Di

Answer 216

Names of the columns

Answer 217

* D1: IDs * D2: Names * D3: DeptIDs

Answer 218

* Represents a relation with k attributes * Each value comes from the domain of the corresponding attribute * Each tuple is distinct * Each attribute contains a single atomic value * Ordering of tuples is immaterial

Answer 219

Refers to the idea that every relation (or table) in a relational database has a name and a defined schema (i.e., a set of named attributes or columns)

Answer 220

The ordering of attributes is significant

Answer 221

D1 is the set of IDs, D2 is Names, and D3 is DeptIDs The relation R is a subset of the Cartesian product of these sets and can be represented by this table with 3 columns Values from domain D3 can only be values allowed in the DeptID column (in the DeptID set), like 'ECS'

Answer 222

Yes, R ∪ S = S ∪ R

Answer 223

Yes, R ∪ (S ∪ T) = (R ∪ S) ∪ T

Answer 224

No, R - S ≠ S - R (except if R=S)

Answer 225

No, R - (S - T) ≠ (R - S) - T

Answer 226

No as locations are different, R X S ≠ S X R

Answer 227

Yes, R X (S X T) = (R X S) X T

Answer 228

Yes, R X (S ∪ T) = (R X S) ∪ (R X T)

Answer 229

Changes attribute names in a relation

Answer 230

ρ_NewRelationName(R) ## Footnote For relation Students(name,age), ρ_S(n,a)(S) renames Students to S and name and age to n and a

Answer 231

Removes attributes not in the specified list

Answer 232

A subset of the relation where tuples satisfy a predicate

Answer 233

* Comparison operators (=, <, >, /=, <=, >=) * Boolean logic operators (∧, ∨)

Answer 234

Like a set, but elements may appear more than once

Answer 235

Combine information from two relations into a new relation ## Footnote Core to relational databases, including Θ-Join, Natural Join, Left Outer Join, Semijoin, Antijoin.

Answer 236

Combines 2 relations using a predicate ## Footnote Equivalent to the Cartesian product followed by a selection using the predicate.

Answer 237

A theta join that only uses the operator = ## Footnote Example: Food's Shop attribute equals Locations' Name attribute.

Answer 238

A join where no predicate is specified ## Footnote Common attributes appear only once.

Answer 239

Cartesian Product of two relations followed by a selection on equality and a projection ## Footnote Involves relations R & S.

Answer 240

Includes tuples from R that don't have corresponding tuples in S ## Footnote Missing values are set to null.

Answer 241

Includes tuples from S that don't have corresponding tuples in R ## Footnote A type of outer join.

Answer 242

Includes tuples from both relations that don't have corresponding tuples in the other relation ## Footnote A type of outer join.

Answer 243

Similar to a natural join, but resulting attributes are only from the left relation ## Footnote Useful for reducing the size of the result.

Answer 244

Contains tuples from R that have no match in S ## Footnote The reverse of a semijoin.

Answer 245

Involves transforming one query expression into another equivalent query expression ## Footnote Aims to enhance efficiency during query processing.

Answer 246

Only the last projection in nested projections is required if not extended ## Footnote Reduces complexity in query execution.

Answer 247

Selections with conjunctive terms can cascade into individual selections ## Footnote Eliminates unnecessary data sooner.

Answer 248

Selection and theta-join are commutative operations ## Footnote Applying the most restrictive filter first reduces rows early.

Answer 249

Joins exhibit associativity ## Footnote Reordering joins helps find the most efficient execution plan.

Answer 250

Selection can be performed on both relations prior to the theta-join ## Footnote If the predicate only involves attributes being joined.

Answer 251

A mathematical system consisting of operands & operators ## Footnote Includes set operations & relation-specific operators.

Answer 252

A relationally complete database query language ## Footnote Can express all relational algebra queries.

Answer 253

SELECT, FROM, and WHERE ## Footnote Fundamental components of SQL queries.

Answer 254

Projection π ## Footnote Used to specify the columns to be returned.

Answer 255

Cartesian Product ## Footnote Specifies the tables involved in the query.

Answer 256

Selection σ ## Footnote Used to filter records based on conditions.

Answer 257

A join where a table is joined with itself ## Footnote Often requires aliases for clarity.

Answer 258

They improve readability and clarity ## Footnote Reduce typing effort in complex queries.

Answer 259

* Created in the FROM list * `FROM AS ` * New names referenced in the select list (SELECT * FROM R AS S ## Footnote Ensures clarity in queries.

Answer 260

Scanning, Parsing & Validating ## Footnote Involves parsing the SQL query into an abstract syntax tree and validating table and attribute names against the system catalogue.

Answer 261

An intermediate form

Answer 262

To find the best way to execute the query

Answer 263

Turns the optimised query into executable code

Answer 264

The executable code generated from the optimised query

Answer 265

* Generating an initial query plan * Optimising it * Translating it into executable code

Answer 266

To improve plan's efficiency without specifying the exact algorithms to be used.

Answer 267

* Using predicate decomposition (breaking down selections with AND conditions) * This technique rewrites a selection with conjunctive predicates (e.g., A AND B) into a sequence of simpler selections. * Example: Instead of `σ_{A AND B}(R)`, we write `σ_A(σ_B(R))`. * This helps in reordering and optimizing the query more effectively. ## Footnote Other techniques include reordering subtrees and combining Cartesian products with adjacent selections.

Answer 268

To minimize the size of intermediate relations, thus reducing the overall cost of the query.

Answer 269

Considering only left-deep trees.

Answer 270

Selecting algorithms for each operator in the logical query plan.

Answer 271

The query plan transitions from an abstract representation to a concrete, executable strategy.

Answer 272

I/O costs and memory usage.

Answer 273

It might make a particular join order more efficient, leading to a different logical plan.

Answer 274

A left-deep tree of products, a conjunctive selection above the products, and a project of the output attributes above the selection.

Answer 275

efficient execution plan.

Answer 276

To predict the resources needed to execute a query plan.

Answer 277

To minimize the sizes of the intermediate relations, thus minimizing the overall cost of the plan.

Answer 278

As the cardinality of its output relation.

Answer 279

The sum of the cardinalities of the intermediate relations, excluding the input relations and the final result.

Answer 280

Statistics about each relation, used to make informed decisions about query costs.

Answer 281

The number of tuples in relation R (cardinality of R).

Answer 282

The number of distinct values for attribute (column) A in relation R.

Answer 283

T(scan(R)) = T(R) and V(scan(R), A) = V(R, A) for all attributes A in R.

Answer 284

T(R × S) = T(R)T(S) and V(R × S, A) = V(R, A) for all attributes A in R.

Answer 285

The projection does not remove duplicate tuples.

Answer 286

It leads to specific filtering of tuples based on the constant value.

Answer 287

Inequality, not equals, conjunction, disjunction

Answer 288

T(R ⋈ S) = 0.

Answer 289

T(R ⋈ S) = T(R).

Answer 290

Divides the product of the number of tuples in R and S by the maximum number of distinct values in either R.A or S.B.

Answer 291

Methods include histograms that provide a more detailed view of data distribution within attributes.

Answer 292

That each attribute value appears with equal frequency.

Answer 293

Equal-Width Histograms.

Answer 294

Sort tuples by attribute, divide them into equal-sized sets, and provide the maximum value for each set.

Answer 295

The top-n most common values for each attribute, along with their frequency counts.

Answer 296

To compare all the possible plans.

Answer 297

The Catalan numbers using the formula: ## Footnote Catalan numbers are a sequence of natural numbers that have many applications in combinatorial mathematics.

Answer 298

n! permutations. So total possible niary join trees with permuations = Catalan number x n! ## Footnote The factorial of n (n!) represents the product of all positive integers up to n.

Answer 299

The number of joins and the order they are executed in.

Answer 300

They produce smaller output relations.

Answer 301

Managing the complexity of join ordering and visualising relationships between relations.

Answer 302

Undirected graph.

Answer 303

The relations (R1, R2, …, Rn) used in the query.

Answer 304

Join predicates between relations.

Answer 305

A predicate of the form a = aj, where a belongs to relation Ri and aj belongs to relation Rj.

Answer 306

A predicate of the form a = const, where a belongs to relation Ri.

Answer 307

To constrain the search space by excluding join orderings that would lead to cross products.

Answer 308

Linear join trees and bushy join trees.

Answer 309

Every join introduces at least one base relation.

Answer 310

Where output signals from an operator are written to disk in a semi-permanent fashion.

Answer 311

Where data is read, processed, and propagated to the next operator in the query plan without materialisation.

Answer 312

Left-deep, right-deep, or zig-zag.

Answer 313

n!C(n-1) = (2n)!/n!.

Answer 314

Use transformation rules to modify the query plan. ## Footnote 1. Start with canonical form 2. Push σ operators down the tree 3. Introduce joins (combine x and σ to create ⋈) 4. Determine join order 5. Push π operators down the tree

Answer 315

A structured representation of a query upon which optimisation techniques can be applied.

Answer 316

left-deep.

Answer 317

To move selection conditions closer to the base relations.

Answer 318

To potentially reduce the number of tuples processed based on join selectivity. ## Footnote How many projects called 'Aquarius'? Probably 1 How many people born after 1957? Quite a lot So move the PROJECT subtree down so it's at the bottom of the query plan

Answer 319

If intermediate relations are to be kept in buffers, reducing the degree of those relations (no. attributes) allows us to use fewer buffer frames

Answer 320

A predicate that relates something in one relation to something in another relation.

Answer 321

The interface that connects operators, how data is propagated between operators, how operators are scheduled within a query plan.

Answer 322

To pipeline data through the operators, processing tuples one at a time to minimize materialization and maximize in-memory operations.

Answer 323

A standard interface for operators that facilitates data flow and pipelining.

Answer 324

* open() * getNext() * close()

Answer 325

Operators don't generate tuples until getNext() is called.

Answer 326

To minimize blocking.

Answer 327

* Push model * Pull model * Stream model

Answer 328

* Data flows from leaves upward. * Producer pushes data as soon as it's ready. * Consumer buffers data until it calls getNext(). * Minimizes idle time; good for pipelining.

Answer 329

* Data flow starts from the root. * Producer waits for getNext() before sending data. * Synchronous style but can include some asynchrony.

Answer 330

* Uses FIFO queues between operators. * Producer adds tuples to queue immediately. * Consumer reads from queue if data is available (non-blocking). * Supports parallelism; asynchronous operators, synchronous streams.

Answer 331

Algorithms that *implement* the basic relational operations used in query plans.

Answer 332

* Structure and size of the relations * Presence of indexes & hashes

Answer 333

Refers to storing tuples from different relations that can be joined (based on shared attributes) in blocks together.

Answer 334

Tuples from the same relation are stored together in blocks, though not necessarily contiguous on disk.

Answer 335

An index that allows tuples to be read in an order that corresponds to their physical order on disk.

Answer 336

The operation of reading all tuples of a relation that satisfy some predicate.

Answer 337

* Table Scan * Index Scan

Answer 338

B(R) disk accesses.

Answer 339

B(R) + B(I_R) disk accesses.

Answer 340

Algorithms that read data from the disk only once and require that at least one argument fits in main memory.

Answer 341

* Unary, tuple at a time * Unary, full-relation * Binary, full-relation

Answer 342

Algorithms that can be applied to a single tuple at a time and are non-blocking.

Answer 343

Generally B(R) or T(R) depending on clustering.

Answer 344

Algorithms that require the entire relation to be examined before producing output and may be blocking

Answer 345

M ≥ B(δ(R)) + 1 blocks of main memory. Costs B(R)

Answer 346

Union, intersection, difference, product, and join.

Answer 347

B(R) + B(S).

Answer 348

A basic join algorithm that typically requires multiple passes over the inner relation for each tuple in the outer relation.

Answer 349

Very expensive, as each tuple requires a separate disk access.

Answer 350

The outer relation is read in chunks, improving performance.

Answer 351

Performance can be improved by reversing the order of the relations in the join.

Answer 352

The cost is reduced as it uses the number of blocks in the cost calculation.

Answer 353

An efficient join method that reads each block of R1 and R2 once only.

Answer 354

When the data is too large to fit in main memory, requiring multiple passes over the data to perform an operation. ## Footnote Two-pass algorithms are essential in scenarios involving large datasets, such as external sorting or joining operations.

Answer 355

Tuples are read, written, read, and written. Each (e.g.) 100 block chunk of R is: * Read * Sorted in memory * Written to disk * Read all the chunks * Merge the chunks * Write them out to a file. ## Footnote Merge Sort is efficient for external sorting due to its sequential access pattern.

Answer 356

Relation is partitioned into M-1 buckets. * Read relation a tuple at a time * Hash each tuple to a bucket * Move full buckets to disk and reinitialize * Join R1, R2. ## Footnote This method reduces the overall memory requirement by processing smaller partitions.

Answer 357

The attribute on which we're joining has an index that fits in memory. ## Footnote Index-based algorithms optimize join operations by leveraging existing indexes to minimize disk accesses.

Answer 358

1 expected matching tuple for each tuple in R1. ## Footnote This scenario is common in relational databases where foreign key relationships are established.

Answer 359

Expected matching tuples = 10,000 / 5,000 = 2. ## Footnote This calculation assumes a uniform distribution of values in R2.

Answer 360

Expected matching tuples = 10,000 / 1,000,000 = 1/100. ## Footnote This reflects a scenario with a much larger domain, indicating sparser matches.

Answer 361

How a DBMS supports multiple users accessing the system simultaneously ## Footnote Concurrency is essential for multi-user environments.

Answer 362

They create a copy, modify it, and then commit the changes back to the central database ## Footnote This process can lead to potential consistency issues if not managed correctly.

Answer 363

A database is in a consistent state when all constraints are satisfied ## Footnote Constraints are derived from application requirements.

Answer 364

Multiple clients modifying the same database can lead to problems ## Footnote If not managed properly, it can compromise consistency and isolation.

Answer 365

A basic action on a database by a client program (read, write) ## Footnote Examples include read(X) and write(X).

Answer 366

Reads a database item Xd into a program variable XT in transaction T ## Footnote This operation retrieves data for processing.

Answer 367

Writes the value of program variable XT in transaction T into the database item Xd ## Footnote This operation updates the database with new information.

Answer 368

Two operations on a database conflict if their execution doesn't commute ## Footnote Conflicts can lead to inconsistencies.

Answer 369

A set of operations meant to be executed together (like in a block) ## Footnote Transactions ensure data integrity.

Answer 370

It terminates successfully; otherwise, it aborts/fails ## Footnote This is critical for maintaining database integrity.

Answer 371

Atomic, Consistent, Isolated, Durable ## Footnote These properties ensure reliable transaction processing.

Answer 372

All operations within a transaction are executed, or none are ## Footnote This ensures that partial transactions do not affect the database.

Answer 373

A transaction must leave the database in a consistent state, adhering to defined constraints ## Footnote This is vital for maintaining data integrity.

Answer 374

Each transaction must read from a consistent database state, isolated from the 'work in progress' of other transactions ## Footnote Isolation helps prevent conflicts between transactions.

Answer 375

The effects of a committed transaction aren't lost ## Footnote This ensures that once a transaction is completed, its changes are permanent.

Answer 376

To execute transactions serially ## Footnote However, this can lead to longer wait times for users.

Answer 377

Users may have to wait a long time before getting an answer from the system ## Footnote This can decrease system throughput.

Answer 378

Operations from different transactions may be interleaved, potentially breaking consistency/isolation ## Footnote This is a challenge in multi-user database environments.

Answer 379

Occurs when concurrent transactions overwrite each other's updates without proper management ## Footnote This leads to data loss or inconsistency.

Answer 380

Occurs when one transaction reads data that is modified by another ongoing transaction ## Footnote For example, checking a balance while a transfer is in progress.

Answer 381

Occurs when a transaction reads data written by another transaction that has not yet been committed ## Footnote This can lead to inconsistencies if the first transaction rolls back.

Answer 382

Each transaction defines a partial order of operations that indicates what needs to be executed before what.

Answer 383

A schedule is a partial order, constructed from the union of operations performed by multiple transactions.

Answer 384

The order of operations as defined within each individual transaction.

Answer 385

To ensure database consistency.

Answer 386

A transaction schedule is considered serialisable if its effect on the database is equivalent to that of a serial schedule.

Answer 387

A schedule that produces the same result as executing transactions one after the other, without any overlap.

Answer 388

A schedule that cannot be rearranged to produce a serial effect on the database.

Answer 389

Receiving a set of concurrent transactions as input and producing a serialisable schedule as output.

Answer 390

* Number of concurrent transactions * Throughput * Consistency

Answer 391

* Locking (and 2PL) * Timestamp Ordering

Answer 392

Locks data items that are currently being accessed by a transaction to prevent others from accessing/modifying the same data until the lock is released.

Answer 393

The requesting transaction waits until the lock is released or if it is aborted.

Answer 394

A lock for reading that allows multiple transactions to hold shared locks on the same data item simultaneously.

Answer 395

A lock for writing that allows only one transaction to hold an exclusive lock on a data item.

Answer 396

A table maintained by the DBMS to track the lock status of each data item.

Answer 397

Whether a lock request can be granted or not.

Answer 398

A concurrency control technique that guarantees serialisable transactions.

Answer 399

* Growing Phase: Transactions obtain locks on data items * Shrinking Phase: Transactions release the locks they have acquired

Answer 400

A situation where two or more transactions are waiting for each other to release a lock on an item.

Answer 401

* Concurrency * Hold * Wait * Mutual dependency

Answer 402

Every transaction locks all items it needs in advance; if an item can't be obtained, no items are locked.

Answer 403

Detecting the deadlock and aborting one of the involved transactions.

Answer 404

A directed graph where each executing transaction is a vertex, and an edge exists if one transaction is waiting on another.

Answer 405

The system assumes that the transaction is deadlocked and aborts it.

Answer 406

Locking larger units reduces concurrency but can prevent deadlocks; smaller units increase concurrency but can lead to more deadlocks.

Answer 407

An alternative to locking that aims to ensure serialisable schedules based on transaction timestamps.

Answer 408

Conflicting operations are executed in order of their transaction timestamps.

Answer 409

If TS(T) >= write-TS(X), the read operation is executed; otherwise, the transaction is aborted.

Answer 410

If TS(T) >= read-TS(X) and TS(T) >= write-TS(X), the write operation is executed; otherwise, the transaction is aborted.

Answer 411

To ensure Atomicity and Durability in the presence of failures/crashes.

Answer 412

A persistent record of changes made during DB operation.

Answer 413

Append-only files.

Answer 414

By reading from the log.

Answer 415

A type of logging used to repair a DB by undoing incomplete transactions after a crash.

Answer 416

must be written to disk.

Answer 417

The value of X is set back to old_value, and incomplete transactions are undone.

Answer 418

* U1: A log record `` must be written before outputting new value. * U2: A `` log record must be written after all changes are output.

Answer 419

It potentially causes more disk I/O operations.

Answer 420

Creates a point ensuring all transactions before that point have committed or aborted.

Answer 421

* Stop accepting new transactions. * Wait for active transactions to commit or abort. * Flush the log. * Write a checkpoint record `` to the log. * Flush the log again. * Resume accepting transactions.

Answer 422

Allows new transactions to enter during the checkpoint process.

Answer 423

* `` * ``.

Answer 424

Disregard the log before the previous ``.

Answer 425

Disregard the log before the start of the earliest incomplete transaction.

Answer 426

Reapplies changes made by transactions that successfully committed before a crash.

Answer 427

Ensures that the effects of committed transactions aren’t lost.

Answer 428

Before modifying a database item X on disk, all log records related to the modification (``, ``) must be written to disk

Answer 429

* Write log record `` and flush log. * Write to disk all database items that have been written to buffers by committed transactions. * Write log record `` and flush log.

Answer 430

Ignore changes made by transactions that committed before the corresponding ``.

Answer 431

Search back to the previous ``.

Answer 432

The situation where DBMS performance is primarily limited by the time taken to access secondary storage (HDD/SSD) ## Footnote Latency of storage, particularly writing to the disk, is the most expensive part of all DB operations.

Answer 433

Has a single processor, tasks may be interleaved but does not achieve true parallelism/concurrency, a single bank of memory, a single buffer pool, and a single disc ## Footnote DB stored on a single storage device.

Answer 434

Tightly coupled, employs a symmetric multiprocessor (SMP) system, multiple processors for executing tasks in parallel, single global memory, and a single buffer ## Footnote Creates a single point of failure.

Answer 435

* Multiple processors for parallel execution * Distributed memory, each processor has its own local memory * Loosely coupled, processors communicate through an interconnection network or a switch * Multiple discs

Answer 436

* Massively parallel * Loosely coupled, processors operate independently * Distributed memory, each processor has its own private memory * Each processor owns part of the data * A data page resides in the buffer pool of only one local memory

Answer 437

* How to divide a query/transaction among nodes * How to partition data across nodes * How to keep the data partition balanced * How to control concurrency and avoid deadlocks * What to do if a node fails/crashes

Answer 438

To overcome the I/O bottleneck by splitting the processing and accessing of data across multiple processors and disks.

Answer 439

Work is divided among several processors, usually with a coordinator process managing execution and multiple worker processes performing parts of the query in parallel.

Answer 440

Involves executing different queries concurrently on different processors.

Answer 441

Involves decomposing a single query into sub-queries to be executed in parallel on different processors.

Answer 442

An operator is executed in parallel on different subsets of the data.

Answer 443

The output of one sub-operator is passed as input to another sub-operator, allowing them to execute concurrently.

Answer 444

Operators decomposed into independent operator instances, performing the same operation on different subsets of data.

Answer 445

Operations are overlapped, allowing data to be pipelined from one stage to the next without materialisation.

Answer 446

Subtrees in query plan executed concurrently.

Answer 447

Refers to how we divide data ## Footnote Aims to reduce execution time and balance I/O load evenly across processors

Answer 448

* Number and capacity of available nodes (not processors) * Query workload * Network topology and distance

Answer 449

Uses modulo calculation to distribute data randomly ## Footnote Can lead to related data being spread across many nodes, making certain queries slow

Answer 450

Applies a hash function to a subset of attributes of a row to determine storage node ## Footnote Guarantees balanced distribution of rows across nodes

Answer 451

Efficient for exact-match queries ## Footnote Example: Fast for WHERE StudentId=X, slow for WHERE Exam > 40

Answer 452

Allows more control over row placement and can optimize queries on the chosen attribute ## Footnote Effectiveness depends on having stable, predictable data ranges

Answer 453

Can lead to imbalanced partitions depending on data distribution

Answer 454

To keep partitions balanced

Answer 455

Worker nodes send relevant data to a central coordinator process ## Footnote The coordinator performs main query operations on aggregated data

Answer 456

Each worker sends its 25,000 tuples to the coordinator, which processes the combined 100,000 tuples

Answer 457

Coordinator sends the query to worker nodes, which execute it locally and send results back ## Footnote The coordinator combines partial results to produce the final answer

Answer 458

Can significantly reduce network traffic if the query is selective

Answer 459

Decomposition for some operators or queries can be difficult

Answer 460

Effectiveness heavily depends on how the data is partitioned

Answer 461

Crucial for making informed decisions about shipping strategy efficiency

Answer 462

Query shipping leverages parallelism of worker nodes, while data shipping may bottleneck at the coordinator

Answer 463

* Scan and union are free * Select costs are Card(R) * Aggregate costs are Card(R) * Join costs depend on indexing

Answer 464

Costs Card(R) to ship relation R to another node

Answer 465

Costs Card(R) due to parallel shipment

Answer 466

Operations taking longer should be depicted with wider blocks for clarity

Answer 467

Optimizing a query plan as if on a single processor

Answer 468

Involves operations where no join between relations is needed, such as aggregations or selections within a single table.

Answer 469

Used wwhen the relations that need to be joined are partitioned on the same key attribute, so corresponding tuples from the two relations reside on the same node.

Answer 470

Used when the relations to be joined are partitioned on their respective join key attribute, but not necessarily the same key, so co-location is not guaranteed.

Answer 471

Used when one of the relations to be joined is relatively small compared to the other.

Answer 472

Used for when the relations are large and not partitioned on the join key.

Answer 473

The ability of a system to maintain data integrity and consistency and to recover from failures.

Answer 474

Ensures that once a transaction is committed, the changes to the database are persistent.

Answer 475

Requires that for any transaction, either all operations are executed successfully, or none are.

Answer 476

They are vital for achieving durability and atomicity.

Answer 477

Transaction T has started execution.

Answer 478

Transaction T has completed successfully and will make no further changes to database items.

Answer 479

Transaction T could not complete successfully. No changes made by T will be copied to disk.

Answer 480

Involves a coordinator and one or more worker nodes to ensure atomicity of transactions.

Answer 481

Coordinator sends a 'prepare T' message to all worker nodes, which execute their part and send back a 'vote-commit T'/'vote-abort T' message.

Answer 482

Coordinator analyzes votes from workers and makes a commit or abort decision.

Answer 483

If all workers vote-commit, the coordinator commits the transaction and sends commit messages to all workers.

Answer 484

If any worker votes-abort, the coordinator aborts the transaction and informs all workers.

Answer 485

Activated when a timeout occurs.

Answer 486

Assumes participants are aware of each other and tries to find out the coordinator's decision after a timeout.

Answer 487

Initiated when a coordinator or participant restarts after a crash.

Answer 488

Allows the coordinator to forget about transactions if the global decision is to abort.

Answer 489

Assumes that if no information about a transaction is in memory, it must have been committed.

Answer 490

persistent, append-only record of changes.

Answer 491

It indicates that a worker is ready to commit or abort changes locally.

Answer 492

A system consisting of multiple machines that are far away from each other, controlled by the same organization, typically in different data centers.

Answer 493

Characteristics include: * Multiple machines (>50) * Homogeneous data format (relational) * Same hardware across machines * No reliance on a central site.

Answer 494

To break down large databases into smaller, more manageable units.

Answer 495

The fragments should be: * Disjoint * Complete.

Answer 496

No tuple of the global relation appears in more than one fragment.

Answer 497

Every tuple in the original global relation must belong to at least one of the fragments.

Answer 498

Goals include: * Reduce overall execution time * Balance I/O load evenly across nodes.

Answer 499

Factors include: * Number and capacity of available nodes * Query workload * Network topology and distance.

Answer 500

It deals with how a query formulated against a global, non-distributed database schema operates on fragments of data across different nodes.

Answer 501

Involves substituting localization programs for relations in the original query, potentially leading to expensive operations.

Answer 502

Techniques employed to optimize queries by reducing the amount of data transferred across the network.

Answer 503

Reduction with selection Reduction with join

Answer 504

Redution with projection

Answer 505

Applied when a query includes a selection operation (WHERE), identifying irrelevant fragments to avoid accessing them.

Answer 506

Optimizes join operations by identifying pairs of fragments whose join will be empty and avoiding unnecessary joins.

Answer 507

Aims to reduce data processed and transferred by eliminating unnecessary fragments for a query involving a projection operation.

Answer 508

Moves only the part of one relation needed for the join.

Answer 509

self-sufficient.

Answer 510

To reconstruct global relations using expressions derived from fragments.

Answer 511

Serialisability, the highest isolation level ## Footnote 2PL is essential for preserving isolation in distributed DBs.

Answer 512

* Centralised 2PL (C2PL) * Distributed 2PL (D2PL) ## Footnote C2PL uses a single site for lock management, while D2PL has lock managers at each site.

Answer 513

The transaction manager sends a lock request to the central lock manager ## Footnote The central lock manager decides whether to grant the lock.

Answer 514

Single point of failure and potential bottleneck ## Footnote Every lock request and release must go through the central lock manager.

Answer 515

To the local lock manager at each participant site ## Footnote Each participant site has its own lock manager.

Answer 516

It sends an 'end of operation' message back to the transaction manager ## Footnote This informs the transaction manager that the operation is complete.

Answer 517

Occurs when 2 or more transactions are waiting for each other to release a lock on an item ## Footnote Deadlocks can severely impact system performance.

Answer 518

* Concurrency * Hold * Wait ## Footnote These conditions must be met for a deadlock to occur.

Answer 519

A directed graph that represents the dependencies between transactions ## Footnote It helps in detecting deadlocks.

Answer 520

The presence of a cycle ## Footnote A cycle in the WFG signifies that transactions are waiting indefinitely.

Answer 521

* Local WFG * Global WFG ## Footnote Local WFGs are per site, while Global WFGs consider all transactions.

Answer 522

Transaction pre-declaration ## Footnote Transactions declare all accessed data items in advance. TM only locks if all items are available.

Answer 523

Requiring transactions to always access resources in a predefined order ## Footnote This can be challenging in dynamic databases.

Answer 524

Using timestamps to decide which transaction to abort when a lock request is denied ## Footnote Rules like WAIT-DIE and WOUND-WAIT are examples of this approach.

Answer 525

An older transaction waits for a younger one, while a younger one is aborted if it requests a lock held by an older one ## Footnote This helps to manage transaction priorities.

Answer 526

One site acts as the deadlock detector, checking for cycles in the Global Wait-For Graph ## Footnote Each site sends its Local WFG to the central detector.

Answer 527

Single point of failure ## Footnote If the deadlock detector fails, deadlocks may go undetected.

Answer 528

Organizing deadlock detectors in a hierarchy for monitoring ## Footnote Local detectors report to higher-level detectors.

Answer 529

Responsibility for deadlock detection is shared among sites ## Footnote This allows for a more robust detection mechanism.

Answer 530

Using timeouts to abort transactions waiting too long for a resource ## Footnote This method assumes that long wait times indicate a deadlock.

Answer 531

Replication is an extension of the fragmentation problem that involves creating multiple copies of data across different geographical locations.

Answer 532

* Latency reduction * Availability * Resilience * Performance

Answer 533

By storing copies of the data closer to users in different geographical areas.

Answer 534

It increases availability by having multiple copies; if one replica fails, data can be accessed from other replicas.

Answer 535

If a node containing a replica fails, transactions can be rerouted to other nodes with copies of the required data.

Answer 536

It balances the read workload across multiple replicas, reducing bottlenecks and increasing throughput.

Answer 537

Each replica has its own transaction management system.

Answer 538

All copies of an item have the same value after the execution of an update transaction.

Answer 539

All copies of an item will eventually have the same value after the execution of an update transaction.

Answer 540

Epsilon defines a bound on the allowed inconsistency, specifically the number of missing writes.

Answer 541

It allows reads as long as they are within a defined bound of time units.

Answer 542

It considers the average/combined temporal difference across multiple items accessed in the same transaction.

Answer 543

Mutual consistency means replicas have the same value, but does not ensure that updates occurred in a single, step-by-step order.

Answer 544

A type of data object that can be used in replicated systems to facilitate lazy distribution of updates without leading to conflicts.

Answer 545

* Associative * Commutative * Idempotent

Answer 546

Updates can be lazily distributed without immediate propagation, ensuring the final state is consistent regardless of update order.

Answer 547

It is practical for systems with network latency or intermittent connectivity.

Answer 548

The dynamic rate of participation of nodes within the network.

Answer 549

It makes fragmentation and replication of data more challenging.

Answer 550

Extreme autonomy with no control of the network topology.

Answer 551

The lack of a central overseer.

Answer 552

* Very resilient * Supports maximum autonomy

Answer 553

* Unpopular items are not replicated enough * Enormous communication cost

Answer 554

To manage data and queries.

Answer 555

The range of hash keys stored in a specific node.

Answer 556

It becomes expensive due to the need for synchronisation.

Answer 557

A network where nodes connect to virtual nodes to control topology.

Answer 558

* Tree * Hypercube * Ring

Answer 559

Each node maintains a routing table that stores the address of one node representative of a different prefix.

Answer 560

Creation of multiple identities by an attacker to manipulate voting processes.

Answer 561

By shifting validation from identity count to computational effort.

Answer 562

Difficulty tuning the puzzle can affect transaction confirmation speed.

Answer 563

A system where participants stake their own economic value to become validators.

Answer 564

They lose their stake.

Answer 565

* Requires validators to reveal identity * Openly auditable transactions

Answer 566

Cryptography.

Advanced Databases Flashcards

(612 cards)