Glossary

3 4

A

Active Server Pages (ASP) page A Web page that creates HTML pages dynamically on the server. ASP pages are commonly used to access Microsoft SQL Server from the World Wide Web.

ad hoc A term used to describe a spontaneous query. Typically, an ad hoc query is not optimized and is typed into Query Analyzer, isql, or osql, not used through an application.

agent A program that runs independently within Microsoft Windows NT or Windows 2000 to perform a service. An agent usually has a particular task, such as scheduling operations or performing replication tasks. The equivalent in the UNIX world is a daemon.

aggregate function A function that performs an operation on a set of values and returns a single value. The SQL Server aggregate functions are AVG, COUNT, DISTINCT, GROUP BY, HAVING, MAX, MIN, STDEV, STDEVP, SUM, VAR, and VARP.

API See application programming interface.

AppleTalk The Apple networking protocol.

application The code that communicates with the user. A database application typically also communicates with the RDBMS.

application programming interface (API) The standard and documented interface that programs are coded to. By coding to an API, developers can create programs that use external functions, such as SQL Server access functions.

article A table or subset of data that is selected to be replicated.

ASP See Active Server Pages (ASP) page.

Asynchronous Transfer Mode (ATM) A network hardware protocol.

B

backup A copy of the contents of a SQL Server database to be used in the event of a system failure. The backup can be copied back into the database in order to return the database to the state it was in when the backup was performed.

bandwidth The throughput capacity of a device or system. This term is typically associated with a network device or computer bus.

batch processing system A system distinguished by scheduled loading and processing of data in groups. These long-running jobs are done off line, with little or no user intervention.

BCP See Bulk Copy Program.

bit The smallest unit of data. A bit is either on (1) or off (0).

block See page.

bottleneck A performance-limiting component. The term comes from an analogy to a bottle containing liquid: the narrowing of the neck slows the flow of liquid.

branch node An intermediate node in an index. Branch nodes are between the root node and the leaf nodes.

buffer cache See cache.

bulk copy A generic term relating to the act of copying large amounts of data to or from a SQL Server database. Typically, the BULK INSERT T-SQL statement is used to perform a bulk copy.

Bulk Copy Program (BCP) A program provided with SQL Server that is used for loading data from text files into the database.

BULK INSERT A T-SQL command used to copy large amounts of data from a data file into a SQL Server table from within SQL Server.

bulk load The act of loading the database by using bulk insert operations. This can be accomplished by using either the BULK INSERT T-SQL statement or the BCP utility.

byte A sequence of 8 binary bits. A byte is the basic unit of data used in computers. It is made up of bits (ones and zeros) that represent numbers, characters, and so forth.

C

cache The RAM used to hold frequently accessed data in order to improve performance. SQL Server contains its own cache made up of Windows memory used to hold commonly accessed pages. This is referred to as the SQL Server buffer cache or page cache. CPU chips contain their own on-board caches, and some I/O controllers also contain caches.

capacity planning The activity of planning for the increased utilization of the system and, by doing so, anticipating capacity that must be added. This task is performed in order to maintain the level of service expected by the user community while the load on the system is increasing.

capacity-planning measurement A collection of performance statistics (counters) used to determine the resource consumption of a workload in order to plan for additional hardware resources. This type of measurement is of a longer duration than a performance measurement (usually hours or even a day).

Cartesian product The result of a join that does not have a where clause. The size of the resulting data set is the number of rows in the first table multiplied by the number of rows in the second table. A Cartesian product is not usually an intentional result of a query.

central processing unit (CPU) The brains of the computer; that is, the chip that processes data.

checkpoint An operation performed to synchronize the data files with the current contents of the database memory cache in order to reduce the recovery time needed in the event of a system failure. The checkpoint process traverses the list of dirty pages and flushes them to disk.

client/server model The programming model in which the user interface portion of the program resides in an executable program on a user's PC and accesses data in a database on a server. The application logic is divided between the client (PC) and the server.

cluster See clustered index, Microsoft Cluster Services.

clustered index A combination of an index and a table. The index is stored as a B-tree, and the data is stored in the leaf nodes of the index.

collision A condition that occurs when two controllers attempt to use the network at the same time. Each controller detects the collision and usually waits a random amount of time before trying again. In a heavily utilized network, collisions can cause a performance bottleneck. This term is usually associated with Ethernet networks.

column A collection of corresponding fields in all rows of data in a table. Each column has a data type associated with it. All the columns in a row make up a database record.

COMMIT The SQL statement that finalizes an SQL transaction. Until the COMMIT statement is issued, the transaction can be undone by the issuing of the ROLLBACK statement.

Component Object Model (COM) COM technologies are a set of APIs and tools developed by Microsoft. These applications run under the control of Microsoft Transaction Services (MTS) on the server system and under the control of Distributed COM (DCOM) on client systems. Developers create COM cli-ents by using platforms such as Microsoft Visual Basic and Microsoft Visual C++. Applications can also be made up of newer technologies such as ASP pages and Internet Server API (ISAPI).

Component Services The component services is a set of products based on the Component Object Model (COM) and Distributed COM (DCOM), Microsoft Transaction Server, Microsoft Internet Information Server, and Microsoft Message Queue Server from Windows NT 4.0. With Windows 2000 the COM and DCOM models have evolved to the next level, COM+. The COM+ applications and other system services make up the Windows 2000 Component Services.

composite index An index that is created on a combination of two or more columns.

configuration parameter A variable that allows you to change the behavior of SQL Server. These parameters are used to tune the SQL Server engine. Configuration parameters affect things such as SQL Server memory allocation, thread counts, and the number of user connections allowed.

constraint A restriction placed on a table to guarantee integrity.

CPU See central processing unit.

cube A multidimensional representation of both detail and summary data. Cubes are typically used in OLAP.

D

database A repository of data. This can be anything from a small list of names to a record of the entire population of the world.

database administrator (DBA) The person or persons responsible for maintaining the database. The DBA's role can vary depending on a company's needs.

database consistency checker (DBCC) A utility for finding and correcting problems in the consistency of the database.

database management system (DBMS) The programs, files, processes, and memory that make up the database. SQL Server is a relational DBMS (RDBMS).

database role A role that helps in the assignment of permissions to users. Rather than assigning individual permissions to individual users, you can create a role that represents the permissions granted to a set of users. When this method is used, only the roles need be assigned to the users.

data definition language (DDL) SQL statements that are used in the definition or declaration of database objects. DDLs include statements such as CREATE DATABASE and DROP DATABASE.

data file The physical operating system file that holds the data associated with a filegroup and, in turn, the database. Under SQL Server, this can be an NTFS file or a Windows NT/2000 raw device.

data integrity See integrity.

data manipulation language (DML) SQL statements that are used to insert data into the database or to update, delete, or retrieve database data.

data mart A decision support system created from company OLTP data. A data mart differs from a data warehouse in that the former contains data that is typically used by one business segment, such as accounts receivable or accounts payable, whereas the latter can contain all a company's data.

Data Transformation Services (DTS) The SQL Server tool for transforming data between systems.

data warehouse A decision support system created from OLTP data. A data warehouse is usually quite large, sometimes containing terabytes of data.

DBA See database administrator.

DBCC See database consistency checker.

DB-LIB The abbreviation of DB-Library; a SQL Server connectivity protocol.

DBMS See database management system.

DDL See data definition language.

decision support system (DSS) A database system that is used to aid in business decisions based on data in the database. These decisions can be based on sales trends, product sales, and so forth.

differential backup A mode of backup that saves only the data that has changed since the previous backup. When you use this mode, you can back up the database faster than you could using a full backup, but a database restore usually takes longer when a differential backup is used.

dirty page A page in the SQL Server cache that has been modified but has not been written to disk.

disk In this book, the term identifies a fixed disk drive. A fixed disk drive is a magnetic medium that is used to store data. A disk provides persistent storage: data remains on the disk even after power is turned off or fails.

distributor The intermediary component in SQL Server replication. A distributor takes replication data from a publisher and presents it to a subscriber.

DML See data manipulation language.

domain A Windows 2000 or Windows NT network group. Systems in a domain use the same user list and passwords.

DSS See decision support system.

DTS See Data Transformation Services.

E

encryption The process of encoding data or stored procedures for security reasons.

enterprise A term for a system that services the entire company.

Enterprise Manager The main utility used to administer the SQL Server RDBMS.

equijoin A join that uses an equality operator in the WHERE clause of the SQL statement performing the join.

Ethernet A popular network hardware protocol.

execution plan The method that the parsed SQL Server statement uses to perform operations on the database.

F

fault tolerance The ability of a subsystem to continue functioning after a component has failed. This is accomplished via the use of redundant components. Fault tolerance is typically associated with the I/O subsystem and RAID controllers, but other subsystems, such as power supplies, can be fault tolerant.

fiberor fiber optics A network hardware protocol.

Fibre Channel A new I/O protocol that operates using either copper wire or fiber-optic connection hardware.

field See column.

filegroup A group of files that is used as a repository for SQL Server objects. A filegroup can be made up of one or more data files. When SQL Server objects are created, they can be assigned to a specific filegroup.

filegroup backup A form of backup, introduced in SQL Server version 7, that enables you to back up and restore a single filegroup, rather than the entire database. This allows for the entire database backup to be performed over several days.

foreign key A field in a table that is also a primary key in another table.

FOREIGN KEY constraint A requirement that a foreign key be a valid primary key in the database. This type of constraint is typically used to verify that referenced data is available.

full backup A backup that saves all data in a database.

full table scan A select operation in which all rows in the table are read in order to find the desired data. A full table scan is not usually desired, because it involves a large amount of I/O overhead.

G

gigabyte (GB) A unit of data equal to 2³⁰ (1,073,741,824) bytes or 1024 megabytes.

graphical user interface (GUI) An application interface that is graphical in nature in order to simplify the use of the application. A GUI can be used to show data in an easy-to-understand format.

GROUP BY An SQL clause used to divide a table into groups of data. These groups can consist of column names or results in computed columns.

GUI See graphical user interface.

H

HBA See host bus adapter.

hint An optional clause you can make to an SQL statement to give Query Optimizer information about how you would like the execution plan to be constructed.

host bus adapter (HBA) A computer adapter used to communicate with an I/O bus.

hub A passive network device that is used to connect the network cards from multiple systems. A hub is an electrical device that has no application logic; thus, hubs are extremely fast.

I

index An auxiliary data structure used to speed access to data within the database.

index scan An operation that occurs when a group of index entries have to be read in order to find the desired data. Such a scan is typically used when a composite index is accessed and not all the columns that make up the index are supplied in the WHERE clause of the query.

input/output (I/O) The movement of data from one computer component to another. The term can be used to describe access to the disk subsystem or other data transfer components.

integrity The accuracy and completeness of data in the database. When data integrity is preserved, only valid data (data consistent with the business model) exists in the database.

Internet Server API (ISAPI) A set of function calls that are designed to provide Internet application developers with a method of extending the functionality of IIS.

Internetwork Packet Exchange Sequenced Packet Exchange (IPX/SPX) A network protocol used in Novell networks.

I/O See input/output.

I/O capacity The amount of work that can be supported by the I/O subsystem. The I/O capacity is constrained by either the number of I/O operations (if random seeks are the limiting factor) or throughput (if seeks are sequential).

IPX/SPX See Internetwork Packet Exchange/Sequenced Packet Exchange.

ISQL An application supplied by Microsoft that allows access into the SQL Server RDBMS. ISQL uses the DB-LIB protocol; OSQL uses the ODBC protocol.

J

JBOD An acronym for "just a bunch of disks"; a non-RAID disk configuration.

join An operation that combines data from two or more tables by taking advantage of the relationship between the tables.

K

key The column or set of columns that is used to define the access point into an index. If a composite index is made up of two columns, those two columns make up the index key.

kilobyte (KB) A unit of data equal to 2¹⁰ (1024) bytes.

L

latency The term used to describe the time a process spends waiting for an operation to be completed. Most often, the term is used to describe the time a thread waits for an I/O operation to be completed.

lazywriter The thread that takes dirty blocks from the SQL Server cache and writes them to disk. It is called the lazywriter because it is a background process that operates according to its own schedule and that has its own priorities.

leaf node The lowest node in an index. The leaf node contains either pointers to the row data or the data itself (in the case of a clustered index).

level 1 (L1) cache The memory built into a CPU chip that is used for holding data and instructions to speed access from the CPU. This cache usually stores 16 KB or 32 KB of data.

level 2 (L2) cache The memory that is either built into or external to a CPU chip and that is used as an overflow for the level 1 cache.

lock An object that is used to allow only one thread to access a resource at a time. Locking is quite important, especially in an SMP system, because many threads might be trying to access resources simultaneously.

log See transaction log.

logical disk drive A virtual disk drive that appears to the operating system as a physical disk drive but that, in reality, is made up of two or more disk drives in a RAID system.

log writer The SQL Server thread that is responsible for reading the log buffer and writing the contents out to the transaction log.

M

medium (plural: media) An object to which data is written. Both physical disk drives and tapes are media.

megabyte (MB) A unit of data equal to 2²⁰ (1,048,576) bytes or 1024 kilobytes.

memory A term used to describe random access memory (RAM) that is allocated and used by the operating system and SQL Server. The term "memory" in this book refers to main memory. The main use of SQL Server memory is as a database cache. Memory is not durable; its contents are lost upon loss of power.

merge replication A replication scheme that has been designed for multidirectional replication. With merge replication it is possible to update data on both publishers and subscribers without the use of MS DTC.

Microsoft Cluster Services (MSCS) A SQL Server add-in program that allows a SQL Server system to operate in standby mode for the primary server. In the event that the primary server fails, the secondary server takes over.

Microsoft Distributed Transaction Coordinator (MS DTC) The Microsoft SQL Server component that coordinates transactions between systems by performing two-phase commits.

Microsoft Transaction Server (MTS) A framework for developing three-tiered distributed applications.

mirroring A fault tolerance technique in which a duplicate of a component is created, thus providing a redundant copy that can be used in the event that the original fails. Mirroring is usually associated with I/O subsystems. Both RAID 1 and RAID 10 use mirroring.

MSCS See Microsoft Cluster Services.

MS DTC See Microsoft Distributed Transaction Coordinator.

MTS See Microsoft Transaction Server.

N

named pipes A SQL Server net-library. Named pipes is required for SQL Server/Windows NT installations. Named pipes is a secure net-library and supports several underlying network protocols, such as IPX/SPX, NetBEUI, and TCP/IP.

network packet A unit of data, including control information, that is transferred across the network as a whole. A SQL Server request might fit into one network packet, or multiple network packets might be required.

NWLINK The SQL Server network library that supports Novell's IPX/SPX network protocol.

O

ODBC See Open Database Connectivity.

OLAP See online analytical processing.

OLAP Services See SQL Server Analysis Services.

OLTP See online transaction processing.

online analytical processing (OLAP) The manipulation of data for analytical purposes. OLAP is usually associated with data marts or data warehouses.

online transaction processing (OLTP) A system of database processing in which many users access various data in real time. Because users are waiting for processing to be completed, response time is critical for OLTP systems.

Open Database Connectivity (ODBC) A Microsoft-designed database connectivity API that can be used to communicate from applications to a multitude of different RDBMS's.

Optimizer See SQL Server Query Optimizer.

ORDER BY A clause that specifies the sort order of columns returned as the result of a SELECT operation.

OSQL An application supplied by Microsoft that allows access into the SQL Server RDBMS. OSQL uses the ODBC protocol; ISQL uses the DB-LIB protocol.

outer join An operation that returns all rows from at least one of the tables or views mentioned in a FROM clause, as long as those rows meet any WHERE or HAVING search conditions.

P

packet See network packet.

page The fundamental unit of data storage in SQL Server. A page is the smallest unit of data that is written to or read from disk or memory. The page size for Microsoft SQL Server 2000 is 8 KB; it was 8 KB for Microsoft SQL Server 7 and 2 KB for Microsoft SQL Server 6.5.

page lock See lock.

paging A process that occurs when the operating system and SQL Server use more memory than is physically available and which involves the temporary copying of data to disk. When paging occurs, the performance of the system is severely degraded. Another term for "paging" is "swapping."

parity The extra data (or pseudodata) that is used to validate or correct the base data. Typically, "parity" refers to a bit that is used to force the underlying data bits to be either even or odd. By checking the data in order to determine if the data is even or odd, the system can validate that it is correct. RAID 5 uses parity to protect its data.

parse A process performed by SQL Server to break down an SQL statement into its fundamental components before passing it to Query Optimizer.

partitioning The process of dividing a table or database into separate components in order to create smaller and more easily managed data sets.

peak utilization period The time a computer is most used during a measurement of utilization or a working day.

performance counter An object that reflects resource usage.

performance measurement A collection of performance statistics (called performance counters) that is used to monitor and tune a computer system. This type of measurement lasts a short time (usually seconds or minutes, but sometimes hours) in order to capture system performance anomalies and correct them.

Performance Monitor (perfmon) A utility that is included with the Windows NT and Microsoft Windows 2000 operating systems that allows you to look at various performance counters within the operating system and SQL Server.

physical memory The chips that are used for memory within the system.

predictive analysis The use of mathematics in order to predict the outcome of a specific change to the system. Predictive analysis can be used to predict the effect of adding users or resources to the system.

primary key A column or combination of columns whose values uniquely identify each row in the table. The column or columns are used to enforce the data integrity of the table.

procedure cache The part of SQL Server memory used for caching stored procedures and parsed SQL statements. The procedure cache is used to improve SQL Server performance.

processor The central processing unit (CPU); the brains of the computer. A standard computer system consists of one or more CPUs, main memory, and disk storage.

Profiler See SQL Server Profiler.

publication A group of articles that is replicated from a publisher.

publisher The SQL Server system that is replicating its data to other systems. The recipient of the replication data is called a subscriber.

pull subscription A form of replication in which the replication is initiated by the subscriber.

push subscription A form of replication in which the replication is initiated by the publisher or distributor.

Q

query An SQL SELECT statement that is used to retrieve data from the database. In this book, "query" refers to a read-only operation.

Query Analyzer See SQL Server Query Analyzer.

Query Optimizer See SQL Server Query Optimizer.

queue The list of processes, or threads, that are awaiting processing.

queuing The act of waiting in line for processing. Queuing occurs in SQL Server, in the operating system, and in hardware.

R

RAID (redundant array of independent disks) A disk array that is used to form one large logical disk drive. Data is partitioned, or distributed, across the physical disk drives. RAID controllers can be configured in various ways. Each type of RAID configuration offers different fault tolerance and performance characteristics.

RAID level The identifier that is used to specify which RAID configuration is being used.

RAID 0 The RAID level that provides data striping. This is the most economical and the fastest RAID level, but it provides no tolerance for disk failures.

RAID 1 The RAID level commonly known as mirroring. RAID 1 volumes consist of two identical disk drives.

RAID 5 The RAID level known as distributed parity RAID. This RAID level creates a parity for each stripe. It is economical and provides fault tolerance.

RAID 10 The RAID level that combines both disk mirroring and disk striping. RAID 10 is sometimes referred to as RAID 0+1 or RAID 1/0.

random access memory (RAM) Nonpersistent storage that is used for data processing. RAM is faster than a disk, but all data is lost when the power is turned off or fails.

raw device A method of accessing a disk drive through a raw interface; that is, by bypassing the file system.

RDBMS See relational database management system.

record A single row in the database.

recovery An operation in which SQL Server is restarted after a system failure; the transaction log is used to roll forward all committed transactions and roll back all uncommitted transactions in order to bring the database to the state it was in at the point of failure.

recovery interval The amount of time SQL Server needs to recover from a system failure.

referential integrity See integrity.

relational database management system (RDBMS) A database system that stores data according to the relational data model. Data is arranged in a hierarchy that reflects on the relations between objects. SQL Server is one example of an RDBMS.

replication A feature of SQL Server that allows you to enable the automatic creation of copies of SQL Server objects or subsets of objects on a system and the propagation of the objects to other systems. Replication comes in three forms: snapshot, transactional, and merge.

response time The length of time between the submission of a request for data (the execution of a transaction) by a user and the returning of that data to the user. Often, the response time is used to judge the performance of the system.

restore An operation in which a SQL Server backup file is copied back into the SQL Server database. This operation takes the database back to the state it was in when the backup was created.

roles See database role.

ROLLBACK The SQL Server statement used to undo a transaction. Prior to committing a transaction, you can issue a ROLLBACK statement that will undo all the activity performed by that transaction.

root node The top node in an index.

rotational latency The amount of time a disk drive requires to rotate to the position at which the requested data resides.

router A network device that passes data from one subnet to another according to the network addresses.

row A single record in the database. A row, or record, of data represents one entry in the database. This entry consists of multiple pieces of data known as columns, or fields.

row lock See lock.

S

SAN See Storage Area Network.

schema A collection of objects associated with a database. Schema objects include tables, indexes, views, and so on.

SCSI See Small Computer System Interface.

seek time The amount of time disk heads require to move to a track where desired data resides.

selectivity The ability of an index to identify objects. An index with few unique values is said to have poor selectivity. A unique index has excellent selectivity.

self-join A join of a table with itself.

service level agreement (SLA) A contract between the provider of computer or database services and the user of those services. This agreement specifies the minimal level of service that is guaranteed, usually by specifying maximum response times for certain transactions.

sizing The task of determining the proper amount of hardware (CPUs, memory, disks, and so forth) for a computer system. A system needs to be sized before it is designed and built. After the system has been in production, capacity planning is performed to plan for future growth.

SLA See service level agreement.

Small Computer System Interface (SCSI) An I/O interface that is highly popular in today's computer systems. SCSI disks are disks that use the SCSI interface.

SMP See symmetric multiprocessor (SMP) system.

SMS See Systems Management Server.

snapshot replication A form of replication in which the entire publication is periodically copied from the publisher to the subscriber.

snowflake schema A schema involving dimension tables that are joined with other dimension tables before being joined with the fact table. Several layers of dimension tables can be created before the dimension tables are joined to the fact table. If you were to diagram this arrangement, you would see that the shape of the diagram resembles the shape of a snowflake.

SPID See system process ID.

split seek A feature of RAID 1 and RAID 10, in which both disks in a mirror can simultaneously seek for data.

SQL See Structured Query Language.

SQL Profiler A SQL Server utility that is used to monitor server performance and activity. SQL Server Profiler is quite useful for tracking events within SQL Server.

SQL Query Analyzer SQL Query Analyzer has replaced ISQL/W as the tool for ad hoc SQL Server access. In the Query Analyzer window, you can type an SQL statement and view the results of the statement in order to debug it. Query Analyzer also displays the execution plan of an SQL statement, allowing you to tune the statement.

SQL Server Microsoft's RDBMS product.

SQL Server Agent A program that performs background tasks, such as scheduling SQL Server jobs and notifying the appropriate persons of problems within SQL Server. The SQL Server Agent Scheduler is used for the execution of other agents, such as the replication agents. SQL Server Agent was known as SQLExecutive in SQL Server 6.5.

SQL Server Analysis Services A SQL Server 2000 add-in program designed to assist you with online analytical processing (OLAP); you can use it to access data in your data warehouses and data marts. In SQL Server 7, this tool was called OLAP Services.

SQL Server Query Optimizer An internal component of SQL Server that analyzes SQL Server statistics and object statistics to determine the optimal execution plan for a query. Users do not access Query Optimizer; instead, the parser passes parsed SQL statements to Query Optimizer.

staging table A temporary table in which to insert data so that it can then be extracted or transformed and copied into permanent tables. Using staging tables is a common way of transforming data within a database.

star schema A single fact table surrounded by dimension tables. The dimension tables are used to form the basis of the analysis of the data in the fact table. Each of the dimension tables is joined to a column in the fact table. If you imagine a fact table surrounded by dimension tables, you can see how the structure resembles the shape of a star. A star schema is quite common for data warehouses.

steady state The average utilization factor of a computer for either a measurement period or a working day.

Storage Area Network (SAN) An I/O subsystem in which multiple computer systems can share a RAID subsystem.

stored procedure One or more T-SQL statements compiled into a single execution plan. This compiled plan is stored in the SQL Server database.

striping The spreading of data, in equal chunks, among two or more disk drives. With RAID controllers, data is distributed in equal pieces among all the drives in the logical volume.

Structured Query Language (SQL) A language commonly used with relational databases. Both the American National Standards Institute (ANSI) and the International Organization for Standardization (ISO) define standards for SQL. Most modern DBMS products support the Entry Level of SQL-92, the latest SQL standard (published in 1992).

subscriber The recipient of SQL Server replication data.

swapping See paging.

switch A network device in which one connector is electronically connected to another via a path determined by network addresses. This differs from a hub, in which all connectors see all traffic, and from a router, in which packets are modified.

symmetric multiprocessor (SMP) system A system with multiple CPUs that all share memory and that function equivalently. Typically, each CPU in an SMP system has cache memory but no main memory associated with it.

system administrator The person or persons responsible for maintaining a company's computer hardware and operating system. The system administrator's role can vary depending on how a company is organized.

system process ID (SPID) A unique integer (smallint) assigned to each user connection when the connection is made. The assignment is not permanent.

Systems Management Server (SMS) Microsoft's enterprise management platform.

T

table lock See lock.

table scan An operation in which all rows in the table are read before the desired data is found. Scanning an entire table is not usually a desired way of performing a query.

TCP/IP See Transmission Control Protocol/Internet Protocol.

terabyte (TB) A unit of data equal to 2⁴⁰ (1,099,511,627,776) bytes or 1024 gigabytes.

thread A feature of the Windows NT and Windows 2000 operating systems that allows application logic to be separated into several concurrent execution paths. All threads are part of the same process and share memory.

throughput The transmission capacity of a data conduit such as a network or CPU bus. The term is used to describe how much data can pass through the conduit in a particular time.

token ring A popular network hardware protocol developed by IBM.

transaction A set of SQL statements that ends with either a COMMIT or a ROLLBACK statement. The term "transaction" is usually used to describe statements that modify data, and the term "query" is usually used to describe SELECT (read-only) statements.

transactional replication A form of replication that operates by duplicating on the subscriber each transaction that has run on the publisher.

transaction log A file in the database that is used to record all modifications to the database. Information about the transactions that perform modifications is also kept in the transaction log. The transaction log stores enough information to undo a transaction (in case a rollback is needed) or to redo a transaction (in case the transaction needs to be rolled forward).

Transact-SQL (T-SQL) The procedural SQL language that is used by SQL Server. Transact-SQL is a superset of SQL.

Transmission Control Protocol/Internet Protocol (TCP/IP) A popular network protocol.

trigger A special type of stored procedure that fires, or runs, automatically whenever a predefined event occurs. This event can be the execution of an UPDATE, INSERT, or DELETE statement on a table.

truncate An operation that removes all rows from an object without deleting the object itself. You can truncate a table by using the TRUNCATE TABLE command. The transaction log can also be truncated.

two-phase commit A protocol used to coordinate a transaction between two independent systems by guaranteeing an all-or-nothing approach to the commit. The two-phase commit splits the commit operation into two parts. The first part is the prepare phase; the second part is the commit phase. These phases are initiated by a commit command from the application.

U

UNION An SQL statement used to combine the results of two or more queries into a single result set consisting of all the rows that are returned from all the queries in the UNION statement.

unique A term used to describe a value of which no duplicate exists. You can create either a constraint or a unique index to enable SQL Server to enforce uniqueness.

Universal Naming Convention (UNC) name A full Windows NT/2000 name of a resource on the network. The UNC name conforms to this syntax: \\servername\sharename. Here servername is the name of the network server, and sharename is the name of the shared resource. The UNC name also usually includes the directory path, as follows: \\servername\sharename\directoryname\filename.

UPDATE The SQL statement used to modify a record in the database.

V

view An auxiliary data structure that is used to create a virtual table. This virtual table is defined by an SQL statement that makes the table appear as either a subset or superset of the underlying objects.

VINES The Banyan network protocol.

virtual memory A memory management scheme that allows Windows NT/2000 to access more memory than is actually present in the system through the use of paging and swapping. At any time, only a portion of the virtual memory is actually in physical memory.

W

WHERE A clause in an SQL statement that you use to define the conditions for the data that you want to retrieve.

wizard A user-friendly tool provided by Microsoft in order to make system administration tasks and configuration tasks easier.

working set The amount of memory that is currently being used by a process.