Load Balancing and Clustering

I am trying to get feedback about the following: Does Navision 4.0 lends itself to clustering or load balancin? I assume we can set up a standard Active / Passive cluster with shared disk, but can we do more? Can we have a separate database server (call it a SQL server) and a set of Nav application servers behind a load balancer? If this is doable, then we can provide better performance in an active active configuration while also getting fault tolerance. What are our options? Ed

Hi Ed! I’m not sure what precisely you want? But I’ll try to give “my two cents” … When you talk about a “Cluster” you mean a SQL Server Failover-cluster, Active/Passive = Server A running, Server B standby, A & B using same disk-subsystem, in case of failover Server B is switched to “active” - right? This is a standard scenario with SQL Server which is independent from the NAVISION application, of course it’s possible. But: the big disadvantage of failover-clusters is, if the disks have a problem, e.g. in case a controller is damaged, then your database is lost - inspite of the cluster. I would prefer a “warm standby” solution, e.g. running Server B on it’s own disks and synchronizing the db via “log shipping” or “db-mirroring” … of course, that depends on your environment … What do you mean with “load balancing”? You want to have multiple NAS connected to the same server, and they should balance some kind of work? Well, basically that’s possible - we have something like this in our Dansih subsidary, load-balancing incoming XML requests. It depends on what you want balance … In an easy case, of course, you could setup one NAS to process “purchase orders”, another NAS processes “sales orders”, and so on … If you have an active/active cluster, then the load balancing is done by the cluster/SQL Server itself, also, independent from NAVISION. Hope that helps a little.

I will second Joerg and reccomend a stand-by server if you want redundancy and a fast way to get back online. If You run SQL Ent. Ed. then log shipping is built in and easy to set up.

Doesn’t log shipping affect performance? That’s not replication right, it’s more of a duplication of the TL on another database?

Log Shipping involves a transaction log backup, a copy to another server and a restore there. I have not seen any performance problems with the backup so far. Anyone else?

Setting up Log Shipping would have little or no performance impact on the production server. It consist of automating the process of a periodic transactional log backup and restore to a remote server. The restore to the backup server is independent of the production server and its performance will not effect the production server. On the other hand, the new Database Mirroring feature in SQL 2005 updates both servers together. Having a slower backup server can impact overall system performance. I find Log Shipping to be a cost-effective way of providing reasonable fault-tolerance/disaster-recovery in a SQL environment. Unlike clustering, it does not require any special hardware. I find most users stop talking about clustering once they see the price tag.