Every major release of Notes/Domino (i.e. R2, R3, R4, R5) has included
significant architectural changes to the database structure, also known as the
On-Disk Structure (ODS). These architectural changes enable us to improve
performance and scalability in each release. Upgrading the databases provides
a lot of benefits, with very low risk, so this should be an important early
step in your R5 upgrade plan.
A common question many administrators ask is, "What happens in a mixed
environment?". R5 databases (i.e. those in the R5 database format) can be
accessed from R4 clients and servers in almost all cases. For example,
replication and mail routing works normally between servers. Since the ODS
format does not replicate (it's local to the system it's stored on, just like
full text indexes), it will not have an impact on other servers and clients.
So, when an R4 client accesses a database on the R5 server, everything works as
expected. The exception to this is if the database uses new R5 design features
-- those features will either be ignored or will generate an error message.
The only thing you need to remember is that the R4 code cannot read the R5
database format locally. What does that mean exactly? Let me give you an
example. If someone creates a database using the R5 client and sends it to you
via email as an attachment, you will not be able to access it from an R4 client
after you've detached it. When you try to open the database, you will get the
error, "Invalid NSF version". However, if the person puts the database on an
R5 server instead, you could access it from your R4 client and make a new copy
or replica. Your local replica would be stored in R4 format on your system and
would continue to replicate seamlessly with the server. That's it. There are
no other known issues with upgrading the database structure.
Of course, there are lots of advantages to the new database structure. In
addition to improved performance and scalability, there are several features
you can enable. Once the database has been compacted once, you can take
advantage of the new on-line, in-place compaction feature. You can also enable
transaction logging, which further enhances performance. Several new database
properties, such as disabling unread marks for large databases can also be
enabled. And this is just the tip of the iceberg.
Okay, so how do I upgrade my databases? Simply run compact and the conversion
is done automatically. Since many administrators run compact on a regular
basis to reclaim white space, you may not even need to do this task
R5 makes it easy to see what version of the ODS your databases are in. From
the Notes client, open Database Properties, and the ODS version is noted on the
Information tab. The ODS version for the shipping release of R5 is 41; the ODS
version for all releases of R4 is 20. We make it even easier in the R5
Administrator Client. From the Files tab of the Admin Client, a column is
displayed for File Format. The release is noted with the ODS version number in
parentheses, e.g. R5 (41:0). This should give you confidence in upgrading the
database structure now that you can tell what version is being used.
So, what do you do if you decide you absolutely need the database in R4 format
(for example, to send in that email attachment)? You have several options to
revert the database structure to a prior release.
1) From the server console, execute the following command:
2) From the operating system prompt, issue the following command(s), depending
on your operating system
ncompact.exe -r for Windows NT (Intel)
acompact.exe -r for Windows NT (Alpha)
/opt/lotus/bin/compact -r for UNIX
3) From the Admin client, access the Files tab, highlight the database(s) you
want to convert, and select Databases from the panel on the right. Click on
Compact, check the box for the "Keep or revert database back to R4 format"
setting and click OK.
4) From the R5 Notes client, make a copy or replica copy of the database and
give it the extension .NS4
It's important to make the distinction between upgrading databases and
upgrading applications. Upgrading databases involves compactin
This was first published in November 2000