When implementing Enterprise Vault for Exchange Server archiving a question often arises in the design phase relating to when scheduled archiving should take place. There are many different schools-of-thought when it comes to picking the ‘right’ time to do it. Is there, in fact, a right time? There are certainly many things to consider so let’s discuss a few of those here, and hopefully it’ll give you food for thought when thinking about when the task should be run in your environment, or an environment that you are designing a solution for.
This list is not in a particular order, they’re just chunks of things that should be considered to help reach a decision.
Geography
One of the things that comes to mind relating to when scheduled archiving should be run is to look at the geography of the environment, and the users who are accessing it. Are the users all based in a single office, or are they spread around a country, a continent or the world?
If users are all based in one geography then the decisions of when to run the archiving may be easier, since it will be easier to figure out when users need access to archived data and therefore the system needing to be most responsive.
If a centralised infrastructure supports users all over the world, then there may not be any ‘quiet’ time. In fact the operation may be 24x7 with little difference in load between the ‘day’ and ‘night’ wherever the infrastructure is located.
A centralised Enterprise Vault infrastructure picking data from remote Exchange servers servicing users in different geographies might also be something that exists in the organisation, and that too can be a little bit easier… unless the users in those remote locations are 24 x 7.
System Load on EV
The point of the scheduled archiving is that it will generate significant load on the Enterprise Vault environment itself. I’m sure many people will have seen that retrieval of items, or Vault Cache builds are somewhat slower when the Enterprise Vault system is busy doing an archiving run over several thousand mailboxes.
One thing that we should try to avoid is impacting the end user experience of getting data out of the archive. It’s also the reason why we should spend time to investigate and use technologies which make backing up the environment faster. Unfortunately there isn’t much we can do in terms of making the scheduled archiving faster.
System Load on other machines
Load is not only generated on the Enterprise Vault environment there is the usual array of user lookups against global catalogue servers, but more importantly there is load on the network, and on the Exchange servers which are being targeted of archiving.
This load should not be underestimated, and should be reviewed from time to time. It’s not just straight system based load there will also be an increase in transaction logs and possibly an increase in size of the Exchange databases themselves, depending on several archiving and storage factors.
These other systems will also have backup cycles that they need to complete. Picking when to do the archiving run in relation to those other systems is a tricky topic in itself, and I have written about that before: https://www-secure.symantec.com/connect/articles/finding-time-archive-mailboxes
Types of user
When looking at when to schedule archiving it’s also important to review the quantity and types of user that exist in the organisation. 9 to 5 office workers would be great, but almost every organisation has a large mixture of different types of user, and the days of 9 to 5 office work seem to be in the past in almost all companies.
Aside from out of hours people accessing the environment the business itself might be 24 x 7, in this situations the load generated may be very similar every hour of the day. (Compare that with a more traditional approach and the access out of hours might be somewhat ‘less’).
This also ties in with geographies. The company itself might not have people in every location 24 x 7, but, they have one office covering one timezone between 9 and 5, and then a different office in a different geography covering 9 to 5. This is all good, especially if the Exchange infrastructure is different for those different geographies, but it is more difficult if the users are accessing centralised infrastructure.
How long do we have?
The final thing to think about is how long the scheduled can run for. It might not be possible to run the scheduled archiving for more than a few hours each day. Some days, though, it might be possible to run the task for much longer.
In other words don’t just consider 6-9pm Monday to Friday.. Consider that perhaps a shorter window is possible during the normal working week, and a longer period is used at the weekends in order to ‘catch up’.
One things for sure that whatever window you choose, on whichever days you choose to run it on, you will need to check the archiving reports which are generated to make sure that the task is getting through all of the mailboxes as often as possible. If a mailbox isn’t ‘reached’, then it will not have been subject to any archiving, and that might lead to problems relating to mailbox quota.
Conclusion
As you can see there are many different things to take into account when thinking about when scheduled archiving can run in a particular environment. One things for sure that whatever time frame you pick you’ll need to come back and revisit your assumptions and choices from time to time. I would suggest doing that every six months or so. Businesses change quite quickly, and so it is a good idea to check that the schedule chosen still fits the needs of the organisation.
When do you run your scheduled archiving? Do you have it the same every day, or longer on some days? Let me know in the comments.
