mysql database Multi-master replication on dynamic ip - php

Situation:
Php application with mysql database running on 2 sites
online -static ip X.X.X.X
localhost (not online most of time and dynamic ip)
application traffic is usually low <10 users.
what i need is that whenever a change is done to the online database, this change is pushed to localhost -if its online or when ever its available- and vise versa (any changes done locally is uploaded online to database when ever there is online connection).
is it possible to setup such replication by mysql ? or do i need to write a custom PHP that ping master server and once its available
thanks very much :).

Yes you can do this with replication. Just pick which server you want to be the master and have the second one send all of its changes to the main one then the main one could send its changes back.
Replication can be a bit daunting to set up but once its up and running its grate. http://dev.mysql.com/doc/refman/5.0/en/replication-howto.html

Let's first analyze your question:
The problem of accessing MySQL with a dynamic ip.
This is very easy. Once you installed MySQL on a server with an ever-changing IP, what you can do is go to NO-IP, DynDNS or any other Dynamic DNS service and register for free with them. Once you've registered with them, you would get a client for your operating system. Install that and then you can access your MySQL server using a domain name.
Example:
Instead of having to access your server at 127.0.0.1, you can access it as mysql-server.easynet.net etc.
Now the second and albeit complex part of your question, how to do available and lazy replication.
This is relatively a bit more complex than the previous step. But, what actually happens is that you have to choose a scheme of replication. Basically what you are looking for here is MASTER-MASTER replication since you have a possibility of changes happening at both the MySQL servers. Thus the updates need to be bi-directional, that's what this scheme of replication does. How to do it? Well, I am providing the links which I've found easier to follow:
Master-Master Replication
Step-by-step MySQL Master Replication
I hope that would ease your plight and answer your question!
Cheers!

Sure, you can
You need to setup both MySQL servers as Master and Slave at the same time.
Configure the online server as Master, and the localhost server as slave, and once replication is OK.
Configure the localhost as Master and the online server as slave.
I already did that on two servers.
About the dynamic IP on the local host, simply you can use any dynamic IP service like: no-ip, and use the dns name instead of the IP.

Here's a post i've written (in french, but you can get the configuration snippets from it) for setting up a MASTER-MASTER replication with a load balancer (mysql proxy) for balancing SQL queries between both nodes.

Related

MySQL replication alternatives

I'm using Curl to send information between local and remote MySQL databases in order to keep them synchronised.
Each row that is sent keeps record of wether or not it has been updated in the central database, and keeps a record of the central database id for its transaction.
Is there a better way to keep many local databases synchronised with a central remote server?
PS I cannot use database replication because of shared hosting server restrictions.
You have not really answered Alfallouji's question. Nor have you said anything about any of the contraints to your approach.
You said you want to replicate the data somewhere else, when the primary database is the hosted solution. What access is there to 'somewhere else' from the hosted server? Does somewhere else have a static IP address? Is it always up?
Can you access the binary log on the primary server?
How much replication lag can you accomodate?
There are lots of solutions.....mysql_proxy, mysql's built-in asynchronous replication (optionally with manual copying of replication logs). Multi-master replication. Or at the device level with DRBD, or write a custom dump tool and rsync the results.....

running the same database both local and online in mysql

I am creating a software for my client in PHP and MySql, The software will be running in the local network, the software should also run in online that is all the data should be viewed online, i would like to know is there a way to update the mysql database from local to online whenever the internet is connected, if the internet is not connected all the data will be stored in local mysql server. they wont be adding any data in the online server and they will only view the contents. Kindly help me in this
It sounds like you are looking for a way to have a read-only, "online" (assuming WAN) MySQL server which is updated from a read-write, "offline" (LAN) MySQL Server which is updated by your users.
If that's the case, you may want to consider a Master/Slave MySQL Replication configuration:
http://www.rackspace.com/knowledge_center/article/mysql-replication-masterslave
you can run a cron job in your local server which monitor and upload data to online server when internet is available. You need to track (set a status) which data is uploaded which not for synchronization
Setup replication between your mysql databases. The local network database would be the slave and the web server would be the master.
See http://dev.mysql.com/doc/refman/5.5/en/replication-howto.html for setting up replication.
I would suggest you setup master/slave replication.
I would make the local server the master and the internet server the slave.
However, if you are using a common shared hosting service, I am not too sure if you have permission to setup such DB replication structure.
Tryhttps://www.digitalocean.com/community/articles/how-to-set-up-master-slave-replication-in-mysql. However, with the mysql versions pre 5.6 (I think) you can't delay the replication from a predefined interval. However, http://alexalexander.blogspot.com/2013/03/mysql-slave-delay-how-to.html suggests using something called a percona tool kit. I haven't used it though.

scaling PHP application with MySQL replication without PHP code change?

I am planning to increase My site performance by adding another MySQL server beside the current one because the current server is too busy.
Is it possible to scale PHP application with MySQL replication without PHP code change? I means all quires will be sent to the master and the master will distribute the load between itself and the slave.
Is there any easy way to send all write quires to the master and distribute read quires between the master and slave?
I think you need to put a load balancer / proxy between your db servers and clients (your code). Example solutions are:
HAProxy: http://haproxy.1wt.eu/
MySQL Proxy: https://launchpad.net/mysql-proxy
If you don't want to do the "load balancing" manually, you might want to look into MySQL Proxy.
I think you should also optimize your application's code (PHP) and then you should optimize your architecture.
First of all you can check your MYSQL queries. Mysql slow query log can help you. If you have a connection issues (MYSQL server has gone away or too many connections etc) you should manage your application's connection pooling mechanism.
And other steps and also your answer is (I think), you can set up MYSQL master-master replication. When you set replication clearly, you can put a load balancer (HAProxy) front of your replication.
You have 2 nodes for mysql (server A and server B, both of them master server)
You can configure HAProxy with server A is master and server B is backup server. Your all MYSQL operations comes server A via HAProxy and your data is automaticly sync with server B.
When server A is down, HAProxy sends all queries server B automaticly.
Also you can configure HAProxy with server A is all insert queries and server B is for all read queries.
All this cases your code should connect MYSQL via HAProxy

PHP/MySql clusters

I am currently planing a web application and I want to plan it to eventually run on a cluster later.
The cluster would be made of a php web cluster and a mysql cluster and a standalone storage unit (maybe a cluster of it I really don't know how that works :s)
I want to know if the code will be different than when php and mysql are on the same machine and what would be different?
The fact that the web and database servers are on different physical machines wouldn't change your code at all. The only place you'd need to change code is where you connect to the database - replacing the localhost reference with the IP address or hostname of the database server.
A clustered web server may need a different approach for storing sessions. If you got multiple webservers behind a load balancer, consequitive requests from the same session may end up on different servers. You should store the session data in a different place, like a central memcache.
Apart from a few of those issues, you should be fine regarding the web server.
As far as I know, MySQL and clustering are no friends. Although I wasn't really involved in the process, I know there has been a lot of trouble to get two database servers run together in our environment and even now they are not really clustered. They syncronize, but only one is actively used while the other is a fallback server.

Can local intranet application (built on php) query mysql database stored in offsite location?

I have a local intranet application which runs off a basic WAMP server in our offices. Every morning, one of our team members manually syncs our internal mysql db with our external mysql db (where our online enrollments occur). If a change is made during the day on the intranet application, it is not reflected on the external db until the following day.
I am wondering if it is possible to (essentially) tunnel to an external mysql connection from say a wamp or xampp server from within our offices and work in 'real-time'.
Anybody had any luck or advice?
Yes
Replication enables data from one MySQL database server (the master) to be replicated to one or more MySQL database servers (the slaves). Replication is asynchronous - slaves need not to connected permanently to receive updates from the master. This means that updates can occur over long-distance connections and even over temporary or intermittent connections such as a dial-up service. Depending on the configuration, you can replicate all databases, selected databases, or even selected tables within a database.
If you use the external server directly, performance is likely to suffer. A Gigabit LAN might be a thousand times faster than your Internet connection - particularly the upload speed of an ADSL connection.
Just make your internal application use the database from the external one. You may need to add permission to the external server to allow connections from your internal server IP, but otherwise this is just like having a webserver and sperate db server that need to access each other.
Can't really tell you how to do this here - it all depends on your specific configuration, something that I would thing is a little complicated (and too specialized) to figure out on SO.

Categories