Mysql is using system file cache, not RAM directly - mysql

Question

I have an Ubuntu 64-bit system and running Mysql version 5.5.32-0ubuntu0.12.04.1-log. The machine has 64 GB ram.
I have a pretty huge database which almost has 90 GB of data and one table with 250.000.000 rows.
My problem is, that the mysqld process does not take much memory. Just between 2-5 GB. But the memory of the system is fully used for the file cache.
I restarted the server, there were 63 GB memory free. Once I started to execute larger queries, the amount of free memory decreases, while the amount of file cached increases.
So mysql is not loading the data in its RAM, but it uses the file cache.
Queries are very slow. When I query the big table, which - according to explain - uses an index and only has 500.000 rows to consider, it takes minutes.
Do you have any explanation for this phenomen?
Here is my my.cnf. (This is actually the config of my slave, but it shows exactly the same behaviour) My tables are all InnoDB:
[client]
port = 3306
socket = /var/run/mysqld/mysqld.sock
[mysqld_safe]
socket = /var/run/mysqld/mysqld.sock
nice = 0
[mysqld]
user = mysql
pid-file = /var/run/mysqld/mysqld.pid
socket = /var/run/mysqld/mysqld.sock
port = 3306
basedir = /usr
datadir = /var/lib/mysql
tmpdir = /tmp
lc-messages-dir = /usr/share/mysql
skip-external-locking
bind-address = 127.0.0.1
key_buffer = 16M
max_allowed_packet = 16M
thread_stack = 192K
thread_cache_size = 8
myisam-recover = BACKUP
query_cache_limit = 1M
query_cache_size = 16M
log_error = /var/log/mysql/error.log
server-id = 2
log_bin = /var/log/mysql/mysql-bin.log
expire_logs_days = 10
max_binlog_size = 100M
[innodb]
innodb_buffer_pool_size = 55G
innodb_log_file_size = 256M
innodb_log_buffer_size = 4M
innodb_flush_log_at_trx_commit=2
innodb_thread_concurrency=0
innodb_file_per_table = 1 # (this is not respected as I changed it after the first database was created). So my ibdata1 file is over 80GB in size
max_connections = 1000 # wenn gleichzeitig viele User zugreifen
table_cache = 10000 # abgeleitet von max_connections, könnte sogar höher sein
tmp_table_size = 5000M # wichtig wg. JOINs, zu kleiner Wert macht die extrem langsam
max_heap_table_size = 5000M
[mysqldump]
quick
quote-names
max_allowed_packet = 16M
[mysql]
[isamchk]
key_buffer = 16M
!includedir /etc/mysql/conf.d/

Answer 1

Increase the key_buffer size to 128M and max_allowed_packet to 128M, as this will speed up the execution of queries

Answer 2

I finally figured out was the problem was. The
innodb_buffer_pool_size = 55G
does not belong in the [innodb] section, but in the [mysqld] section.
All the innodb configs needed to be placed in the mysqld section.