Results 1 to 5 of 5
Like Tree1Likes
  • 1 Post By cioannides

Thread: Large Scale Deployment query

  1. #1
    cioannides Guest

    Default Large Scale Deployment query

    Hi everyone. I need some help designing the infrastructure for deploying a large scale Sugar installation. The company I work for is a financial institution that has users across the globe. We will have initially 400 users, but most probably this number will increase. Most of these users are based in Europe in a single location. As we expand in Asia, we need to keep another installation in our datacenter there for our users there. Ideally we will need to synchronize the two database installations with a master-master configuration. We were thinking of a single installation in Europe but ping times from Asia are way to high so we are going with 2 installations (one in Europe and one in Asia). Both installations need to have a fall back in case of one installation going offline.

    Initially we were thinking of having 4 Oracle installations (2 on each datacenter, live and backup in each location). The cost however for licensing Oracle for such scenario is massive. So we were thinking of going with MySQL. Our problem is that we have never used MySQL for remote replication and we wanted to know if anyone has experience in setting up Sugar in similar cases (and MySQL for that matter). The requirement is that there should be 4 installations: 2 datacenters, 2 installations in each datacenter (one live and one backup). Sugar response times should be kept to minimum.

    Any help would be greatly appreciated!

  2. #2
    christianknoll's Avatar
    christianknoll is offline Sugar Community Member
    Join Date
    Nov 2008
    Location
    Vienna
    Posts
    1,382

    Default Re: Large Scale Deployment query

    The dual master setup each with an additonal slave should work but has some implications where you shoudl be prepared on sync issues. We are happy to help and have other quite complex and large installations. Also with mirrors etc. Not yet one like that but we like the challenge. Comtact me at info@kreporter.org if you are interested to start a conversation.

    regards, christian.

    Quote Originally Posted by cioannides View Post
    Hi everyone. I need some help designing the infrastructure for deploying a large scale Sugar installation. The company I work for is a financial institution that has users across the globe. We will have initially 400 users, but most probably this number will increase. Most of these users are based in Europe in a single location. As we expand in Asia, we need to keep another installation in our datacenter there for our users there. Ideally we will need to synchronize the two database installations with a master-master configuration. We were thinking of a single installation in Europe but ping times from Asia are way to high so we are going with 2 installations (one in Europe and one in Asia). Both installations need to have a fall back in case of one installation going offline.

    Initially we were thinking of having 4 Oracle installations (2 on each datacenter, live and backup in each location). The cost however for licensing Oracle for such scenario is massive. So we were thinking of going with MySQL. Our problem is that we have never used MySQL for remote replication and we wanted to know if anyone has experience in setting up Sugar in similar cases (and MySQL for that matter). The requirement is that there should be 4 installations: 2 datacenters, 2 installations in each datacenter (one live and one backup). Sugar response times should be kept to minimum.

    Any help would be greatly appreciated!

  3. #3
    cioannides Guest

    Default Re: Large Scale Deployment query

    Thanks for your reply christian. We actually took some professional help and decided to go with the following:

    The application servers will be configured as described in my first post. The database setupthough will be configured in the following manner:

    * There will be one master located in one of the datacenters. This master will only serve as a write-only database. All updates/inserts will be committed in this database.
    * Slaves will act as read-only and will be located in the same datacenters as the application servers.
    * Sugar installations will be configured to write on the master but read from the slaves.

    In this way we eliminate most of the master-to-master sync issues whilst keeping the fast local read response times. Write/update response times are not as essential (for our scenario) so the write (master) db server can be located in a remote datacenter. In case of fail over, one of the slaves will be converted to master to handle all writes/updates. Also this setup is more scalable than master-master setup.

    I will try and update this thread when I have some results from the actual implementation (it can take a couple of months)
    Chris_C likes this.

  4. #4
    agcopley is offline Sugar Community Member
    Join Date
    Nov 2007
    Location
    Santiago, Chile
    Posts
    423

    Default Re: Large Scale Deployment query

    Quote Originally Posted by cioannides View Post
    Thanks for your reply christian. We actually took some professional help and decided to go with the following:

    The application servers will be configured as described in my first post. The database setupthough will be configured in the following manner:

    * There will be one master located in one of the datacenters. This master will only serve as a write-only database. All updates/inserts will be committed in this database.
    * Slaves will act as read-only and will be located in the same datacenters as the application servers.
    * Sugar installations will be configured to write on the master but read from the slaves.

    In this way we eliminate most of the master-to-master sync issues whilst keeping the fast local read response times. Write/update response times are not as essential (for our scenario) so the write (master) db server can be located in a remote datacenter. In case of fail over, one of the slaves will be converted to master to handle all writes/updates. Also this setup is more scalable than master-master setup.

    I will try and update this thread when I have some results from the actual implementation (it can take a couple of months)

    Unless your replication is basically instantaneous, how can you avoid your detail and list views showing old data after saving an update in editview?

    Rgds
    Andrew

  5. #5
    cioannides Guest

    Default Re: Large Scale Deployment query

    This will be an issue and it will depend on the connection we will have between the master and slaves. We are prepared to accept this cost since most users will open, edit, save and return to list view and each user will only deal with his assigned leads. However this will be a problem; we just don't know how big of a problem it will be. This will show after we perform the initial testing.

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Similar Threads

  1. Large-scale SugarCRM CE development
    By sugar_crm_consultant in forum Classifieds
    Replies: 0
    Last Post: 2012-09-06, 06:02 PM
  2. Large Install - Query problem
    By blackradius in forum Developer Help
    Replies: 1
    Last Post: 2010-05-26, 12:32 AM
  3. Scale way up???
    By dogfuel in forum Developer Help
    Replies: 0
    Last Post: 2008-06-28, 03:26 AM
  4. Graphics scale
    By ldebernardini in forum Help
    Replies: 2
    Last Post: 2006-08-22, 05:56 PM
  5. Replies: 0
    Last Post: 2006-04-01, 09:28 AM

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •