From 42b353ecfd47b62d0221fa8dd1613cc5dfabc82d Mon Sep 17 00:00:00 2001
From: Kresten Krab Thorup <krab@trifork.com>
Date: Mon, 23 Jan 2012 00:49:07 +0100
Subject: [PATCH] Implement sequential/random reader API

---
 src/lsm_btree_level.erl  | 14 +++++++-------
 src/lsm_btree_merger.erl |  4 ++--
 src/lsm_btree_reader.erl | 15 +++++++++++++--
 src/lsm_btree_writer.erl |  2 +-
 4 files changed, 23 insertions(+), 12 deletions(-)

diff --git a/src/lsm_btree_level.erl b/src/lsm_btree_level.erl
index f25ab13..5d28932 100644
--- a/src/lsm_btree_level.erl
+++ b/src/lsm_btree_level.erl
@@ -119,14 +119,14 @@ initialize(State) ->
             file:delete(BFileName),
             ok = file:rename(CFileName, AFileName),
 
-            {ok, BT} = lsm_btree_reader:open(CFileName),
+            {ok, BT} = lsm_btree_reader:open(CFileName, random),
             main_loop(State#state{ a= BT, b=undefined });
 
         {error, enoent} ->
             case file:read_file_info(BFileName) of
                 {ok, _} ->
-                    {ok, BT1} = lsm_btree_reader:open(AFileName),
-                    {ok, BT2} = lsm_btree_reader:open(BFileName),
+                    {ok, BT1} = lsm_btree_reader:open(AFileName, random),
+                    {ok, BT2} = lsm_btree_reader:open(BFileName, random),
 
                     check_begin_merge_then_loop(State#state{ a=BT1, b=BT2 });
 
@@ -134,7 +134,7 @@ initialize(State) ->
 
                     case file:read_file_info(AFileName) of
                         {ok, _} ->
-                            {ok, BT1} = lsm_btree_reader:open(AFileName),
+                            {ok, BT1} = lsm_btree_reader:open(AFileName, random),
                             main_loop(State#state{ a=BT1 });
 
                         {error, enoent} ->
@@ -173,7 +173,7 @@ main_loop(State = #state{ next=Next }) ->
                     SetPos = #state.b
             end,
             ok = file:rename(FileName, ToFileName),
-            {ok, BT} = lsm_btree_reader:open(ToFileName),
+            {ok, BT} = lsm_btree_reader:open(ToFileName, random),
             reply(From, ok),
             check_begin_merge_then_loop(setelement(SetPos, State, BT));
 
@@ -276,7 +276,7 @@ main_loop(State = #state{ next=Next }) ->
             % then, rename C to A, and open it
             AFileName = filename("A",State2),
             ok = file:rename(CFileName, AFileName),
-            {ok, BT} = lsm_btree_reader:open(AFileName),
+            {ok, BT} = lsm_btree_reader:open(AFileName, random),
 
             main_loop(State2#state{ a=BT, b=undefined, merge_pid=undefined });
 
@@ -398,7 +398,7 @@ start_range_fold(FileName, WorkerPID, FromKey, ToKey) ->
     PID =
         proc_lib:spawn( fun() ->
                                 erlang:link(WorkerPID),
-                                {ok, File} = lsm_btree_reader:open(FileName),
+                                {ok, File} = lsm_btree_reader:open(FileName, sequential),
                                 do_range_fold(File, WorkerPID, self(), FromKey, ToKey),
                                 erlang:unlink(WorkerPID),
 
diff --git a/src/lsm_btree_merger.erl b/src/lsm_btree_merger.erl
index 58cab47..51af987 100644
--- a/src/lsm_btree_merger.erl
+++ b/src/lsm_btree_merger.erl
@@ -16,8 +16,8 @@
 -define(LOCAL_WRITER, true).
 
 merge(A,B,C, Size, IsLastLevel) ->
-    {ok, BT1} = lsm_btree_reader:open(A),
-    {ok, BT2} = lsm_btree_reader:open(B),
+    {ok, BT1} = lsm_btree_reader:open(A, sequential),
+    {ok, BT2} = lsm_btree_reader:open(B, sequential),
     case ?LOCAL_WRITER of
         true ->
             {ok, Out} = lsm_btree_writer:init([C, Size]);
diff --git a/src/lsm_btree_reader.erl b/src/lsm_btree_reader.erl
index b00bb9e..894d686 100644
--- a/src/lsm_btree_reader.erl
+++ b/src/lsm_btree_reader.erl
@@ -3,15 +3,26 @@
 -include_lib("kernel/include/file.hrl").
 -include("lsm_btree.hrl").
 
--export([open/1,close/1,lookup/2,fold/3,range_fold/5]).
+-export([open/1, open/2,close/1,lookup/2,fold/3,range_fold/5]).
 -export([first_node/1,next_node/1]).
 
 -record(node, { level, members=[] }).
 -record(index, {file, root, bloom}).
 
 open(Name) ->
+    open(Name, random).
 
-    {ok, File} = file:open(Name, [raw,read,read_ahead,binary]),
+%% this is how to open a btree for sequential scanning (merge, fold)
+open(Name, sequential) ->
+    {ok, File} = file:open(Name, [raw,read,{read_ahead, 1024 * 512},binary]),
+    open2(Name, File);
+
+%% this is how to open a btree for random access
+open(Name, random) ->
+    {ok, File} = file:open(Name, [raw,read,binary]),
+    open2(Name, File).
+
+open2(Name, File) ->
     {ok, FileInfo} = file:read_file_info(Name),
 
     %% read root position
diff --git a/src/lsm_btree_writer.erl b/src/lsm_btree_writer.erl
index 2998a8d..2502a13 100644
--- a/src/lsm_btree_writer.erl
+++ b/src/lsm_btree_writer.erl
@@ -56,7 +56,7 @@ init([Name,Size]) ->
 %    io:format("got name: ~p~n", [Name]),
 
     case file:open( lsm_btree_util:index_file_name(Name),
-                               [raw, exclusive, write, delayed_write]) of
+                               [raw, exclusive, write, {delayed_write, 512 * 1024, 2000}]) of
         {ok, IdxFile} ->
             {ok, BloomFilter} = ebloom:new(erlang:min(Size,16#ffffffff), 0.01, 123),
             {ok, #state{ name=Name,