MySQL Cluster slower than MySQL Memory Engine? - mysql

Question

I spent the last days studying alternatives to update my database , I have studied databases as MemSQL , VoltDB , and many other databases. I have a specific problem , in which case I have real-time data updates , and need to show this data quickly to my user. However I can not migrate to NoSQL solutions for the needs of Joins , ACID , Group By , among others.
I'm currently using the following structure:
1 Server 64 GB Ram 1600 Mhz / Xeon E5 3.7 Ghz / 160 GB SSD
MySQL InnoDB engine to save the data and not lose them and MySQL Memory Engine to perform the Selects in real time updated via Trigger.
This structure is good , but do not increase much my performace and my amount of records is increasing too .
From that, I started to test other database, looking for a solution that could scale and found MySQL Cluster . I have tested MySQL Cluster the last days and the performace it is far below the MySQL Memory Engine, comparative 4-5x slower for some queries .
Have my infrastructure for MySQL Cluster are 4 machines with the same configuration above , one NDB Manager , MySQL SQL Node 1 , Node 2 Date ( 1 Node Group with replication) . In this case I have a super hardware, because as I said , I have a major concern with real time.
A case I'm testing is as follows , I have a table with 5 million records , with one column I need to perform a " ( count (distinct (column ) ) " are using Index BTREE . My query is basically the following
MySQL Memory Engine :
select sql_no_cache COUNT(DISTINCT(id)) from my_data;
/* Affected rows: 0 Registros encontrados: 1 Avisos: 0 Duração de 1 query: 0,109 sec. */
MySQL Cluster :
select sql_no_cache COUNT(DISTINCT(id)) from my_data;
/* Affected rows: 0 Registros encontrados: 1 Avisos: 0 Duração de 1 query: 0,485 sec. */
For MySQL Memory Engine, i have a my.cnf boosted , some values are bigger than I need , but it is working as well , see below:
[mysqld]
socket = /var/run/mysqld/mysqld.sock
basedir = /usr
datadir = /var/lib/mysql
tmpdir = /tmp
lc-messages-dir = /usr/share/mysql
skip-external-locking
# * Fine Tuning
key_buffer = 16G
max_allowed_packet = 4G
table_open_cache = 16384
sort_buffer_size = 8G
read_buffer_size = 8G
read_rnd_buffer_size = 16G
myisam_sort_buffer_size = 16G
thread_cache_size = 256
thread_concurrency = 128
thread_stack = 8192K
myisam-recover = BACKUP
max_connections = 5000
# * Query Cache Configuration
query_cache_limit = 2G
query_cache_size = 8G
tmp_table_size = 16G
max_heap_table_size = 32G
# * Logging and Replication
general_log_file = /var/log/mysql/mysql.log
general_log = 1
log_error = /var/log/mysql/error.log
log_slow_queries = /var/log/mysql/mysql-slow.log
long_query_time = 1
log-queries-not-using-indexes
log_bin = /var/log/mysql/mysql-bin.log
expire_logs_days = 10
max_binlog_size = 100M
# INNODB
innodb_buffer_pool_size = 48G
innodb_additional_mem_pool_size = 32G
innodb_file_per_table = 1
innodb_file_format = barracuda
[mysqldump]
quick
quote-names
max_allowed_packet = 8G
[mysql]
#no-auto-rehash # faster start of mysql but no tab completition
[isamchk]
key_buffer = 16G
[myisamchk]
key_buffer_size = 24G
sort_buffer_size = 24G
read_buffer = 16G
write_buffer = 16G
As for MySQL Cluster , the Config.ini file is as follows:
[ndbd default]
# Options affecting ndbd processes on all data nodes:
NoOfReplicas=2 # Number of replicas
DataMemory=32G # How much memory to allocate for data storage
IndexMemory=16G # How much memory to allocate for index storage
# For DataMemory and IndexMemory, we have used the
# default values. Since the "world" database takes up
# only about 500KB, this should be more than enough for
# this example Cluster setup.
MaxNoOfTables=4096
MaxNoOfTriggers=3500
MaxNoOfAttributes=10000
MaxNoOfOrderedIndexes=1000
MaxNoOfUniqueHashIndexes=1000
MaxNoOfConcurrentIndexOperations=32K
MaxNoOfConcurrentOperations=128K
SharedGlobalMemory=1G
DiskPageBufferMemory=1G
[ndb_mgmd]
NodeId=20
# Management process options:
hostname=192.168.0.1 # Hostname or IP address of MGM node
datadir=/var/lib/mysql-cluster # Directory for MGM node log files
[ndbd]
#Data Node 1
NodeId=1
NodeGroup=0
hostname=192.168.0.2 # Hostname or IP address
datadir=/usr/local/mysql/data # Directory for this data nodes data files
[ndbd]
#Data Node 2
NodeId=2
NodeGroup=0
hostname=192.168.0.3 # Hostname or IP address
datadir=/usr/local/mysql/data # Directory for this data nodes data files
[mysqld]
# SQL Node 1
NodeID=11
hostname=192.168.0.4 # Hostname or IP address
So what's wrong for MySQL Cluster not have the same performace MySQL Memory Engine?