How to speed up table lookup?

Question

MySQL

There is a MySQL table

CREATE TABLE IF NOT EXISTS `users` (
  `IDClient` int(11) NOT NULL,
  `Phone` varchar(15) NOT NULL,
  `passw` varchar(32) NOT NULL,
  `ip` varchar(50) DEFAULT NULL COMMENT 'IP адрес'
) ENGINE=InnoDB DEFAULT CHARSET=utf8
/*!50100 PARTITION BY KEY (Phone)
PARTITIONS 10 */;
ALTER TABLE `users`
  ADD PRIMARY KEY (`Phone`),
  ADD KEY `IDClient` (`IDClient`),
  ADD KEY `ip` (`ip`);

It has about 3 million records. When the site is not loaded search

SELECT * FROM users WHERE phone='9123456789' LIMIT 1

takes about 0.03 seconds, but when the site is reloading this time can reach up to 5 seconds.
Who in the subject, can you advise something? After all, even 0.03 seconds, IMHO, is a lot for such a banal request.
UPD. MySQL server
Server: Localhost via UNIX socket
Server type: Percona Server Server
version: 5.6.26-74.0-log - Percona Server (GPL), Release 74.0, Revision 32f8dfd
Explain

id     select_type  table     type          possible_keys       key             key_len          ref         rows     Extra
1      SIMPLE        users    range         PRIMARY             PRIMARY       47               NULL     1 	 NULL

Maybe a real partition of the table into 2,5,10 tables will help?

Answer 1

Use caching. Memkeshed or radish will save the father of Russian democracy. And so, well, you can tune your muscle "according to the photograph", but there is zero sense in this.

Answer 2

What is the server doing during prime time?
big iowait?
What is the type of the Phone field?
line? And you need a number.

Answer 3

Using where; Using filesort

filesort doesn't bother you?
An intimate question - why do you need Limit 1? Do you have several identical phones? If so, what is the point of such a request?

Answer 4

`Phone` varchar(15) NOT NULL
If space on the server allows, it is better to change to
Since in this case it will be indexed by a fixed length key (in the case of int it is even better, it will be indexed by 4/8 bytes instead of 15 ) - accordingly, the search will be faster.