One of the questions that Ed Brill loves to ask during his presentations is how many administrators in the audience don't have any mail quotas set for their users. He usually gets a fair number of hands going up and it makes him smile. He is a big proponent of not putting quotas on users' mail files and his reasoning is sound. Good thing he doesn't work for MicroSloth.
Many customers tell me that they do not impose mailbox and/or message size limits on the vast majority of their user population. This poses a problem because you will never be able to adequately design a storage subsystem or maintain the desired database size limits, let alone meet any service level agreements that may be in place.
Additionally, most of these customers do not have defined, measurable, and actionable Service Level Agreements (SLAs) or Operating Level Agreements (OLAs) in place to manage user expectations (i.e. messaging as a service), and thus cannot measure a real cost/benefit with regards to the user population demands. Ultimately, the lack of mailbox size limits and message size limits can lead to instability in the messaging architecture.
So let me understand this if I can: unless mailbox quotas are implemented, it is impossible to design or maintain a stable Exchange environment. I might be able to agree with the need for message size limitations (we all have horror stories of the marketing team trying to send out 20 MB presentations to the whole company and crashing a mail server in the process), but the fact that you lack mailbox quotas will lead to your system becoming unstable is an absurd statement, even if it is true. The only reason I ever like to impose mailbox quotas is when we are tight for disk space and are waiting for more to be installed. Since Domino is not reliant upon a SCOS like Exchange, the problems with disk space are things that can be easily monitored and solved by throwing money at the problem. Disk is cheap!! And any performance impact is on a mailbox by mailbox basis. An accounting user with a 25 MB mailbox isn't impacted if the VP of HR let's her mailbox grow to over 1 GB. The other thing to remember is that mailbox performance is impacted greater by the number of emails rather than the size of the mailbox. The more emails you have, the larger the view and folder indexes have to be and the longer they take to update.
Ross goes on to point out a number of reasons why not having quotas are a bad thing. Most of his ideas point to the main problem with Exchange in most organizations: it is no longer a tool to enhance a user's productivity. Does making a user worry about keeping his mailbox below a certain size make him more productive? Does searching looking through multiple PSTs to find some long lost email make a user more productive? Does the loss of email in PSTs due to a hard drive crash or laptop theft increase productivity? The more limitations you have to impose on users just to allow your system to stay up and running severely hinder their productivity. It is the same problem when companies try to implement "kill all email older than x days" policies to try to reduce their exposure to litigation. The sooner the management of a company sees email as a productivity tool and treats it that way, the better off users will be.
If you can afford the hardware, there should be no limitation on mailbox sizes. Since Domino does field level replication, the size of a mail file will have no impact on the amount of data being transmitted over the network during se rver to server or client to server replication. One thing you do have to keep in mind is that disk is not the only cost, but backup media, backup time, and restore time is also impacted when the size of the mail stores grow. Depending on your backup solution, SLAs may have to be written to take mail file sizes into account.