Digital Information Gateway (DIG^®) Scalability

The ability of a multi-tier application server to scale with large numbers (10,000+) of users and large numbers of simultaneous transactions (queries, etc.) is affected by several factors. It is important to note that "scalability," as perceived by end-users, includes much more than just the application server. The "system" includes back-end databases, document indexes, network switches/routers, WAN/Internet bandwidth, and the application server. The databases, document indexes, local network and connection bandwidth must also be capable of handling these 10,000+ users. The real response time for each user will only be as fast as the slowest component in the "system." Therefore, the entire system, including the application server (DIG, in this case), must be well optimized. It is critical that your application server not be the weakest link in the system. The following factors are important considerations in application server scalability.
	Clustering - The ability to cluster multiple physical servers to operate in parallel. This allows the application to grow/scale in conjunction with greater numbers of users by simply adding servers as necessary. In contrast, applications bound to a single server cannot scale past the number of users that a single server can support.
	Statelessness - Linear scalability (i.e., application responds in 1 second with 500 users, 2 seconds with 1000 users, 4 seconds with 2000 users, and so on...for a given set of hardware). Although true linear scalability is not entirely possible, near linear scalability has been proven time and again with stateless architectures. Stateless architectures do not require copying/synchronization of information between multiple application servers. Rather, persistent state (like tracking currently authenticated users) is looked up in a common location, like a back-end database. This enables the multi-server cluster to truly work in parallel because users can be dynamically routed to the server with the least load. This enables all resources on all servers to be utilized to their fullest extent.
	Execution Speed - Code efficiency (i.e., how quickly the code executes on each server). The faster the code executes on each server, the more users each server can support requiring fewer servers to support n number of users.
The Digital Information Gateway (DIG®) software architecture supports all of these features. This means that DIG can be clustered with an unlimited number of physical servers that work in parallel to fulfill user requests. In benchmark tests at Visual Analytics, DIG boasted 100+ queries per second on a single Dell Server priced under $10K. To translate this into an approximate number of users, we will assume that 1000 concurrent users are each performing a new query every 10 seconds.
For more information about how DIG can help solve information retrieval needs in your organization, contact Visual Analytics, Inc. at sales@visualanalytics.com.