Relation between CMOD Library Server and CMOD Object Server

[obaltus]

Hello

I was reading some documentation regarding distributed CMOD architecture where you have a library server and one or more object servers.
And a few questions came to mind for which I found no answer. Where should storage sets be defined? On the library server? Or on the Object Server? Or both?

I suppose that the main purpose of this type architecture is offloading the library server from the the load activities? And ARSLOAD is then executed on the Object Server? Am I right?

Thanks for your help,

Olivier

[Justin Derrick]

That's a big question. 

First, let me start with an unpopular truth:  Object servers are an obsolete idea. 

The architecture of CMOD was designed in the early 90's, when storage and bandwidth were exceedingly scarce and expensive - there were also very hard physical limits on how much storage you could connect to a single server.  The first CMOD server I was responsible for had 4 112Mhz CPUs, 512MB of RAM, 24GB of hard drives, and a 100GB tape library -- and it was one of the most powerful midrange systems in the company -- https://en.wikipedia.org/wiki/IBM_RS/6000#Type_7013_and_7016 .

What should you do instead? 

Run a single CMOD server, and use any of the cloud-ish storage tech or Cheap and Deep filesystem storage.  Storage, CPU, Bandwidth...  All are relatively cheap.  The only remote loading you should be doing is PDFs on a Windows server.

-JD.