On 1/2/08, Tim Howe wrote:
At least Microsoft has single storage of duplicated attachments (although I haven't heard how intelligent it is, e.g., if the same document is attached to a different message rather than forwarded).
I think it only does that on message receipt. So, if you send a message to ten people, and then send another copy of the exact same message (with the exact same attachments) to ten other people, you still get multiple copies stored.
I have recently interviewed with a company that is doing long-term message archiving & handling e-discovery for their clients (think financial trading firms that have seven year storage requirements), and they have a special process they've developed to go in and actively harvest all those extra copies of the same attachment and coalesce those into just one copy with multiple links.
Nick Christenson and I had looked into doing the same kind of thing back in 2000, for the invited talk that I presented at LISA that year. But we couldn't make a good case for it, since disk capacities are continuing to increase at unbelievable rates but disk I/O throughput is not, and you'd be trading space for additional I/O operations, which would be exactly the reverse of what you'd want to do for a scalable e-mail system.
But this company I interviewed with sees this as a way to reduce help desk calls from all these customers who've signed contracts for X amount of storage for Y number of years, and they need to keep the information archived while also reducing their disk quota problems. Nick and I hadn't seen the cost of the customer support side of this issue, so it didn't figure into our equations.
If only the message store didn't fall apart given a slight breeze.
Indeed. ;( -- Brad Knowles <brad@xxxxxxxxxxxxxxxxx> LinkedIn Profile: <http://tinyurl.com/y8kpxu>