SIS and collection files

[Reginald_D_Souz]

I had a question regarding SIS (Single instance storage) and how it works with collection files. Question 1. We do archiving based on quota limits. A mail with an attachment is sent out to 10 users. Now since archiving is done based on quota, suppose the mail (which was sent to 10 users) of only 2 users gets archived and the remaining users doesn't get archived. And after 10 days the collection file runs and all the archive files get collected into a cab file. What would happen after 15 days when if remaining users mail gets archived. Would the mail be still treated as SIS or would a new file be created for that or would it point to the file within the cab.
Question 2. If a user would like to restore a mail which is collected in a cab file; does the whole cab file get extracted or just that mail. And what would happen during the next cycle that the mail gets archived. What would happen.

[Tony_Uren]

Reginald,

Q1. I am unsure myself. Keep an eye out for the answer, I will be.

Q2. I am pretty sure that the whole cab gets extracted. This is then available for the user to access the item in question. This extracted cab is only temporary though. At the next archiving run, the system recognises it as a temp archive (it is extracted with a different file extension I think) and deletes the temp archive. The original cab is left as is. I am sure there are a few technotes on this.

If I am wrong I am sure someone will correct me. ;)

[Micah_Wyenn]

Hmmm,

1) So a quick disclaimer...SIS is complicated, and I'm probably wrong about what I'm about to say. Your mileage will vary. As far as I've been told, as long as the vault can update the metadata in both SQL and on an individual DVS, then SIS "can" work. It's very dependant on the items header and body being the same to the vault. I've seen it work where I expected it, and then sometimes not work where I expected it. So I can't really claim to know either way. But still, cabbing a file shouldn't prevent SIS. Now the caveat with that is, that it won't SIS on a closed partition (I believe).

2) Sounds right to me. :)

[Michael_Bilsbor]

ok, shall we get definative :)
had a question regarding SIS (Single instance storage) and how it works with collection files. Question 1. We do archiving based on quota limits. A mail with an attachment is sent out to 10 users. Now since archiving is done based on quota, suppose the mail (which was sent to 10 users) of only 2 users gets archived and the remaining users doesn't get archived. And after 10 days the collection file runs and all the archive files get collected into a cab file. What would happen after 15 days when if remaining users mail gets archived. Would the mail be still treated as SIS or would a new file be created for that or would it point to the file within the cab.

A new file would be created. For EV to SIS we have to update the DVS file (with lots of per user data). Since you cannot update CABs then the DVS you get a new DVS file when the item gets archived and a new CAB later on down the line when it gets collected. Remember, if we did keep on updateing the CAB it would continue requiring backup which is one thing you're trying to avoid by using collections.

Question 2. If a user would like to restore a mail which is collected in a cab file; does the whole cab file get extracted or just that mail. And what would happen during the next cycle that the mail gets archived. What would happen.

We open up the cab and just pull the item out, we don't 'undo' the whole collection or anything like that. I think we store this item temporarily as a .archdvs file so if user views the item again then we don't have to open the cab.

Your question1. is a good one and is one reason to look to use partitions over collections.

ok?
Mike

[Micah_Wyenn]

Ug...collections just got a whole lot less useful in my eyes. Time to email my senators. Why can't you just tug the cab outa the collection, update the metadata, and toss it back in?

micah

[Michael_Bilsbor]

1. you cannot update a cab no windows cab api for that
2. If you change the cab it needs backing up. which is what you wanted to try and avoid...

That's why it's important to get the number of days before items are collected right. The greater the value the longer before the item gets collected and therefore the greater amount of SIS that will occur till it goes into a cab.