ng their CDP capabilities.
But what is CDP really? It is a method or process of performing
data backup and in many cases it is a feature of an existing
backup product. To provide Continuous Data Protection, a
backup program must do two things:
1. Backup data as the data is written to the hard drive.
2. And, the backup program must keep revisions.
What is the goal of Continuous Data Protection?
The goal of Continuous Data Protection is to constantly backup
data as the data is created or modified. So, a true CDP solution
eliminates the possibility of losing data between backup sessions
because in theory, no time goes by between the backup
sessions. A backup is always occurring to proactively protect all
new data.
What are the problems that CDP solves?
Lets look at what happens when you perform a single backup
every night at midnight. Employees come in and create and
modify data throughout the day. All of the work that has been
performed during the day is at risk until it is backed up at
midnight. What happens if you have a hard drive failure at
lunch time? In this situation, all of the data created this morning
is lost because you can only restore to the point of the last
backup.
CDP solves this problem by backing up the data on-the-fly; as
the data is created and changed. So, in this scenario, the work
during the morning would be recoverable because the CDP
backup features would be capturing the data throughout the
morning.
What are the problems that CDP creates?
Using a CDP based solution creates several issues that can
become large problems in a relatively short time frame.
Consider the following before choosing to
implement a CDP solution.
Increased Load on Resources
The first issue that comes up is the
increase in network and individual
computer resource load. Since all data
that is written to the hard drive has to be
backed up to a storage device at the same
time; you have effectively doubled the
load on the computer running the CDP
solution. In other words, every time data
is written to the hard drive a backup event
is occurring. An increase in the amount
of data written to disk results in an equal
increase in the load put on the computer
because this simultaneous backup process
is running.
It is important to note that the computers
that have the greatest need for CDP tend
to be those computers that are under the
greatest load demand to begin with. The
increased load caused by the CDP process
may overwhelm these computer
resources, resulting in a disruption in their
normal operation.
Difficulty with Revision
Management
Another issue that you can encounter with
a CDP based solution relates to file
revision management. Lets assume you
have a database that is continually
updated. Since this database is
continually changing, the CDP system is
backing up a near infinite amount of little
changes, as they occur. At some point,
these minute changes need to be pulled
together to form a revision of the database
Page 1 of 3 Engineering
Highly Reliable Systems does not gaurantee this information to be accurate or fit for any purpose. Use this information at your own risk. Copyright (c) 2006 by Highly Reliable Systems. All Rights Reserved.
WHITE PAPERS
in order to fulfill the requirement of point-in-time restoration.
But at what point do you assemble these small changes to represent
one, single revision? Lets assume your CDP solution creates
restore points every 10 minutes. The question becomes, How
many restore points can be stored given our current storage
capacity? If 25 revision points may be stored, then your system
will allow you to restore to any point in the last 25 revisions. These
revisions are created every 10 minutes, which covers a little over 4
hours. Anything beyond the 4 hour time frame is lost. So, if your
database becomes corrupted, you have a 4 hour window until every
revision held by the CDP backup system is also corrupted;
eliminating your ability to recover a clean copy of your database.
What is Near CDP?
You have probably heard of the term Near CDP. Near CDP is
defined as a backup solution that performs the backup process
when a file closes. The key difference between CDP and Near
CDP products is that True CDP backs up all of the changes to files
as they occur whereas Near CDP performs one backup at the time
the file is closed and editing is complete.
There are 2 key advantages that Near CDP products have over
True CDP products. The first advantage is that with Near CDP
products you have a significant reduction in load on the computer
being backed up. The second advantage is that the problems with
True CDPs revision management are cleared up because the
revisions are clearly tied to file close events.
But, there are problems with Near CDP solutions. The single
largest problem with Near CDP occurs when a file is continuously
edited and never closes. For example, Microsofts Exchange
Server never closes the Mailbox Stores. As a result, the Near CDP
backup trigger never occurs because the files are never closed and
therefore they are never backed up. Another example of files that
are continuously in use is seen in MS-SQL files. Again, these files
are ignored by the Near CDP solution because they never close to
trigger the backup process.
As a work around, the Near CDP solutions also carry a backup
scheduler to backup data that is overlooked by the file closing
trigger. So effectively, the Near CDP
solution operates as a traditional scheduled
backup program does and you do not
enjoy any of the CDP-related benefits for
files that are continuously in use.
It is important to point out that Near CDP
solutions can introduce a disruption in the
normal operation of your production
servers. Since the Near CDP backup is
triggered by a file close event, if an
application closes a file, and then re-opens
the file, that application may be denied
access to the re-opening of the file. This
can happen because the first close event
triggers the Near CDP solution to start the
backup and then your request to re-open
the file is denied because the backup
program has locked the file during the
backup process. For example, you may
update data in an accounting program and
when you post your update, the accounting
program may save its files and then
generate a report based on your new
information. The report production may
be disrupted because the Near CDP
backup program has gained access to the
account program data.
Compare CDP to other backup
methods.
Lets compare CDP backup solutions to 2
other popular data protection methods:
schedule-based backup and mirroring.
CDP and Scheduled File Backup
Weve mentioned that True CDP operates
on-the-fly, so therefore it is always
working and does not follow a schedule to
initiate the backup process. Both True
Page 2 of 3 Engineering
Highly Reliable Systems does not gaurantee this information to be accurate or fit for any purpose. Use this information at your own risk. Copyright (c) 2006 by Highly Reliable Systems. All Rights Reserved.
WHITE PAPERS
CDP and Near CDP solutions create problems of their own when it
comes to backing up files that are in continuous use. Schedule
based backup overcomes these issues by allowing you to determine
when and how frequently you want to backup. And schedule
based backup programs can more clearly correlate backup events
with revision management.
CDP and Mirroring
It is important to note that while both mirroring and backup are
both data protection methods, they are not one and the same.
Mirroring refers to the process of making a duplicate copy of data
to a second storage device. The key difference between mirroring
and backup is that mirrored data is a straight copy of the original
files and backup programs tend to encode the stored copy using
some combination of encryption and compression which
necessitates using a restore program to gain access to the backup
data. Generally speaking, mirrored data is a copy of the raw data
which is immediately ready for use and does not require you to use
a restore program to gain access to it. Mirroring differs from CDP
because mirroring typically does not contain revisioning. In a
mirrored environment, when an original file changes, the mirror
copy is overwritten with the new file. In a CDP environment, the
new file is added to the revision history and the data may be
encrypted and compressed before it is stored.
When is CDP interesting and when should it be
avoided?
If you have been reading all the latest industry news about CDP,
you may be attracted to these solutions because you want to
minimize the amount of data that can be lost since the last backup
occurred. If your business has a critical need to protect data on-
the-fly, as the data changes, you may consider a CDP solution to be
the answer. But, be aware that the CDP solution may also
introduce new problems; such as increased load on network
resources and the shortening of the time window to restore clean
data in the event of corruption.
It is very common for businesses to implement multiple data
protection (backup) solutions to provide a greater level of data
protection. Most companies find that one
solution does not fit all of the needs of their
computing environment. For example, you
may have a few servers for which CDP
makes sense and the rest of your servers are
best protected with traditional schedule
based backup. And some businesses
implement more than one backup sol