The primary purpose of failover clustering is to guarantee uptime for a server that is used to persist data on its disk subsystem. To completely protect BizTalk Server against hardware failures, failover clustering should be implemented in the areas listed in Table 9.1.
Table 9.1 Areas to Implement Failover Clustered Solutions
Message queues Message queues should be looked at with a critical eye due to the importance of queuing. BizTalk receives or sends its critical data using Microsoft's Message Queuing (MSMQ). Queues are used where high throughput and reads and writes are very important, if not critical. MSMQ can be clustered and should be worked into your clustered BizTalk design.
SQL Server databases We learned in previous chapters and doing installs of BizTalk that BizTalk Server needs four databases to function properly: Message Management, Shared Queue, Tracking, and Orchestration. You would of course want to cluster the database on which these essential databases reside.
File shares BizTalk will store, send, and receive text files as part of the processing service of BizTalk. If these files are not available, BizTalk Server cannot process this data. A file share can optimize the ability to perform after a failure and keep the server running. It is most important to protect against this by clustering the files by clustering a share.
The WebDAV repository Document specification, which we learned about in earlier chapters, is critical to BizTalk production work. If the WebDAV repository is either unavailable or damaged, how can business continue? Set failover-based clustering to make sure you have a redundant repository. The WebDAV (Web Distributed Authoring and Versioning) repository for BizTalk Server must have a reliable storage clustering in case of failure.
If you do not intend to use either flat files or Message Queuing queues, then it is unnecessary to cluster these resources.
You might hear that you can optimize BizTalk performance by scaling it.What exactly is scaling, and should you choose to scale out or up? Scaling out is adding more servers to the equation so that you can expand out.You can also scale up, which means using more powerful hardware such as CPUs and memory to increase performance.
Was this article helpful?