Benefits of UNICODE vs ANSI PST file format - are there any?

[Conor McKeown]

The only 'advantages' I can see are:

1. Support for multiple character sets
2. 'Unlimited' PST/OST file size

Point 1 is valid, although I'm not really sure that this would be enough of
a reason for us to migrate our current 6000 users to the new format. In the
next 12 months we will be looking at migrating another 40,000 users to
Outlook from mix of various POP & IMAP clients and while we could easily
start these users off on UNICODE format PST/OST files, trying to convert
6000 existing users represents a serious dilemma for us. We already have
about 5000 users in Asia Pacific using Outlook 2000 so I don't think they
have missed not having multiple character sets.

Point 2 is something that I would appreciate as a home user, savvy enough to
know how to check for errors regularly etc, although after many years of
using Outlook I only have 3 * 500MB PST files on my home machines. On my
current work machine I only have 1GB of PST and that's after nearly 2 years
here. However, at least 20% of our current 6000 Outlook XP users fill up 2GB
worth in 3-6 months and perhaps another 50% fill up 2GB of PST every 12
months. The remainder have usage similar to my own.

We are a bit of a unique company in that over 95% of our users only use a
laptop and all data is stored locally (synched from E2K server + PST files,
+ all other data) - that's another issue...

In the enterprise environment that I support, having PST files larger than
2GB would solve a lot of problems that our users hate - i.e. having to
spread their data over multiple PST files, but it creates other problems for
us:

a) Time to restore from backup system in event of disk failure / stolen
laptop.
b) As we have seen, the performance accessing PST files becomes quite poor
when they get reach >1.2GB

I know that we can limit the size of UNICODE PST files and if we do start
using that format, we will most enforce a 2GB (maybe 5GB) limit, but what I
really want to know is the following:

i) Is there any performance benefit of UNICODE over ANSI PST file format?
i.e. quicker indexing, faster reads/writes?
ii) Any other benefits?

Please responses post to the list - address is invalid.

Regards,

Conor.

 

[Charles Beauregard [MSFT]]

Here are some highlights of what has been done in Outlook 2003 to the
Unicode (Old) and Ansi (New) PSTs:

- The block and page cache can grow larger than they did before. This
means that there is a better chance that the data will be retrieved from the
cache thus increasing read performance
- We now use multiple (5) indices for tables. In practical terms, this
means that performance is greatly increased when switching folders or
sorting a table, since there are more chances that the current sort exists
in a cache .
- Table indices are no longer recreated unless the table has changed
significantly; this means that an index is rarely recreated from scratch.
- Indices no longer stored for small tables --> reduces overhead when not
needed.
- New table notification algorithm makes addition of n items to a folder
take O(n log n) time vs. O(n). Especially noticeable in folder with lots of
items.
- Larger sort buffer: no longer limited to a 64k buffer, making sorting
much faster in folders with lots of items.
- PST grows by blocks of 256k (up to 20Mb) and 2Mb blocks after the PST
reaches 20Mb. This significantly reduces on-disk fragmentation because the
PST used to assign hard disk space using a "Just In Time" and "Just Enough"
algorithm.
- Implemented NID recycling, meaning that a PST can now have more than 128
Million items during its lifetime, (but still no more than 128 Million items
at a time)

The above changes were made for BOTH Unicode and Ansi PSTs. This means that
in most cases, the Ansi pst will be (marginally) faster than the Unicode pst
because Ansi data is lightweight.

Here are some changes that are specific to the Unicode PST:
- Size up to 33 TeraBytes (33,000 GigaBytes)
- Unlimited amount of items per folder (used to be 64k)
- Full Unicode support

Hope this helps

--
Charles Beauregard-Tellier
Software Test Engineer
Microsoft Office

This posting is provided "AS IS" with no warranties, and confers no rights.
Please do not send e-mail directly to this alias. This alias is for
newsgroup purposes only.