Camera ready version.

doc/paper3/LLADD.tex

@@ -762,28 +762,42 @@ set of byte ranges with their new values.  Note that we still allow
 code that invokes operations to examine the page file, just not during
 the redo phase of recovery.
 
-Recovery works the same way as before, except that it now computes
-a lower bound for the LSN of each page, rather than reading it from the page.
-One possible lower bound is the LSN of the most recent checkpoint.  
-Alternatively, \yad could occasionally store its list of dirty pages 
-and their LSNs to the log (Figure~\ref{fig:lsn-estimation}).  
+Recovery works the same way as before, except that it now estimates
+page LSNs rather than reading them from pages.  One safe estimate
+is the LSN of the most recent archive or log truncation point.
+Alternatively, \yad could occasionally store its {\em dirty page
+table} to the log (Figure~\ref{fig:lsn-estimation}).  The dirty page
+table lists all dirty pages and their {\em recovery LSNs}.  It
+provides a lower bound of the LSN of the first log entry that must be
+applied to each page in the page file.  It is used by ARIES to reduce
+the amount of work that must be performed during REDO.
 
-Each dirty list is an
-accurate sparse representation of the LSNs of the entire page file.
-If a page is present in the most recent list of dirty pages then we use
-the LSN in the list as our estimate.  If the page is not in the list then
-the page was not updated between the most recent update to the on-disk
-version (the ``true'' LSN of the page), and the point at which the
-list was written to log.  Therefore, we use the LSN of the log entry that contains the list.  
-The buffer pool must maintain the dirty list whether or not LSN-free
-pages are in use, so we expect the runtime overhead to be minimal.
+The recovery LSN (RecLSN) is the LSN of the log entry 
+that caused a clean (up-to-date on disk) page to become dirty.  No log
+entries older than the RecLSN need to be applied to a page during
+redo.  Therefore, redo can safely estimate the page LSN with any number less than RecLSN.
+If a page is not in the table, redo can use the LSN of the log
+entry that contains the table, since the page must have been clean
+when the log entry was produced.  \yad writes the dirty page table to
+log whether or not LSN-free pages are in use, so we expect the runtime
+overhead to be minimal.
+
+%Each dirty list is an
+%accurate sparse representation of the LSNs of the entire page file.
+%If a page is present in the most recent list of dirty pages then we use
+%the LSN in the list as our estimate.  If the page is not in the list then
+%the page was not updated between the most recent update to the on-disk
+%version (the ``true'' LSN of the page), and the point at which the
+%list was written to log.  Therefore, we use the LSN of the log entry that contains the list.  
+%The buffer pool must maintain the dirty list whether or not LSN-free
+%pages are in use, so we expect the runtime overhead to be minimal.
 
 \begin{figure}
 \includegraphics[%
   viewport=0bp 0bp 460bp 225bp,
   clip,
   width=1\columnwidth]{figs/lsn-estimation.pdf}
-\caption{\label{fig:lsn-estimation}LSN estimation.  If a page was not mentioned in the log, it must have been up-to-date on disk.}
+\caption{\label{fig:lsn-estimation}LSN estimation.  If a page was not mentioned in the log, it must have been up-to-date on disk.  Here, RecLSN is the LSN of the entry that caused the page to become dirty.  Subtracting one yields a safe estimate of the page LSN.}
 \vspace{-12pt}
 \end{figure}
 
@@ -1564,17 +1578,20 @@ concurrently with their children~\cite{omtt}.
 QuickSilver is a distributed transactional operating system.  It
 provides a transactional IPC mechanism, and
 allows varying degrees of isolation, both to support legacy code, and
-to implement servers that require special isolation properties.  
+to provide an appropriate environment for custom transactional software~\cite{recoveryInQuickSilver}.
+By providing an environment that allows multiple, 
+independently written, transactional programs to interoperate, QuickSilver
+would complement \yad nicely.
 %It
 %supports transactions over durable and volatile state, and includes a
 %number of different commit protocols.  
-Interestingly, its shared logging facility does not
-hard-code log format or recovery algorithms, and supports a number
-of interesting optimizations such as distributed
-logging~\cite{recoveryInQuickSilver}.  QuickSilver's logging mechanism 
-is general enough to support \yad.  
+%Interestingly, its shared logging facility does not
+%hard-code log format or recovery algorithms, and supports a number
+%of interesting optimizations such as distributed
+%logging~\cite{recoveryInQuickSilver}.  QuickSilver's logging mechanism 
+%is general enough to support \yad.  
 
-The QuickSilver project showed that transactions meet the demands of most
+The QuickSilver project showed that transactions can meet the demands of most
 applications, provided that long-running transactions do not exhaust
 system resources, and that flexible concurrency control policies are
 available.  Nested transactions are