Tag Archives: MySQL

My favourite DB, MySQL

MySQL 5.5 (aka superfast InnoDB) is GA

Oracle has finally announced the long time pending MySQL 5.5 as GA. The 5.5x branch has the following super improvements.

  • Super Fast & Stable Innodb: Code from Innodb plugin 1.0.4x onwards, Multiple Read and Write threads for Innodb, Native Async I/O in linux. more…
  • Semi Synchronous Replication: Ability to make sure the data committed to master has atleast reached one Slave Node: more…
  • .. and more fixes.

uuhhoooo.. !!!

MySQL makes community server edition difficult to find and download!

Read this post today, http://mysqlha.blogspot.com/2010/02/dude-where-is-my-link.html. And I was amazed to find that on mysql.com it was so hard to download the community version of MySQL Server.

On the home page, could see links for

  • Enterprise Products
  • Resources
  • Consulting
  • Training
  • News and Events

Couldn’t find a link for Community Server Download or something similar, I’ve been working with MySQL for over 5 yrs, I know that the downloads and most of the useful stuff for open source devs like me is on http://dev.mysql.com. The Developer Zone hosts the following for us

  • Community Server Downloads
  • Documentation (which keeps changing its links)
  • Articles
  • Forums
  • Beginner articles
  • MySQL forge
  • … and much more

I’m really surprised and ashamed of this MySQL’s or Ora@#$’s move to hide such a widely used link.

I’ve done my part in adding rel='nofollow' attribute to http://mysql.com

Redis, Memcached, Tokyo Tyrant and MySQL comparision (rectification skip-name-resolve)

My previous post Redis, Memcache, Tokyp Tyrant, MySQL comparison had a flaw as pointed out by this comment. The MySQL was taking a huge time for doing a reverse DNS lookup.

I turned on the skip-name-resolve parameter in the my.cnf and the Throughput of MySQL grew considerably, almost more than double.

Here are the new results.




MyISAM vs InnoDB

Nothing much has changed in the above test. Except for the fact InnoDB starts leading the way when there are high number of concurrent Inserts/Updates or Writes on the table. As seen from the “Set” graph InnoDB starts closing for MyISAM’s write efficiency around 30 concurrent requests and then by 60 concurrent requests its already ahead in throughput of writes – 1284/s against 825/s. Further I had put a watch on processlist and was watching the processess, there were times during MyISAM when the inserts took over 6seconds to finish, which also means that if you are in a need of an application which requires quicker response during heavy loads / heavy concurrency… You need to check the MyISAM vs. InnoDB scenario really closely. At low concurrency MyISAM is well ahead in writes, and in Reads, both MyISAM and InnoDB perform equally well.

Again you need to make sure that you check ur test conditions really well before just taking InnoDB for granted.

Redis, Memcached, Tokyo Tyrant and MySQL comparision

I wanted to compare the following DBs, NoSQLs and caching solutions for speed and connections. Tested the following

My test had the following criteria

  • 2 client boxes
  • All clients connecting to the server using Python
  • Used Python’s threads to create concurrency
  • Each thread made 10,000 open-close connections to the server
  • The server was
    • Intel(R) Pentium(R) D CPU 3.00GHz
    • Fedora 10 32bit
    • Intel(R) Pentium(R) D CPU 3.00GHz
    • #1 SMP
    • 1GB RAM
  • Used a md5 as key and a value that was saved
  • Created an index on the key column of the table
  • Each server had SET and GET requests as a different test at same concurrency

Results please !

Work sheet

throughput set

throughput get

I wanted to simulate a situation where I had 2 servers (clients) serving my code, which connected to the 1 server (memcached, redis, or whatever). Another thing to note was that I used Python as the client in all the tests, definately the tests would give a different output had I used PHP. Again the test was done to check how well the clients could make and break the connections to the server, and I wanted the overall throughput after making and breaking the connections. I did not monitor the response times. I didnt change absolutely any parameters for the servers, eg didn’t change the innodb_buffer_pool_size or key_buffer_size.


MySQL lacked the whole scene terribly, I monitored the MySQL server via the MySQL Administrator and found that hardly there were any conncurrent inserts or selects, I could see the unauthenticated users, which meant that the client had connected to MySQL and was doing a handshake using MySQL authentication (using username and password). As you could see I didn’t even perform the 40 and 60 thread tests.

I truncated the table before I swtiched my tests from MyISAM to InnoDB. And always started the tests from lesser threads. My table was as follows

CREATE TABLE `comp_dump` (
  `k` char(32) DEFAULT NULL,
  `v` char(32) DEFAULT NULL,
  KEY `ix_k` (`k`)


For Tokyo Tyrant I used a file.tch as the DB, which is a hash database. I also tried MongoDB as u may find if u have opened the worksheet, But the server kept failing or actually the mongod failed after coming at an unhandled Exception. I found something similar over here. I tried 1.0.1, 1.1.3 and the available Nightly build, but all failed and I lost my patience.

Now what

If you need speed just to fetch a data for a given combination or key, Redis is a solution that you need to look at. MySQL can no way compare to Redis and Memcache. If you find Memcache good enough, you may want to look at Tokyo Tyrant as it does a synchronous writes. But you need to check for your application which server/combination suits you the best. In Marathi there is a saying “मेल्या शिवाय स्वर्ग दिसत नाही”, which means “You can’t see heaven without dieing” or need to do your hard work, can’t escape that 😉

I’ve attached the source code used to test, if anybody has any doubts, questions feel free to ask

MySQL DB Pie Graph

Its same as what http://blog.olindata.com/2009/02/using-the-google-graph-api-with-mysql-stored-functions/ had posted, I’ve just updated a little. And there are no smart quotes here. So u can copy paste 😉


  p_chart_type CHAR,
  p_height INT,
  p_width INT) RETURNS varchar(3000) CHARSET latin1
  /* Author:    Walter Heck - OlinData */
  /* Date:      20090216 */
  /* Note:      After an idea by Alex Gorbachev - Pythian */
  /*            http://www.pythian.com/blogs/1490/google-charts-for-dba-tablespaces-allocation */

  /* variable declaration */
  DECLARE v_done BOOLEAN default false;
  DECLARE v_url varchar(3000);

  DECLARE v_schema_name varchar(3000);
  DECLARE v_data_length_sum int;
  DECLARE v_data_length_total int;
  DECLARE v_legend_labels varchar(3000);
  DECLARE v_chart_labels varchar(3000);
  DECLARE v_chart_data varchar(3000);

  /* Cursor declaration */
  DECLARE c_schema_sizes cursor for
        round(sum(t.data_length + t.index_length) / 1024 / 1024) as data_length_schema
        information_schema.tables t
      group by
      order by

  /* Handler declaration */

  /* Initialize the variables */
  SET v_legend_labels = '';
  SET v_chart_labels = '';
  SET v_chart_data = '';

  /* Get the total data length + index_length for all tables */
      round(sum(t.data_length + t.index_length) / 1024 / 1024) as data_length_total
      information_schema.tables t;
  /* Open the cursor */
  OPEN c_schema_sizes;

  /* Loop through the cursor */
  get_data: LOOP

      /* Fetch the next row of data into our variables */
      FETCH c_schema_sizes INTO v_schema_name, v_data_length_sum;

      /* if there is no more data, v_done will be true */
      IF v_done THEN
        /* Exit the loop */
        LEAVE get_data;
      END IF;

      /* Add the schema name to the labels for the legend */
      IF v_legend_labels = '' THEN
        SET v_legend_labels = v_schema_name;
        SET v_legend_labels = concat(v_legend_labels, '|', v_schema_name);
      END IF;

      /* Add the total size of the schema to the labels */
      IF v_chart_labels = '' THEN
        SET v_chart_labels = v_data_length_sum;
        SET v_chart_labels = concat(v_chart_labels, '|', v_data_length_sum);
      END IF;

      /* Get the percentage of the total size as the graph's data */
      IF v_chart_data = '' THEN
        SET v_chart_data = ROUND(v_data_length_sum / v_data_length_total, 2) * 100;
        SET v_chart_data = concat(v_chart_data, ',', ROUND(v_data_length_sum / v_data_length_total, 2) * 100);
      END IF;

  END LOOP get_data;

  /* Close the cursor */
  CLOSE c_schema_sizes;
  /* Build up the google graph url */
  SET v_url = 'http://chart.apis.google.com/chart?';
  SET v_url = CONCAT(v_url, 'cht=', p_chart_type);
  SET v_url = CONCAT(v_url, '&chs=', p_width , 'x', p_height);
  SET v_url = CONCAT(v_url, '&chtt=Database Sizes (MB)');
  SET v_url = CONCAT(v_url, '&chl=', v_chart_labels);
  SET v_url = CONCAT(v_url, '&chd=t:', v_chart_data);
  SET v_url = CONCAT(v_url, '&chdl=', v_legend_labels);

  /* return the url as the function's result */
  RETURN v_url;


PHP, Python Consistent Hashing

I found out the hashing algorithm used in PHP-Memcache is different from that of Python-Memcache. The keys went to different servers as the hash created by python and php were different.

I posted a question on the memcache groups and was lucky to find this wonderful reply.

import memcache
import binascii
m = memcache.Client(['', '
', ''])

def php_hash(key):
    return (binascii.crc32(key) >> 16) & 0x7fff

for i in range(30):
       key = 'key' + str(i)
       a = m.get((php_hash(key), key))
       print i, a

This is the only thing that has to be done on Python’s end, change the way the hash is calculated. The coding on PHP end remains same. All you guys using PHP for web based front-end with MySQL and Python for back-end scripts shall find this helpful.

Thanks Brian Rue.

Reference: http://groups.google.com/group/memcached/msg/7bb75a026c44ec43

MySQL, Memcache, Replication and Delay

An interesting article on Facebook, Which deals with MySQL’s replication system, Memcache and the replication delay.

… by setting a cookie in your browser with the current time whenever you write something to our databases. The load balancer also looks for that cookie and, if it notices that you wrote something within 20 seconds, will unconditionally send you to California. Then when 20 seconds have passed and we’re certain the data has replicated to Virginia, we’ll allow you to go back for safe pages.

The following is just a dump of the URL: http://www.facebook.com/notes.php?id=9445547199

I joined Facebook in April 2007 and, after getting settled over the course of a few weeks, my manager Robert Johnson approached me. We talked for a while but the conversation boiled down to:

Bobby: “So, Jason, we’re going to open a new datacenter in Virginia by 2008. Do you think you can help?”
Me: “Uh…. yes?”
Bobby: “Great!”

My first project at Facebook was a tad more involved then I was expecting, but I think that is one reason why we have such a great engineering organization; we have a lot of hard problems to solve and everyone here is excited to jump in and tackle them. I set out to really understand why we were building a new datacenter and what problems we had to overcome to make it work.

Why Bother?

The primary reason for building a new datacenter on the east coast was latency. It takes about 70 milliseconds to send a packet across the country on a high-speed link, and it can be much longer for an average internet user. By putting servers in Virginia we could reduce the time to send a page to users on the east coast and in Europe by a noticeable amount.

Secondary concerns were space, power, and disaster recovery. We were running out of physical space in our primary datacenters in California and the Virginia site would give us lots of room to grow. We were having a similar problem with getting enough electricity to power all those servers. Finally, restricting ourselves to only one location meant that, in the event of a disaster (power failure, earthquake, Godzilla), Facebook could be unusable for extended periods of time.

Build It!

Before we could go to work on the application level challenges our operations team put in a heroic effort to build out the servers and the physical space in Virginia. They also brought up the intra-datacenter network and the low latency inter-datacenter fiber channel link. This work was an enormous undertaking but our operations team is top-notch and made it all look easy.

With the network and hardware in place we set up our standard 3 tier architecture: web server, memcache server, and MySQL database. The MySQL databses in Virginia were going to run as slaves of the west coast databases, so we spent a couple weeks copying all the data across the country and setting up replication streams.

Now that the hardware, network, and basic infrastructure was set up it was time to face the two main application level challenges: cache consistency and traffic routing.

Cache Consistency

A bit of background on our caching model: when a user modifies a data object our infrastructure will write the new value in to a database and delete the old value from memcache (if it was present). The next time a user requests that data object we pull the result from the database and write it to memcache. Subsequent requests will pull the data from memcache until it expires out of the cache or is deleted by another update.

This setup works really well with only one set of databases because we only delete the value from memcache after the database has confirmed the write of the new value. That way we are guaranteed the next read will get the updated value from the database and put it in to memcache. With a slave database on the east coast, however, the situation got a little tricky.

When we update a west coast master database with some new data there is a replication lag before the new value is properly reflected in the east coast slave database. Normally this replication lag is under a second but in periods of high load it can spike up to 20 seconds.

Now let’s say we delete the value from Virginia memcache tier at the time we update the master database in California. A subsequent read from the slave database in Virginia might see the old value instead of the new one because of replication lag. Then Virginia memcache would be updated with the old (incorrect) value and it would be “trapped” there until another delete. As you can see, in the worst case the Virginia memcache tier would always be one “version” behind of the correct data.

Consider the following example:

1. I update my first name from “Jason” to “Monkey”

2. We write “Monkey” in to the master database in California and delete my first name from memcache in California and Virginia

3. Someone goes to my profile in Virginia

4. We don’t find my first name in memcache so we read from the Virginia slave database and get “Jason” because of replication lag

5. We update Virginia memcache with my first name as “Jason”

6. Replication catches up and we update the slave database with my first name as “Monkey”

7. Someone else goes to my profile in Virginia

8. We find my first name in memcache and return “Jason”

Until I update my first name again or it falls out of cache and we go back to the database, we will show my first name as “Jason” in Virginia and “Monkey” in California. Confusing? You bet. Welcome to the world of distributed systems, where consistency is a really hard problem.

Fortunately, the solution is a lot easier to explain than the problem. We made a small change to MySQL that allows us to tack on extra information in the replication stream that is updating the slave database. We used this feature to append all the data objects that are changing for a given query and then the slave database “sees” these objects and is responsible for deleting the value from cache after it performs the update to the database.

How’d we do it? MySQL uses a lex parser and a yacc grammar to define the structure of a query and then parse it. I’ve simplified the following for ease of explanation, but at the highest level this grammar looks like:

statement END_OF_INPUT {};

| analyze
| backup
| call
… (insert, replace, select, etc.)

Pretty straightforward, right? A query is a statement which breaks down to one of the MySQL expressions we all know and love. We modified this grammar to allow appending memcache keys to the end of any query, as follows:

statement mc_dirty END_OF_INPUT {};

| MEMCACHE_DIRTY mc_key_list;

mc_key_list ‘,’ text_string { Lex->mc_key_list.push_back($3); }
| text_string { Lex->mc_key_list.push_back($1); };

A query now has an additional component; after the statement comes the mc_dirty which is either empty or a keyword MEMCACHE_DIRTY followed by a mc_key_list. A mc_key_list is just a comma-separated list of strings and the rule tells the parser to push all the strings one-by-one on to a vector named mc_key_list which is stored inside a per-query parser object.

As an example, an old query might look like:
REPLACE INTO profile (`first_name`) VALUES (‘Monkey’) WHERE `user_id`=’jsobel’
and under the new grammar it would change to:
REPLACE INTO profile (`first_name`) VALUES (‘Monkey’) WHERE `user_id`=’jsobel’ MEMCACHE_DIRTY ‘jsobel:first_name’

The new query is telling MySQL that, in addition to changing my first name to Monkey, it also needs to dirty a corresponding memcache key. This is easily implemented. Since the per-query parser object now stores all memcache keys we tack on to a query, we added a small piece of code at the end of mysql_execute_command that dirties those keys if the query is successful. Voila, we’ve hijacked the MySQL replication stream for our own purpose: cache consistency.

The new workflow becomes (changed items in bold):

1. I update my first name from “Jason” to “Monkey”

2. We write “Monkey” in to the master database in California and delete my first name from memcache in California but not Virginia

3. Someone goes to my profile in Virginia

4. We find my first name in memcache and return “Jason”

5. Replication catches up and we update the slave database with my first name as “Monkey.” We also delete my first name from Virginia memcache because that cache object showed up in the replication stream

6. Someone else goes to my profile in Virginia

7. We don’t find my first name in memcache so we read from the slave and get “Monkey”

Page Routing

The other main problem we had to address was that only our master databases in California could accept write operations. This fact meant we needed to avoid serving pages that did database writes from Virginia because each one would have to cross the country to our master databases in California. Fortunately, our most frequently accessed pages (home page, profiles, photo pages) don’t do any writes under normal operation. The problem thus boiled down to, when a user makes a request for a page, how do we decide if it is “safe” to send to Virginia or if it must be routed to California?

This question turned out to have a relatively straightforward answer. One of the first servers a user request to Facebook hits is called a load balancer; this machine’s primary responsibility is picking a web server to handle the request but it also serves a number of other purposes: protecting against denial of service attacks and multiplexing user connections to name a few. This load balancer has the capability to run in Layer 7 mode where it can examine the URI a user is requesting and make routing decisions based on that information. This feature meant it was easy to tell the load balancer about our “safe” pages and it could decide whether to send the request to Virginia or California based on the page name and the user’s location.

There is another wrinkle to this problem, however. Let’s say you go to editprofile.php to change your hometown. This page isn’t marked as safe so it gets routed to California and you make the change. Then you go to view your profile and, since it is a safe page, we send you to Virginia. Because of the replication lag we mentioned earlier, however, you might not see the change you just made! This experience is very confusing for a user and also leads to double posting. We got around this concern by setting a cookie in your browser with the current time whenever you write something to our databases. The load balancer also looks for that cookie and, if it notices that you wrote something within 20 seconds, will unconditionally send you to California. Then when 20 seconds have passed and we’re certain the data has replicated to Virginia, we’ll allow you to go back for safe pages.

Looking Back

Nine months after our first user viewed a page in the Virginia datacenter we’re still running the same architecture with good success. There were bumps along the way, of course; for the first month or two the cache consistency infrastructure was very shaky and would periodically force us to divert traffic away from Virginia while we diagnosed and fixed bugs. Over time, however, we’ve ironed out the issues and now serve a substantial portion of Facebook’s traffic out of this datacenter.

The main scaling challenge with this architecture is pretty obvious: all write operations must happen in one location. Going forward we’re very excited to develop new technologies that will let us perform writes in any location. We’re also thinking a lot about how to use our new datacenter as a disaster recovery site in case Godzilla decides to attack our California locations! Interested in helping us out? www.facebook.com/jobs!