A Flexible Virtual Archive using Open Text Content Server

No replies
Dave Kinchlea
Dave Kinchlea's picture
Offline
Joined: 2009-04-22

I've not touched load balancers for quite some time, but a recent engagement has made me take another look as I learn about a relatively new entry, the Citrix Netscaler. This product promises to bring a lot of technology to the table to offer Availability, Performance, and Security to a Livelink installation. I'm looking to use this technology to provide an archiving service that is at the same time simple for clients to use, simple for IT to setup and configure yet highly flexible. The desire is to supply a single archiving virtual service that provides 4 to 5 distinct archiving applications all sharing a single set of physical resources. The archiving applications are a mix of both interactive and automated systems and there is a requirement to offer preferential treatment to interactive users.

The easiest way to achieve the preferential treatment would be to provide a number of different URLs and dedicate physical resources to meet each application, but that approach not only makes for a confusing client landscape it also makes it difficult to reassign resources if/when business objectives change. Not only would client configuration potentially have to change, so too would the backend resources have to be redeployed ... these activities not only speak to a lot of IT effort but also a significant change management problem. Thus the "easiest" method provides the least flexibility and and overly costly infrastructure.

Another, common approach to this problem would be to use scheduling and governing techniques at the client end to restrict some of the automated archiving activity to certain times of day/night when there is historically less interactive use. While this approach typically is very manageable, it doesn't offer a lot of flexibility. Furthermore, while the archiving resources may be less used at certain times, it is common for other IT activities to be active during those times. Thus this approach often leads to scheduling difficulties and conflicts with other automated IT tasks like daily backups.

The advantage of the Virtual Archive is the simplicity and flexibility it provides. All archiving clients are configured identically and all Content Server instances are identically configured. The Virtual Resource Pools are defined and managed within the Load Balancer, each pool using a different number of dedicated Content Server instances as befits the application the pool is designed to meet. Finally, the Load Balancer uses Layer 7 (application) load balancing to determine the correct pool to use...that is, the URI of the incoming request is examined and the "func=X" parameter is used to route the request to the appropriate pool.

Visual representation of the Virtual Archive

 

There are clear advantages to this approach, particularly in a complex IT environment. It allows the sharing of physical resources while still allowing for preferential treatment of some transactions over another. It also provides an easy method to redeploy resources on the fly to meet changing business needs without complex and time consuming reconfiguration of client or backend resources.

The disadvantage is that the load balancer itself becomes much more complex demanding more knowledge and skills from the administrator.

In subsequent articles I'll talk about how this architecture is managed using the Citrix Netscaler as the load balancer.