Scribd
Upload
82 views

L16: Logging: Frans Kaashoek & Nickolai Zeldovich 6.033 Spring 2012

This document discusses various techniques for logging transactions in a database including: 1) Using shadow copies and renaming to implement all-or-nothing transactions. 2) Introducing transaction terminology like "begin", "commit", and "abort" and checking for sufficient funds before committing. 3) Reading from a log of committed transactions to return the most recent value for a variable.

Uploaded by

Steven Swafford
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
82 views

L16: Logging: Frans Kaashoek & Nickolai Zeldovich 6.033 Spring 2012

This document discusses various techniques for logging transactions in a database including: 1) Using shadow copies and renaming to implement all-or-nothing transactions. 2) Introducing transaction terminology like "begin", "commit", and "abort" and checking for sufficient funds before committing. 3) Reading from a log of committed transactions to return the most recent value for a variable.

Uploaded by

Steven Swafford
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

L16: Logging

Frans Kaashoek & Nickolai Zeldovich 6.033 Spring 2012

All-or-nothing using shadow copy


xfer(bank, a, b, amt): copy(bank, tmp) tmp[a] = tmp[a] amt tmp[b] = tmp[b] + amt rename(tmp, bank)

Shadow copy abort vs. commit


xfer(bank, a, b, amt): copy(bank, tmp) tmp[a] = tmp[a] amt tmp[b] = tmp[b] + amt if tmp[a] < 0: print Not enough funds unlink(tmp) else: rename(tmp, bank)

Transaction terminology
xfer(bank, a, b, amt): begin bank[a] = bank[a] amt bank[b] = bank[b] + amt if bank[a] < 0: print Not enough funds abort else: commit

Read with a log


read(log, var): commits = { } for record r in log[len(log)-1] .. log[0]: if (r.type == commit): commits = commits + r.tid if (r.type == update) and (r.tid in commits) and (r.var == var): return r.new_val

Read / write with cell storage


read(var): return cell_read(var) write(var, value): log.append(cur_tid, update, var, read(var), value) cell_write(var, value)

Recovering cell storage from log


recover(log): done = { } for record r in log[len(log)-1] .. log[0]: if r.type == commit or r.type == abort: done = done + r.tid if r.type == update and r.tid not in done: cell_write(r.var, r.old_val) # undo

Recovering cell storage from log


recover(log): done = { }, aborted = { } for record r in log[len(log)-1] .. log[0]: if r.type == commit or r.type == abort: done = done + r.tid if r.type == update and r.tid not in done: cell_write(r.var, r.old_val) # undo aborted = aborted + r.tid for tid in aborted: log.append(tid, abort)

Cached read / write


read(var): if var not in cache: # may evict others from cache to cell store cache[var] = cell_read(var) return cache[var] write(var, value): log.append(cur_tid, update, var, read(var), value) cache[var] = value

Recovery for cached writes


recover(log): done = { } for record r in log[len(log)-1] .. log[0]: if r.type == commit or r.type == abort: done = done + r.tid if r.type == update and r.tid not in done: cell_write(r.var, r.old_val) # undo for record r in log[0] .. log[len(log)-1]: if r.type == update and r.tid in done: cell_write(r.var, r.new_val) # redo

You might also like