diff --git a/rabin/global/config.c b/rabin/global/config.c
index 9f9f058..28543ed 100644
--- a/rabin/global/config.c
+++ b/rabin/global/config.c
@@ -175,7 +175,8 @@ read_config(char *configfile, archive_config_t *cfg)
 	// Default
 	cfg->verify_chunks = 0;
 	cfg->algo = COMPRESS_LZ4;
-	cfg->chunk_cksum_type = DEFAULT_CKSUM;
+	cfg->chunk_cksum_type = DEFAULT_CHUNK_CKSUM;
+	cfg->similarity_cksum = DEFAULT_SIMILARITY_CKSUM;
 	cfg->pct_interval = DEFAULT_SIMILARITY_INTERVAL;
 
 	fh = fopen(configfile, "r");
@@ -262,11 +263,19 @@ read_config(char *configfile, archive_config_t *cfg)
 				fclose(fh);
 				return (1);
 			}
+		} else if (strncmp(line, "SIMILARITY_CKSUM") == 0) {
+			cfg->chunk_cksum_type = get_cksum_type(pos);
+			if (cfg->chunk_cksum_type == CKSUM_INVALID) {
+				fprintf(stderr, "Invalid CHUNK_CKSUM setting.\n");
+				fclose(fh);
+				return (1);
+			}
 		}
 	}
 	fclose(fh);
 	cfg->compress_level = get_compress_level(cfg->algo);
 	cfg->chunk_cksum_sz = get_cksum_sz(cfg->chunk_cksum_type);
+	cfg->similarity_cksum_sz = get_cksum_sz(cfg->similarity_cksum);
 
 	/*
 	 * Now compute the remaining parameters.
@@ -330,13 +339,15 @@ write_config(char *configfile, archive_config_t *cfg)
 }
 
 int
-set_config_s(archive_config_t *cfg, compress_algo_t algo, cksum_t ck, uint32_t chunksize,
-		  size_t file_sz, int pct_interval)
+set_config_s(archive_config_t *cfg, compress_algo_t algo, cksum_t ck, cksum_t ck_sim,
+	     uint32_t chunksize, size_t file_sz, int pct_interval)
 {
 	cfg->algo = algo;
 	cfg->chunk_cksum_type = ck;
+	cfg->similarity_cksum = ck_sim;
 	cfg->compress_level = get_compress_level(cfg->algo);
 	cfg->chunk_cksum_sz = get_cksum_sz(cfg->chunk_cksum_type);
+	cfg->similarity_cksum_sz = get_cksum_sz(cfg->similarity_cksum);
 	cfg->chunk_sz = chunksize;
 	cfg->chunk_sz_bytes = RAB_BLK_AVG_SZ(cfg->chunk_sz);
 	cfg->pct_interval = pct_interval;
diff --git a/rabin/global/config.h b/rabin/global/config.h
index 58574f8..bd3025f 100644
--- a/rabin/global/config.h
+++ b/rabin/global/config.h
@@ -29,7 +29,8 @@ extern "C" {
 #endif
 
 #define	DEFAULT_SIMILARITY_INTERVAL	5
-#define	DEFAULT_CKSUM		CKSUM_BLAKE256
+#define	DEFAULT_CHUNK_CKSUM	CKSUM_SHA256
+#define	DEFAULT_SIMILARITY_CKSUM	CKSUM_BLAKE256
 #define	DEFAULT_COMPRESS		COMPRESS_LZ4
 #define	MIN_CK 1
 #define	MAX_CK 5
@@ -44,8 +45,10 @@ typedef struct {
 	int verify_chunks; // Whether to use memcmp() to compare chunks byte for byte.
 	int algo; // Which compression algo for segments.
 	compress_algo_t compress_level; // Default preset compression level per algo.
-	cksum_t chunk_cksum_type; // Which digest to use for hash based chunk lookup.
+	cksum_t chunk_cksum_type; // Which digest to use for hash based chunk comparison.
+	cksum_t similarity_cksum; // Which digest to use similarity based segment lookup.
 	int chunk_cksum_sz; // Size of cksum in bytes.
+	int similarity_cksum_sz; // Size of cksum in bytes.
 	int pct_interval; // Similarity based match intervals in %age.
 			// The items below are computed given the above
 			// components.
@@ -60,9 +63,15 @@ typedef struct {
 	void *dbdata;
 } archive_config_t;
 
+typedef struct _segment_entry {
+	uint64_t offset;
+	uint32_t length;
+	uchar_t *cksum;
+} segment_entry_t;
+
 int read_config(char *configfile, archive_config_t *cfg);
 int write_config(char *configfile, archive_config_t *cfg);
-int set_config_s(archive_config_t *cfg, compress_algo_t algo, cksum_t ck,
+int set_config_s(archive_config_t *cfg, compress_algo_t algo, cksum_t ck, cksum_t ck_sim,
 		      uint32_t chunksize, size_t file_sz, int pct_interval);
 
 #ifdef	__cplusplus
diff --git a/rabin/global/initdb.c b/rabin/global/db.c
similarity index 81%
rename from rabin/global/initdb.c
rename to rabin/global/db.c
index 2fd548b..a61a26b 100644
--- a/rabin/global/initdb.c
+++ b/rabin/global/db.c
@@ -31,7 +31,7 @@
 #include <allocator.h>
 #include <pthread.h>
 
-#include "initdb.h"
+#include "db.h"
 #include "config.h"
 
 #define	ONE_PB (1125899906842624ULL)
@@ -43,10 +43,11 @@
  * Hashtable structures for in-memory index.
  */
 typedef struct _hash_entry {
-	uchar_t *cksum;
+	segment_entry_t *seg;
 	struct _hash_entry *next;
 	struct _hash_entry *lru_prev;
 	struct _hash_entry *lru_next;
+	uchar_t cksum[1];
 } hash_entry_t;
 
 typedef struct {
@@ -54,12 +55,13 @@ typedef struct {
 } htab_t;
 
 typedef struct {
-	htab_t *htablst;
+	htab_t *list;
 	pthread_mutex_t *mlist;
 	hash_entry_t *lru_head;
 	hash_entry_t *lru_tail;
 	uint64_t memlimit;
 	uint64_t memused;
+	int hash_entry_size;
 } htablst_t;
 
 archive_config_t *
@@ -83,14 +85,15 @@ init_global_db(char *configfile)
 
 archive_config_t *
 init_global_db_s(char *path, uint32_t chunksize, int pct_interval,
-		      compress_algo_t algo, cksum_t ck, size_t file_sz, size_t memlimit)
+		 compress_algo_t algo, cksum_t ck, cksum_t ck_sim, size_t file_sz,
+		 size_t memlimit)
 {
 	archive_config_t *cfg;
 	int rv;
 	float diff;
 
 	cfg = calloc(1, sizeof (archive_config_t));
-	rv = set_config_s(cfg, algo, ck, chunksize, file_sz, chunks_per_seg, pct_interval);
+	rv = set_config_s(cfg, algo, ck, ck_sim, chunksize, file_sz, chunks_per_seg, pct_interval);
 
 	if (path != NULL) {
 		printf("Disk based index not yet implemented.\n");
@@ -100,24 +103,25 @@ init_global_db_s(char *path, uint32_t chunksize, int pct_interval,
 		uint32_t hash_slots, intervals, i;
 		uint64_t memreqd;
 		htablst_t *htablst;
+		int hash_entry_size;
 
 		// Compute total hashtable entries first
 		intervals = 100 / pct_interval - 1;
 		hash_slots = file_sz / cfg->segment_sz_bytes + 1;
 		hash_slots *= intervals;
+		hash_entry_size = sizeof (hash_entry_t) + cfg->similarity_cksum_sz - 1;
 
 		// Compute memory required to hold all hash entries assuming worst case 50%
 		// occupancy.
-		memreqd = hash_slots * (sizeof (hash_entry_t) + cfg->chunk_cksum_sz +
-			sizeof (hash_entry_t *) + (sizeof (hash_entry_t *)) / 2);
+		memreqd = hash_slots * (hash_entry_size + sizeof (hash_entry_t *) +
+			(sizeof (hash_entry_t *)) / 2);
 		memreqd += hash_slots * sizeof (hash_entry_t **);
 		diff = (float)pct_interval / 100.0;
 
 		// Reduce hash_slots to remain within memlimit
 		while (memreqd > memlimit) {
 			hash_slots -= (hash_slots * diff);
-			memreqd = hash_slots * (sizeof (hash_entry_t) +
-					cfg->chunk_cksum_sz + sizeof (hash_entry_t *) + 
+			memreqd = hash_slots * (hash_entry_size + sizeof (hash_entry_t *) + 
 					(sizeof (hash_entry_t *)) / 2);
 			memreqd += hash_slots * sizeof (hash_entry_t **);
 		}
@@ -126,17 +130,18 @@ init_global_db_s(char *path, uint32_t chunksize, int pct_interval,
 		// each having hash_slots / intervals slots.
 		htablst = calloc(1, sizeof (htablst_t));
 		htablst->memlimit = memlimit;
-		htablst->htablst = (htab_t *)calloc(intervals, sizeof (htab_t));
+		htablst->list = (htab_t *)calloc(intervals, sizeof (htab_t));
 		htablst->mlist = (pthread_mutex_t *)malloc(intervals * sizeof (pthread_mutex_t));
+		htablst->hash_entry_size = hash_entry_size;
 
 		for (i = 0; i < intervals; i++) {
-			htablst->htablst[i].htab = (hash_entry_t **)calloc(hash_slots / intervals,
+			htablst->list[i].htab = (hash_entry_t **)calloc(hash_slots / intervals,
 							sizeof (hash_entry_t *));
 			htablst->memused += ((hash_slots / intervals) * (sizeof (hash_entry_t *)));
 			pthread_mutex_init(&(htablst->mlist[i]), NULL);
 		}
 		cfg->dbdata = htablst;
-		slab_cache_add(sizeof (hash_entry_t));
+		slab_cache_add(hash_entry_size);
 		slab_cache_add(cfg->chunk_cksum_sz);
 	}
 	return (cfg);
diff --git a/rabin/global/initdb.h b/rabin/global/db.h
similarity index 89%
rename from rabin/global/initdb.h
rename to rabin/global/db.h
index f51b968..2220c18 100644
--- a/rabin/global/initdb.h
+++ b/rabin/global/db.h
@@ -18,8 +18,8 @@
  * moinakg@belenix.org, http://moinakg.wordpress.com/
  */
 
-#ifndef	_INITDB_H
-#define	_INITDB_H
+#ifndef	_DB_H
+#define	_DB_H
 
 #ifdef	__cplusplus
 extern "C" {
@@ -27,7 +27,8 @@ extern "C" {
 
 archive_config_t *init_global_db(char *configfile);
 archive_config_t *init_global_db_s(char *path, uint32_t chunksize, int pct_interval,
-			compress_algo_t algo, cksum_t ck, size_t file_sz, size_t memlimit);
+			compress_algo_t algo, cksum_t ck, cksum_t ck_sim, size_t file_sz,
+			size_t memlimit);
 
 #ifdef	__cplusplus
 }