greg/stasis-aries-wal
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Tweaks from "final pass".

Browse source
This commit is contained in:
Sears Russell 2006-09-05 21:49:32 +00:00
parent 73f67a2a62
commit decf605332
1 changed files with 11 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

22
doc/paper3/LLADD.tex
View file

@ -338,7 +338,7 @@ extensions and novel variants, which we cover in the next two
sections. sections.
As with other systems, \yads transactions have a multi-level As with other systems, \yads transactions have a multi-level
structure. Multi-layered transactions were originally proposed as an structure. Multi-layered transactions were originally proposed as a
concurrency control strategy for database servers that support high concurrency control strategy for database servers that support high
level, application specific extensions~\cite{multiLayeredSystems}. level, application specific extensions~\cite{multiLayeredSystems}.
In \yad, the lower level of an operation provides atomic updates to regions of In \yad, the lower level of an operation provides atomic updates to regions of
@ -359,7 +359,7 @@ updates bootstrap transactions that are too large to be
applied atomically. In particular, write-ahead logging (and therefore applied atomically. In particular, write-ahead logging (and therefore
\yad) relies on the ability to write entries to the log \yad) relies on the ability to write entries to the log
file atomically. Transaction systems that store sequence numbers on pages to file atomically. Transaction systems that store sequence numbers on pages to
track version information also rely on atomic page writes. track versions rely on atomic page writes in addition to atomic log writes.
In practice, a write to a disk page is not atomic (in modern drives). Two common failure In practice, a write to a disk page is not atomic (in modern drives). Two common failure
modes exist. The first occurs when the disk writes a partial sector modes exist. The first occurs when the disk writes a partial sector
@ -406,7 +406,7 @@ but no logging is required.
This approach performs poorly because we {\em force} the page to disk This approach performs poorly because we {\em force} the page to disk
on commit, which leads to a large number of synchronous non-sequential on commit, which leads to a large number of synchronous non-sequential
writes. By writing redo information to the log before committing writes. By writing redo information to the log before committing
(write-ahead logging), we get {\em no force} transactions and better (write-ahead logging), we get {\em no-force} transactions and better
performance, since the synchronous writes to the log are sequential. performance, since the synchronous writes to the log are sequential.
Later, the pages are written out asynchronously, often Later, the pages are written out asynchronously, often
as part of a larger sequential write. as part of a larger sequential write.
@ -460,7 +460,7 @@ to build full transactions.
To recover a multi-page transaction, we simply recover each of To recover a multi-page transaction, we simply recover each of
the pages individually. This works because steal/no-force completely the pages individually. This works because steal/no-force completely
decouples the pages: any page can be written back early (steal) or decouples the pages: any page can be written back early (steal) or
late (no force). late (no-force).
\subsection{Concurrent Transactions} \subsection{Concurrent Transactions}
\label{sec:nta} \label{sec:nta}
@ -689,7 +689,7 @@ The record allocator and the region allocator each contain custom lock
management. The lock management prevents one transaction from reusing management. The lock management prevents one transaction from reusing
storage freed by another, active transaction. If this storage were storage freed by another, active transaction. If this storage were
reused and then the transaction that freed it aborted, then the reused and then the transaction that freed it aborted, then the
storage would be double allocated. storage would be double-allocated.
%If transaction A frees some storage, transaction B reuses %If transaction A frees some storage, transaction B reuses
%the storage and commits, and then transaction A aborts, then the %the storage and commits, and then transaction A aborts, then the
%storage would be double allocated. %storage would be double allocated.
@ -785,7 +785,7 @@ and their LSNs to the log (Figure~\ref{fig:lsn-estimation}).
viewport=0bp 0bp 460bp 225bp, viewport=0bp 0bp 460bp 225bp,
clip, clip,
width=1\columnwidth]{figs/lsn-estimation.pdf} width=1\columnwidth]{figs/lsn-estimation.pdf}
\caption{\label{fig:lsn-estimation}LSN estimation. If a page is not mentioned in the checkpoint, it must be up to date on disk.} \caption{\label{fig:lsn-estimation}LSN estimation. If a page was not mentioned by the checkpoint, it must have been up-to-date on disk.}
\vspace{-12pt} \vspace{-12pt}
\end{figure} \end{figure}
@ -1087,8 +1087,8 @@ reduce runtime overheads such as log bandwidth. Berkeley DB's
hash table is a popular, commonly deployed implementation, and serves hash table is a popular, commonly deployed implementation, and serves
as a baseline for our experiments. as a baseline for our experiments.
Both of our hash tables outperform Berkeley DB on a workload that bulk Both of our hash tables outperform Berkeley DB on a workload that populates
loads the tables by repeatedly inserting $(key, value)$ pairs the tables by repeatedly inserting $(key, value)$ pairs
(Figure~\ref{fig:BULK_LOAD}). (Figure~\ref{fig:BULK_LOAD}).
%although we do not wish to imply this is always the case. %although we do not wish to imply this is always the case.
%We do not claim that our partial implementation of \yad %We do not claim that our partial implementation of \yad
@ -1228,7 +1228,7 @@ Alternatively, we could arrange for the object pool
to update atomically the buffer to update atomically the buffer
manager's copy of all objects that share a given page. manager's copy of all objects that share a given page.
The third plugin variant, ``delta'', incorporates the update/flush The third plugin variant, ``delta,'' incorporates the update/flush
optimizations, but only writes changed portions of optimizations, but only writes changed portions of
objects to the log. With \yads support for custom log objects to the log. With \yads support for custom log
formats, this optimization is straightforward. formats, this optimization is straightforward.
@ -1396,7 +1396,7 @@ engines automatically.
Object-oriented database systems~\cite{objectstore} and Object-oriented database systems~\cite{objectstore} and
relational databases with support for user-definable abstract data relational databases with support for user-definable abstract data
types (such as in Postgres~\cite{postgres}) provide functionality types (such as in POSTGRES~\cite{postgres}) provide functionality
similar to extensible database toolkits. In contrast to database similar to extensible database toolkits. In contrast to database
toolkits, which leverage type information as the database server is toolkits, which leverage type information as the database server is
compiled, object-oriented and object-relational databases allow types compiled, object-oriented and object-relational databases allow types
@ -1475,7 +1475,7 @@ when the parent aborts. (We believe that recent proposals to use
open, linear nesting for software transactional memory will lead to a open, linear nesting for software transactional memory will lead to a
programming style similar to \yads~\cite{nestedTransactionPoster}.) programming style similar to \yads~\cite{nestedTransactionPoster}.)
However, logical undo gives the programmer the option to compensate However, logical undo gives the programmer the option to compensate
for nested top action. We expect that nested transactions could be for nested top actions. We expect that nested transactions could be
implemented with \yad. implemented with \yad.
\subsubsection{Distributed Programming Models} \subsubsection{Distributed Programming Models}