The graph below demonstrates the scalability and performance of a MOG managed asset pool starting with one client and climbing up to 150 machines actively performing pipeline activities. Each of these machines ran a burst test designed to point out potential bottlenecks in a network environment.
In the test we pushed assets ranging in size from 1k to 10 MB to find hard limitiations of the network environment.
- 1 kilobyte files maxed out the SQL server (SQL 2005, 1GB ram, dual 1 Ghz) at close to 50 simultanious connections.
- 10 megabyte files maxed out the fileserver at 10 connections due to harddrive limitations in our server.
- The sweet spot for MOG, the SQL server and fileserver seemed to be at the 1 megabyte asset sizes running 80 clients.
In all of our tests, we found that the primary bottleneck in network performance to be in the SQL machine allowing us to conclude that MOG's scalability over a large team is directly linked with the hardware specs of an SQL server with significant gains seen when adding cpu's.
While this graph demonstrates a slice of our current scalability and performance given a high number of concurent burst users, it should be noted that the MOG Server has been tested with over 450 active connections further demonstrating the scalability of the MOG pipeline as being a team solution that can withstand the demands of any current or future project.
