In most circumstances Ava is the primary repository of business
knowledge for either temp agency staffing or staff bank activity.
It is the core of our customers' profitability and continued
existence. It's at the heart of the reasons why to choose Ava. Over
the years Ava has consistently had the outstanding record for
reliability and security. Accurately measured in a large and
complex environments Ava has continuously out performed in the 98%
of other supplied software irrespective of size and complexity and
matched the remaining product.
Allied to Ava's security and reliability is the process of
backing up data.
This is probably best done with the free application SQLBackupAndFTP. You should also have an email account set up so that notifications can be sent to you.
Where a bespoke database administrator post is overkill and an
unnecessary overhead: Ava provides advice on backup strategy. This
article is part of that advice.
Setting up backups and maintenance plans is essential for data
security and continued high performance. This process is different
in MS SQL 2008 having evolved over previous versions of MS SQL.
This article takes the IT savvy user simply through the multistep
process of setting up the basic maintenance plans and notifications
necessary in a well run system.
Ensure that you are up to date with SQL2008 service packs. There
was an issue which concerned the following setup procedure which
was resolved in a very early patch to the RTM version of SQL
2008.The patch is included in SQL 2008 SP1. Ava recommend
that if you are applying SP1, you also apply the latest cumulative
update for SP1 namely CU3 SP1 CU3
Create a folder on a physically separate drive (from the database
and log files) mounted on your server. This is for your backup
files. You don't really need to worry about fragmentation of this
drive (to a point). It should not be compressed (this will slow the
backup process down). It must have sufficient free capacity for all
your (Ava / SQL) backups.
It is certain that if the extremely rare, serious error, occurs:
you want to know about it as soon as possible. SQL 2008
standard edition (upwards) allows for database mail to send email
notifications to your (IT) network support or other designated
persons. In all probability a problem of this nature with MS SQL
will actually be caused by a problem with your network or will be
network related. It makes sense for all these error messages to go
to one person or place.
(Note: The Ava application has an entirely separate process
which is designed to give us specific information to help develop
our software and refine new features.)
The following information is required for the database email
system:

To set this up:
Log in to SQL server with SQL Server Management Studio
(SSMS)
Expand the management node in the top left pane and right mouse
click on 'Database mail'

A wizard will be launched. Click next.

If this is the first time, you will be asked if you want to enable
database mail. Seems redundant to say this is a major part of the
object of the exercise!

You will now get the opportunity to create an email profile for
any messages you need. I would suggest that you might have
two profiles, one for the "oh F$%!!$%" messages which goes straight
to a VIP's mobile and one for system ok notifications which
probably goes to a mail box where it probably isn't touched, but
may be useful at some time in the future for finding the date/ time
when a process notified of successful completion.
We are interested in the "oh F$%!!$%" messages here. The profile
should be labelled appropriately. For your information "oh F$%!!$%"
is not a registered trade mark of Ava and may be used as an account
name / description without copyright referral.

Now enter the smtp account details

click OK
Once again:
If not logged in, log in to SQL server with SQL Server Management
Studio (SSMS)
If not expanded, expand the management node in the top left pane
and right mouse click on 'Maintenance Plans'

A wizard will be launched. Click next.

You now get the opportunity to label you plan: you could call it
"I have a cunning plan 1", but better something like
"DatabseName(s) full backup". Seems redundant, but a comment
like "after office close" could be helpful (especially if office
hours change).
Then click on the [change] button: this process needs to be scheduled to actually occur at some (repeated) point in a day.

The default schedule is not Daily. Unless you have a system where
there is not sufficient time to schedule a daily full backup (in
which case I'd be worried if you were reading this for anything but
the jokes!) do it daily backup.
By default it will happen at midnight. There might be a better
time, especially if some tape backup kicks off before midnight.
Tape backup should occur after database backup completes allowing
the backup to be quickly copied to an alternate repository.
Click ok and you will be returned to the main wizard work flow.

It will now show a description of the schedule.

Click next.
You now get a choice of a bunch of tasks.

If your database is in full recovery mode, then you will need to
do a separate maintenance wizard run for log backup. So repeat this
wizard process but use a frequency of (say) every 5 minutes. If it
is in simple recovery mode: this additional (log backup) step is
redundant.
You should also do a maintenance wizard run for clean up tasks,
such as deleting old backups to reclaim disk space.
You should also have tasks to check database integrity (this can
and should be done prior to backup as a database which does not
pass database consistency checks (dbcc) can backup (good) but the
backup may not restore (very very bad and it renders the backup
process almost useless).
Restoring a backup which errors in the restore process and fails
to complete.
The next window allows you to re-order tasks if they are run on
the same schedule. This may happen if you've done this wizard for
multiple tasks run on a single schedule. It is the default option
and way back earlier in the article we showed it as the
option.
Here we are concerned with the first run for the basic backup
task.

Choose the database to back-up. There is no point in backing up
'tempdb' (this is recreated every time the server restarts). Back
up 'master' , 'msdb' and 'model'. It should be pointed out that
these databases only change under specific circumstances. If this
instance of SQL server is just running the Ava software, then a
single backup up 'model' is sufficient as Ava does not implement
the functionality of 'model' because of possible conflict with
other applications which may rely on the 'model' database. For more
information on the purpose of these databases this article was written for
SQL7 and its still totally relevant today.

Also you must choose the location of the backup databases. Raid 0
is a good choice for this drive.
What is raid?
http://en.wikipedia.org/wiki/RAID
http://en.wikipedia.org/wiki/Standard_RAID_levels
The location should be mounted on the server.

You can backup to a unc path,
however this will probably hammer the network and in addition if
the backup path then fails, could have a substantial continued
negative effect on the SQL service. In addition the SQL
service must be configured to have network access and this may give
rise to a security issue.
It is best to have a physically separate local hard drive. Please
note, drives are really cheap. A 1 Tera Byte drive can cost
less than £100. Put it another way 10p per GB or on a 4GB database
backup: 40p (and this is reusable). Putting the backup on a
drive which is a logical split of the database files drive (or
array of drives) is a very false economy. If the data drive goes in
this situation then so does the backup. That defeats the whole
point of having a backup.
Ideally the drive should be on a separate disk controller (or
controller channel) this will minimise any impact on database
performance if it becomes necessary to backup (for some reason)
whist normal operations are in progress.
The disk backup can then go to tape or other in a leisurely
fashion.
Its always good to keep a report of these things and its always wise to clean out reports that are simply too old to be useful - so a cleanup task for reports (later).

Click next

Finish (first time anyway)
Have I set up maintenance plans for:
Database dbcc check
Database backup
Log backup
Update stats
Reorganise indexes
Rebuild indexes
cleanup task for database backup
cleanup task for log backup
cleanup task for reports
cleanup history