ovsdb: Use previous snapshot size as an additional factor for compact…

…ion.

Until now, the minimum database size before automatically compacting has
been 10 MB, regardless of the inherent size of the data in the database.
A couple of people have pointed out that this won't scale well to larger
databases.  This commit changes this criterion to 4 times the previously
compacted size of the database, with 10 MB as a minimum.

The 4x factor is suggested by Diego Ongaro's thesis, "Consensus: Bridging
Theory and Practice", section 5.1.2 "When to snapshot".

Signed-off-by: Ben Pfaff <blp@ovn.org>
Acked-by: Justin Pettit <jpettit@ovn.org>

ovsdb/file.c

@@ -1,4 +1,4 @@
-/* Copyright (c) 2009, 2010, 2011, 2012, 2013 Nicira, Inc.
+/* Copyright (c) 2009, 2010, 2011, 2012, 2013, 2016 Nicira, Inc.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -73,6 +73,7 @@ static struct ovsdb_error *ovsdb_file_create(struct ovsdb *,
                                              struct ovsdb_log *,
                                              const char *file_name,
                                              unsigned int n_transactions,
+                                             off_t snapshot_size,
                                              struct ovsdb_file **filep);
 
 /* Opens database 'file_name' and stores a pointer to the new database in
@@ -182,7 +183,6 @@ ovsdb_file_open__(const char *file_name,
                   struct ovsdb_file **filep)
 {
     enum ovsdb_log_open_mode open_mode;
-    unsigned int n_transactions;
     struct ovsdb_schema *schema = NULL;
     struct ovsdb_error *error;
     struct ovsdb_log *log;
@@ -201,7 +201,16 @@ ovsdb_file_open__(const char *file_name,
 
     db = ovsdb_create(schema ? schema : ovsdb_schema_clone(alternate_schema));
 
-    n_transactions = 0;
+    /* When a log gets big, we compact it into a new log that initially has
+     * only a single transaction that represents the entire state of the
+     * database.  Thus, we consider the first transaction in the database to be
+     * the snapshot.  We measure its size to later influence the minimum log
+     * size before compacting again.
+     *
+     * The schema precedes the snapshot in the log; we could compensate for its
+     * size, but it's just not that important. */
+    off_t snapshot_size = 0;
+    unsigned int n_transactions = 0;
     while ((error = ovsdb_log_read(log, &json)) == NULL && json) {
         struct ovsdb_txn *txn;
 
@@ -219,6 +228,10 @@ ovsdb_file_open__(const char *file_name,
             ovsdb_log_unread(log);
             break;
         }
+
+        if (n_transactions == 1) {
+            snapshot_size = ovsdb_log_get_offset(log);
+        }
     }
     if (error) {
         /* Log error but otherwise ignore it.  Probably the database just got
@@ -234,7 +247,8 @@ ovsdb_file_open__(const char *file_name,
     if (!read_only) {
         struct ovsdb_file *file;
 
-        error = ovsdb_file_create(db, log, file_name, n_transactions, &file);
+        error = ovsdb_file_create(db, log, file_name, n_transactions,
+                                  snapshot_size, &file);
         if (error) {
             goto error;
         }
@@ -500,14 +514,15 @@ struct ovsdb_file {
     long long int last_compact;
     long long int next_compact;
     unsigned int n_transactions;
+    off_t snapshot_size;
 };
 
 static const struct ovsdb_replica_class ovsdb_file_class;
 
 static struct ovsdb_error *
 ovsdb_file_create(struct ovsdb *db, struct ovsdb_log *log,
                   const char *file_name,
-                  unsigned int n_transactions,
+                  unsigned int n_transactions, off_t snapshot_size,
                   struct ovsdb_file **filep)
 {
     struct ovsdb_file *file;
@@ -532,6 +547,7 @@ ovsdb_file_create(struct ovsdb *db, struct ovsdb_log *log,
     file->file_name = abs_name;
     file->last_compact = time_msec();
     file->next_compact = file->last_compact + COMPACT_MIN_MSEC;
+    file->snapshot_size = snapshot_size;
     file->n_transactions = n_transactions;
     ovsdb_add_replica(db, &file->replica);
 
@@ -582,10 +598,13 @@ ovsdb_file_commit(struct ovsdb_replica *replica,
     /* If it has been at least COMPACT_MIN_MSEC ms since the last time we
      * compacted (or at least COMPACT_RETRY_MSEC ms since the last time we
      * tried), and if there are at least 100 transactions in the database, and
-     * if the database is at least 10 MB, then compact the database. */
+     * if the database is at least 10 MB, and the database is at least 4x the
+     * size of the previous snapshot, then compact the database. */
+    off_t log_size = ovsdb_log_get_offset(file->log);
     if (time_msec() >= file->next_compact
         && file->n_transactions >= 100
-        && ovsdb_log_get_offset(file->log) >= 10 * 1024 * 1024)
+        && log_size >= 10 * 1024 * 1024
+        && log_size / 4 >= file->snapshot_size)
     {
         error = ovsdb_file_compact(file);
         if (error) {

ovsdb/ovsdb-server.1.in

@@ -126,8 +126,11 @@ These commands are specific to \fBovsdb\-server\fR.
 Causes \fBovsdb\-server\fR to gracefully terminate.
 .IP "\fBovsdb\-server/compact\fR [\fIdb\fR]\&..."
 Compacts each database \fIdb\fR in-place.  If no \fIdb\fR is
-specified, compacts every database in-place.  Databases are also
-automatically compacted occasionally.
+specified, compacts every database in-place.  A database is also
+compacted automatically when a transaction is logged if it is over 4
+times as large as its previous compacted size (and at least 10 MB),
+but not before 100 commits have been added or 10 minutes have elapsed
+since the last compaction.
 .
 .IP "\fBovsdb\-server/reconnect\fR"
 Makes \fBovsdb\-server\fR drop all of the JSON\-RPC