StoreGrid Clustering Deployment – Some Best Practices

From managing complex data centers to providing rapid uptime in times of server crashes and also scaling to the dynamic backup industry is no mean a trivial thing for a service provider. We understand that for you MSP’s, failure to provision any of these can mean lost credibility, lost revenues, or worse, lost clients. StoreGrid’s Round Robin Algorithm based Clustering support for StoreGrid backup and replication servers is designed to deliver high availability services and ease and automate administration in complex datacenters. Here are some of the key highlights:

  • Accelerate Performance –  Achieve optimized response time for an overall improved performance with StoreGrid’s ability to provide tiered architecture by bringing greater efficiency to your data center.

  • Flexible Scaling during peaks and dips –  Scale up cluster servers seamlessly during demand spikes and scale down unused servers when demand subsides and dramatically reduce costs and no downtime.

  • Dramatic Cost and Power Savings  - The flexibility to scale down unused servers when servers are idle or during downtime helps reduction in disks, power, and space enabling dramatic savings in CAPEX and OPEX.

  • Visibility & Simplified Management – With StoreGrid’s user friendly User interface, it takes just minutes to configure a cluster,  add servers to the existing cluster, edit cluster settings, scale in and scale out servers with no disruption to the entire environment.

  • Leverage Existing Infrastructure –   Easily convert your existing setup into a clustered environment with no added setup costs. StoreGrid allows easy configuration of clusters and automatic server load balancing.

The  work flow of StoreGrid’s Clustering strategy was discussed some time back on our blog. And we are back with  some recommended ways for setting up your backup and replication servers to effectively use Clustering to maximize RTO for your datacenter during disasters.

TIP # 1 Start with one or two servers to deploy Clustering. Add multiple servers as your business grows, as you can easily scale up/down the servers based on the backup load.

TIP # 2 Install MySQL service in a dedicated Server with enough hardware resource to handle huge requests rather than installing MySQL on StoreGrid primary node. With this configuration, even if the primary backup server machine went down, you can continue your business by redirecting the backup/restore requests to other backup server nodes.

TIP # 3 Keep the MySQL MetaData Files in a common backup data storage location. This will help you to rebuild your MySQL server within minutes, if your MySQL server went down due to any data corruption or any disaster.

TIP # 4  Specify the dedicated Public IP Address/Public DNS Name of the machine, while configuring a cluster.

The below figure shows the typical configuration of Clustering in a datacenter.

 

The ideal way to set up a cluster behind a single public IP/Domain Name

 Though the clients need to directly connect to the individual backup servers in the cluster to start backing up, you don’t need to have a unique public IP/domain name for each of the backup servers. Instead, you can setup each backup server to listen on a different (unique) backup port.

For example, Primary server has been installed with 32004 for Backup port, 32007 for SSL and 32005 for GUI, then you need to specify Backup, SSL and GUI ports as 42004, 42007 and 42005 respectively for the secondary server node and so on.

The NAT in the network where cluster is present can be configured to forward the TCP traffic coming in those specific ports to the appropriate backup server. Like for example the NAT could be configured to forward all traffic coming on port 32004 to backup server A, traffic coming on port 42004 to Server B and so on. A client machine assigned to backup server A will initiate a connection to the cluster at the public IP/DNS name on port 32004, a client assigned to backup server B will initiate a connection on port 42004. The illustration below explains this.

So we are set with the configuration of the cluster. Now how do we ensure the best uptime for your datacenter during disaster ?

Case 1 : If the server node other than the primary node went down, then no worries!. The backups which are backing up to that node will automatically start backing up to the other active node(based on round robin basis) on the next schedule.

Case 2 : Incase the Primary node went down, No need to panic! Just redirect the requests received for Primary Server to a Secondary Server Node by following the simple steps below,

If you have assigned separate Public DNS Name/Public IP Address for each of your server nodes.

Step 1: Assign the Primary server’s Public DNS Name/Public IP Address to your Secondary server node(Server B).

Step 2: Login to StoreGrid backup/replication server console and goto “Server Management -> Cluster Management -> List Cluster Nodes -> Click ‘Edit Cluster Node’ action” and update the Primary server Node’s Public DNS Name/Public IP Address for Secondary server Node and save the configuration.

Step 3: Go to StoreGrid installation location and edit ‘conf/SGConfiguration.conf’ file and update the Primary server Node’s Public DNS Name/Public IP Address in ‘NodeLookupNameIP’ tag and restart StoreGrid service.

If your cluster backup/replication servers are configured with single Public DNS Name/Public IP Address

Step 1: Configure NAT to redirect the requests received for Primary server Node  to Secondary Server Node.

Step 2: Update the Primary Server Node’s ports to Secondary Server Node by logging in to your StoreGrid backup/replication server console.

Step 3: Go to StoreGrid installation location and edit ‘conf/SGConfiguration.conf’ file and update the Primary server Node’s Public DNS Name/Public IP Address in ‘NodeLookupNameIP’ tag and restart StoreGrid service.

So until you recover your Primary Server Node, your Secondary Server Node is set to handle all the requests from your clients.

Check out clustering in StoreGrid with the free trial. Post your comments and let us know whether you found the above blog post useful and what other feature tips would you like to see in the blog.

 

Share and Enjoy

  • Facebook
  • Twitter
  • Delicious
  • LinkedIn
  • StumbleUpon
  • Add to favorites
  • Email
  • RSS