Jul 26, 2014

Validating Your Disk Performance

Databases and data stores in general (yes, including the leading the leading NoSQL products such as MongoDB and Cassandra) love IO. The more RAM and SSD they have, the better performance you will get...

If you are short of RAM and get a poor performance, it may be a good time to verify your disks performance and see if they do match your needs.

How much IO do you need?
To analyze your needs you should use iostat. This utility will show for each mounted disk the actual number of blocks that are read from disk and are written to disk.
If the numbers are constantly higher than 1,000 blocks/sec, you should consider improving your application design or move from magnetic disks to SSD. In the case below for example the system has a write issue.

> iostat 2
Linux 2.6.32-358.11.1.el6.x86_64 (example.com)         07/26/2014      _x86_64_        (16 CPU)

avg-cpu:  %user   %nice %system %iowait  %steal   %idle

           2.02    0.00    0.92    0.01    0.00   97.04

Device:            tps   Blk_read/s   Blk_wrtn/s   Blk_read   Blk_wrtn

sda              92.88       599.18      3920.00  547971262 3584968522
sdb               0.00         0.00         0.00       3104          0

What are Your Disks Limits?
If you are not sure how much can you get from your disks, Bonnie++ is a great tool to explore your current limits. 

Installing Bonnie++
> wget http://www.coker.com.au/bonnie++/bonnie++-1.03e.tgz
> tar xvfz bonnie++-1.03e.tgz
> cd bonnie++-1.03e
> make
> make install

Running Bonnie++
Select your target disk, the size of written data in MB (the -s flag), the number of repeats (the -n flag) and the server RAM size in MB (the -r flag). Please notice that you should define a write to disk value that is larger than the actual RAM value in order to avoid false results:
> sudo ./bonnie++ -u root -d /tmp -s 1500 -n 3 -r 750

Detecting an issues
As you will be able to see in the following lines Bonnie++ is very useful in detecting disk performance issue and failed disks (see Server C that has a very poor write performance)
Server A (SAS Disks) Results
Version 1.03e       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
serverA        100M 52626  99 +++++ +++ +++++ +++ 61374  99 +++++ +++ +++++ +++
                    ------Sequential Create------ --------Random Create--------
                    -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP

Server B (SSD Disks) Results
Version 1.03e       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
serverB        100M 61864  99 +++++ +++ +++++ +++ 58338  99 +++++ +++ +++++ +++
                    ------Sequential Create------ --------Random Create--------
                    -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                  3 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++

Server C (Failed SSD Disks) Results
Version 1.03e       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
serverC        100M  3892  99 104779  99 98847  99  3865  99 +++++ +++ +++++ +++
                    ------Sequential Create------ --------Random Create--------
                    -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                  3 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++

Bottom Line
When you plan your next move, you should verify your servers can handle it. These tools will help you verify at least your disk can take care of the your future tasks.

Keep Performing,
Moshe Kaplan

Jul 24, 2014

9 steps to Implement Offline Backup at Azure IaaS: Backup a MySQL to Azure Storage

A good backup is probably something you will thank for when s%$t hits the fan.
If you chose MySQL as your data infrastructure and Microsoft Azure as your cloud infrastructure, you will probably thank this procedure (that I actually turned into a a script).

Chosen Products
Our first task is choosing the backup and automation products. I selected the following two:
  1. Percona XtraBackup: a leading backup product by Percona. The product creates an hot backup of the database that is equivalent to disk copy. This method is much faster to backup and recover than mysqldump. It also support increment backup.
  2. Azure SDK Tools Xplat: a node.js based SDK that enables command line interface to the various Azure services including Azure Storage.
Backup Implementation Guide
  1. Install Percona XtraBackup
    sudo wget http://www.percona.com/redir/downloads/XtraBackup/XtraBackup-2.2.3/binary/debian/precise/x86_64/percona-xtrabackup_2.2.3-4982-1.precise_amd64.deb
    sudo dpkg -i percona-xtrabackup_2.2.3-4982-1.precise_amd64.deb
    sudo apt-get update
    sudo apt-get install -f
    sudo dpkg -i percona-xtrabackup_2.2.3-4982-1.precise_amd64.deb

    The CentOS
    sudo rpm -Uhv http://www.percona.com/downloads/percona-release/percona-release-0.0-1.x86_64.rpm
    sudo yum -y install percona-xtrabackup.x86_64
  2. Install Azure SDK Tools Xplat
    sudo apt-get updatesudo apt-get -y install nodejs python-software-propertiessudo add-apt-repository ppa:chris-lea/node.jssudo wget https://npmjs.org/install.sh --no-check-certificate | sudo shsudo apt-get install npmsudo npm config set registry http://registry.npmjs.org/sudo npm install -g azure-clisudo ln -s /usr/bin/nodejs /usr/bin/node
  3. Install a backup procedure
    1. Get a publish settings file from Azure (can be done from the console).
    2. Get account name and the matching base64 key from the Azure console.
    3. Import the publish setting filesudo azure account import /opt/mysqlbackup/mysqlbackup.publishsettings
    4. Create a Storage Container
      sudo azure storage container create --container container_name -a account_name -k base64_account_key
  4. Run the backup
    1. The xtrabackup way
      1. Run a full backup and prepare it twice to make it ready for recovery
        sudo xtrabackup --backup
        sudo xtrabackup --prepare
        sudo xtrabackup --prepare
      2. Add the frm files and mysql database to your backup
        sudo chmod -R +r /var/lib/mysql/
        sudo cp -R /var/lib/mysql/myql/* /mnt/backup/mysql/
        sudo cp -R /var/lib/mysql/yourdb/*.frm /mnt/backup/yourdb/
    2. The innobackupex way:
      1. Run full backup and prepare (adjust the memory usage for your case):
        sudo rm -rf /mnt/datadrive/mysqlbackup/*
        sudo innobackupex --user=DBUSER --password=DBUSERPASS 
        sudo innobackupex --apply-log /mnt/datadrive/mysqlbackup/ --use-memory=500M
  5. Tar the files into a unique daily name_now=$(date +"%Y_%m_%d")
    tar cvzf "$_file" /mnt/datadrive/mysqlbackup/
  6. Copy the folder to Azure Storage using
  7. azure storage blob upload -f "$_file" -q --container container_name -a account_name -k base64_account_key
  8. Create a cron that will run it daily:
    > sudo cron -e
    * 0 * * * /opt/mysqlbackup/daily.backup.sh >/dev/null 2>&1
Recovery Guide
  1. Bring back the files from the Azure storage to /mnt/backup/
    sudo cd /mnt/backup
    sudo azure storage blob download --container container_name -a account_name -k base64_account_key -b $file_name
  2. Uncompress the files
    sudo tar xvfz $file_name
  3. Copy the files to your data folder (/var/lib/mysql) after shutting down the MySQL
    1. The xtrabackup waysudo service mysql stopsudo rsync -avrP /mnt/backup/ /var/lib/mysql/
    2. The innobackupex way:
      sudo service mysql stopsudo rm -rf /var/lib/mysql/*sudo innobackupex --copy-back /mnt/datadrive/mysqlbackup/sudo chown -R mysql:mysql /var/lib/mysql
  4. Verify the folder permissions
    sudo chown -R mysql:mysql /var/lib/mysql
  5. Restart the MySQL and verify everything is working.
    sudo service mysql start
Bottom Line 
It may be tough. It may be resource and time consuming. However, you must have good recovery process to keep your a$$...

Keep Performing,

Jul 17, 2014

How to Disable MySQL Binlog

@ MySQL, clean the the old bin log files 
Take the File name

@ The /etc/my.cnf
Comment out log-bin
> #log-bin = ....

Restart the mysql
> sudo service mysql restart

Keep Performing,

Jul 14, 2014

Keep Your Website Up and Running During an Upgrade

All of us want to have our website up and running.
However, upgrading a software version requires downtime (even if it minimal).
What should we do?

The answer is simple: follow the following process and gain a highly available system based on load balancing and session off loading, as well as a no downtime procedure to upgrade your system.

Keep Your Site Up and Running @ AWS

  1. Place more than a single web instance behind a load balancer. The Elastic Load Balancer (ELB) will be fine for that.
  2. Make sure you are using session offloading in your system implementation.
  3. Install elbcli on your machine sudo apt-get -y install elbcli
  4. Create a credentials file /var/keys/aws_credential_file and secure it sudo chmod 600 /var/keys/aws_credential_file
  5. Upgrade the servers one after the another.
  6. Before upgrading a server, take it out of the ELB
    sudo elb-deregister-instances-from-lb MyLoadBalancer --instances i-4e05f721  --aws-credential-file=/var/keys/aws_credential_file
  7. And after completing the server upgrade take it back in.
    sudo elb-register-instances-with-lb MyLoadBalancer --instances i-4e05f721  --aws-credential-file=/var/keys/aws_credential_file
Bottom Line
A simple process and careful design will keep you system up and running.

Keep Performing,

Jul 10, 2014

Scale Out Patterns for OpenStack (and other Cloud) based Systems

It is a common question these days how one should design their next system to support elastic and growth requirements in the cloud era.

As I got this specific query today, I would like to share with you my answer based on various materials I created in the last few years:

1. Avoid File Storage SPOF: Use AWS S3 or its equivalent open source OpenStack Swift as a central file server repository when needed

How to use OpenStack Swift for your business case

2. Avoid Data Store SPOF: Use clustered data store that recovers automatically w/o a DBA or an operator intervention. Fine examples are AWS MySQL RDS, MongoDB and Cassandra

MongoDB HA concepts

3. Avoid Static Servers: Leverage Autoscale features or customize it for your needs

How to implement your application logic to auto scale your app

4. Avoid Messing with Cache: Use a central sharded cache to avoid cache revocation and reduce data stores load using MongoDB, CouchBase or Redis.

5. Offload your servers sessions: in order to avoid users log off and lost transactions:

How to implement a session offloading using a central store

6. Avoid Service Downtime due to Servers Downtime: More issues that can be found in my extensive presentation:

 - Use DNS Load Balancing for Geo LB and DRP (Slide 10)
 - Use CDN to offload network traffic (Slides 16-19)
 - Perform Session Offloading by cookies or a central store (Slides 62-65)

Bottom Line
You can scale your app! just follow the right recommendations

Keep Performing,

Moshe Kaplan


Intense Debate Comments

Ratings and Recommendations