Handshake to Paul Robichaux on Exchange Storage
The write-up of my experience with Exchange storage usage has engaged Paul Robichaux, and I wanted to address his insights and tie up some loose ends. As an author of many technical texts on Microsoft Windows and Exchange Server, he is quite an established expert. He and I may not agree on the value of different collaboration platforms, but I extend a handshake to him across the blogosphere.
After I had posted on some observations of the Exchange data store, Ed Brill picked it up, and then commented on Paul's blog about it. So, now I'm working my way through Paul's blog, my blog and Ed's blog—to weave together some of the strongest responses.
-
Some of the comments have been careful to include view indexes as part of the Domino store (the NIF). Which is true. A mail file can be inflated by building all the indexes, and especially by creating a full-text index with every option selected. So, ok, it is technically supported to rebuild all indexes. Domino files can get huge, and not just from unnecessary views. I'm sometimes jealous of the Exchange admins, because they can just say “no.” I, on the other hand, might need to support user mail files that exceed 8 Gs. Nevertheless, (1) Domino does support larger mail files than any other platform, and (2) the idea of pointing out the view-indexing mis-use is not really anything other than identifying deliberately bad administration. It reminds me of the admin that allowed anonymous access to the Domino directory and then loaded up HTTP (this was written up in InfoWorld). Bad idea.
-
The point of my original entry, was to note that on the same piece of hardware that hosted Exchange, with Domino it actually used less disk storage, and then I pointed out the there are some other peculiarities with the Exchange data store. I'm not insisting I performed a wide-ranging study on different setups and migrations. I'm just stating what I witnessed, that in this instance the data store was smaller for Domino than for Exchange. Interestingly, nobody bothered to respond about my observation on CPU resources, which were hugely disparate.
- Paul notes that for the most part, Exchange doesn't need to be compacted, because it reuses the whitespace. So does Domino. However, I think it is interesting that with Domino we can turn off the database properties to overwrite free space to speed up the transactions. With Domino, the attribute of overwriting the free space is more a security feature than an enhancement for disk efficiency. But, to suggest that Exchange doesn't really need to be compacted reminds me of how for years Microsoft said their disk system didn't need defragmentation. It's safe to say that Exchange servers can run a long time without compacting their databases (which I've seen with Domino servers, as well), but I can't imagine that anyone sees forestalling compacting as a best practice.
Natually, Domino and Exchange have their unique architectures, and the significant differences are so much more than the message store (even though I thought it was an interesting measurement). Right here, in this dialogue about Domino and Exchange with Paul, Ed and I, we can see the bigger distinctions. Here's how I would demonstrate what makes Domino, Domino.
Paul is as smart as they come when it comes to Microsoft technology. That's not an opinion or cheap flattery—it's a fact. I would love to have an afternoon with him to learn what he could tell me. However, Paul's blog doesn't use Microsoft's collaboration platform with Exchange, but runs on Linux and PHP .
My blog runs on Windows 2003 using Domino and an open-source Notes database from OpenNTF. When I get a new post or comment, an e-mail is sent to me. I have the blog replicated on my laptop so that I can enter drafts when I'm offline (or review the referrals). I can use a Macintosh, Windows, or Linux Notes client.
With Domino, I can walk the walk.
Addendum: Thanks to Brian Green for catching a syntax error.
- 


Comments
Posted by Paul Robichaux At 11:55:05 AM On 08/18/2006 | - Website - |