By Mark Ferelli
The data backup world has changed dramatically in recent years. No change has been more dramatic or rapid than the shift from traditional tape-based backup technology to disk-to-disk (D2D) backup. Disk-based backup has enabled shorter backup windows and more rapid data recovery which has opened the way for more sophisticated backup and recovery software technologies that were not possible with tape backup systems. Software vendors have responded to the technology potential of disk-based backup with new enhanced functionality, such as point-in-time snapshots and local and remote replication in an effort to reduce the vulnerability of data loss in between scheduled backup sessions.
Beyond the pure speed advantages, disk backup is also the right technology at the right time to address the convergence of two business trends: the necessity for 24/7 data access in a global wired economy and the increasing use and importance of remote offices, meaning that more remote data is at risk than ever. According to the Enterprise Strategy Group an estimated 60 to 70 per cent of mission-critical data is stored and used at offsite locations. Enterprise IT managers face the challenge of how to protect and manage all remote data in an era of tight budget constraints and the reality that the geographically distributed locations typically lack the IT staff to manage and monitor backup systems and to verify that the backup operations were completed successfully.
Against this backdrop of vulnerable data stored at remote offices, IT managers are also facing a fundamental change in the concept of backup data. Disk-based backup has largely removed the issue of the backup window and the focus has moved to what has always been the most critical aspect of the backup process: data recovery. Driven by business objectives in terms of application and data availability, and the disaster recovery and business continuity planning, as well as compliance with government regulations and legal discovery, data backup processes are now focused on two new metrics: Recovery Time Objective (RTO) and Recovery Point Objective (RPO). The RTO establishes a maximum duration for how long the restore process will take while the RPO sets a goal for the maximum age of the data that will be used for a restore operation. The challenge is to drive the RTO to as close to zero as possible and to have the RPO be minimal, while being able to afford it.
Continuous Data Protection (CDP) Gains Momentum
The disk-based backup and recovery strategy gaining traction in data centers of various sizes to meet the RPO and RTO challenge is CDP. The traction is especially visible among users of Exchange, where the management and compliance challenges are driving elements of the CDP marketplace.
As with any new technology there is not yet universal agreement in how CDP is designed, deployed or even an exact definition of the technology. International Data Corporation defined it this way:
Continuous data protection (CDP), also referred to as continuous backup, pertains to products that track and save data to disk so that information can be recovered from any point in time, even minutes ago. CDP uses technology to continuously capture updates to data in real time or near real time, offering data recovery in a matter of seconds. The objectives of CDP are to minimise exposure to data loss and shorten time to recover.
While many echo this definition, not all makers and users of CDP are on board. The underlying problem, it seems, is the issue of granularity how many points are needed (in a given time) for protection to be considered continuous? A competing definition from the Storage Networking Industry Association (SNIA) reads:
A methodology that continuously captures or tracks data modifications and stores changes independent of the primary data, enabling recovery points from any point in the past.
A CDP product is one that will continuously monitor an object for changes and will preserve copies of all prior versions of the object. The user will have the ability to view and access these prior versions as required. The time to perform recovery changes from hours or days to seconds or minutes. The backup window is no longer a problem because there is no longer the concept of a backup window.
Specifically, though, CDP is a cross between disk-based backup and replication. CDP continually captures all changes made to a file, and engages in tagging (versioning) objects so that they can be specifically rolled back to a particular point in time. The business value of CDP lies in the ability to restore data objects to a point before a data corruption or interruption event takes place. If you experience a data loss at 12:34:14 am, a storage admin can restore back to that particular time, or somewhere very close. CDP protects/captures data as it is written to disk. One of the great myths of CDP is the unspoken assertion that CDP is for every kind of data, all the time. This is of course untrue, since the value of data changes as a matter of time, urgency, and business dynamics. A vital document today could well be worthless three days from now, and important again in another six months.
There are two approaches to CDP object protection: file-system centric or block-based. Historically, the CDP file system approach started on the Windows environment, where most applications utilised files to hold their data. Block based CDP started in the UNIX (and now Linux) community, where database applications traditionally bypassed the file system and operated directly at the disk/block level.
Block-based CDP products started in intelligent arrays such as those from EMC and HDS and soon moved to host-based or appliance-based platforms. These CDP products operate as a layered feature of the underlying storage infrastructure, and usually operate independently of the host's file system and volume manager. Recovery is typically storage or database administrator driven, is provided through capabilities outside of the platform being protected, and is managed by the CDP implementation.
One important scenario to keep in mind when considering the implementation of CDP is that of centralised backup for the remote or branch office. Too often, basic IT tasks like monitoring the backup server and changing tapes can be missed when assigned to remote office clerical staff not skilled in IT. Using a CDP strategy over the WAN to protect branch office file servers removes the requirement for tape drive and media handling at the remote site.
What about Recovery?
There are two general principles that govern all recovery policy-making: the recovery point objective (RPO) and the recovery time objective (RTO). The RPO defines how much data you are willing to lose when you recover data. For example, if you back up twice daily your RPO would be 12 hours, which is the maximum amount of data loss that could occur between backup images. The RTO defines how long it will take to recover your business processes from a data failure. This includes not only the data recovery, but restarting the servers or applications that depend on that data. These recovery considerations must also be applied to local and remote recovery strategies.
A true CDP product protects every data change as it takes place, and the RPO approaches zero. On the other hand, with the vast amount of data being recoverable, how you choose the recovery point affects your RTO.
Some recovery points are based on time, a particular hour or minute. More usefully, however, they are event-based. Since every data change is protected, a loss event can be absorbed and yield a recovery event.
Implementing CDP
Along with the debate over an exact definition of CDP there is no agreement on the best method to deploy CDP within an organisations storage and data protection infrastructure, how to approach what is actually protected and the level of granularity required. CDP solutions are designed to be block-based, file-based or application-based. Block- and file-based CDP solutions have the advantage of functioning with a range of different applications, while application-based CDP is optimised and tightly integrated with a specific application, such as Microsoft Exchange. Potential CDP buyers should also be aware of the level of recovery granularity a particular CDP solution provides, as all CDP products are not created equal on this issue. Some products only support recovery of servers, volumes or folders and lack the granularity to recover a single file or email message.
CDP is deployed most frequently either as an appliance or as a software solution running on a server or switch with agents. A dedicated CDP appliance can deliver good performance without impacting application servers, but the hardware cost can be extremely high, especially when an enterprise needs to scale its CDP capabilities and add more appliances. Appliances can carry price tags around USD50,000. The software solutions are billed using various licensing strategies, frequently per server. License fees of USD5000 to USD25,000 per server can be readily found. The software solutions also involve agents that must reside on each server to be protected. The more servers a user has, the more agents that have to be purchased and managed a stumbling block to the SMB, a potential struggle for the enterprise, that might have to manage agents on hundreds (or thousands) of servers. There is a significantly better way.
Host-based CDP software eliminates the hardware expense of CDP appliance but comes with its own set of cost and complexity issues. The software solutions require that agents be installed on each server to be protected, creating management overhead and additional costs. The pricing model for this type of CDP is typical of most enterprise backup software that charges a license fee for each server or database that is protected, regardless of how often the CDP functionality is actually used by each server.
The third CDP architectural alternative is to incorporate the CDP functionality as a feature in a full-featured backup and recovery software suite. Asigra Televaulting is a product in this category.
Asigras implementation of CDP focuses on remote office CDP and is designed to work over a WAN. The CDP functionality is integrated as a feature of Asigra's Televaulting software. The functionality is available for file data (Windows and UNIX file systems) with agentless CDP and for Microsoft Exchange emails.
We are not going to delve deeply into this particular solution here but the link below will take you to Asigra's web site for further information.
However, there are a few important points of universal relevance. When CDP is enabled, the Asigra DS-Client monitors the specified data source when changes are detected, data is automatically backed up offsite. The client also provides automatic retention policy enforcement, an essential feature for a business whose retention practices are dictated by regulatory requirements. Retention enforcement is enabled for both regular and CDP backup. Separate retention configurations can be established for both local and offsite data storage and eliminates the accumulation of vast amounts of data requiring storage. CDP without retention rules could necessitate an additional investment in disk hardware [and lead to breaches of the data Protection legislation in Europe -Ed].
Asigra's CDP implementation is a two-stage continuous backup that without agents backs up any changes on Windows or Unix or Linux servers to a local server or servers as they occur. Backup starts with a change event, and granularity is available to the pace at which data is written to disk in a consistent state. The local servers aggregate the changes, de-duplicate, compress, encrypt with up to AES 256 prior to offsite transit to the centralised storage repository at a data centre, and the data remains encrypted at-rest. This DS-System piece of software at the repository automatically checks and verifies data for consistency and recoverability. If Autonomic Self Healing determines a file can't be recovered, it automatically asks for it again from the local server.
Selecting a CDP Product at a Glance
One way to determine if CDP is right for your remote office or branch office locations is to ask yourself a set of qualifying questions. For example, are you worried about meeting the business SLAs established by the CIO which include remote sites? Perhaps you are being asked to measure business impact of downtime at remote sites or are looking to modify or improve your remote branch office backup site and strategy? Do you have rapidly changing data at remote sites that is critical to business operations, and are you worried about shrinking backup windows to protect that data? If you answered yes to one or more of these questions, you should be seriously investigating CDP technology for your remote sites.
There are several features a robust CDP product brings to the market. These include:
The Best Reason of All
CDP as a feature in a backup/recovery software package carries an obvious financial advantage but perhaps the most important reason of all for CDP to be part of a data protection package is that it is not a universal replacement for other forms of backup, replication or disaster recovery. CDP isnt RAID, replication, or mirroring. Each of these technologies has its place in the data centre data life cycle, but copy-based strategies only allow you to recover the most recent copy of data. CDP allows you to restore previous versions back to a specific time or event.
CDP has an important place in the data protection hierarchy and ensures the recoverability of all crucial business information. But it is not for data that is relatively static or is less than mission-critical. In some cases, CDP does not provide the user any advantage over conventional backup and recovery. If the data is not changing regularly, it doesnt matter if it is continuously backed up.
If data is not changing enough to require continuous backup, making a business case for a CDP appliance is difficult to impossible. But if CDP is a feature in existing backup software, it can be enabled or disabled as the IT manager thinks best depending on the importance of the data. If an enterprise or SMB is experiencing dynamic data growth (the average is over 100 per cent every 6 to 18 months) CDP can be a lifeline, keeping data not only protected but up to date. CDP is a powerful, in many cases an essential, data protection feature, but it should be just that: a feature integrated with powerful, flexible, backup and recovery software.
Conclusion
Continuous Data Protection is the latest piece in an enterprises data protection arsenal. It complements existing backup, replication and snapshot technology. CDP will quickly become an integral part of a total backup solution for business.
[This is clearly an important type of solution because it improves the 'horses for courses' choice of back-up solutions, each tailored to the degree of dynamism and value pertaining to each data type or object. The usual security measures need to be considered - encryption of course, but also how the company's data retention policy can be brought to bear on data backed up through this method. Ensuring secure deletion at the end of the retention period can be even more important than ensuring dynamic back-up.
An equally important consideration is the protection of the logs associated with this type of fast back-up and the protection against exploitable time slices. It must be impossible to make changes, use the changed data for nefarious purposes like implanting malware, and then restoring the state to one prior to the introduction of the changes without the system at least recording this securely and indelibly. If the back-up system fails to prevent or detect this type of thing it can become a convenient tool in the hands of criminals, automatically and practically hiding their tracks. Hence, the security of the back-up system itself, is of the utmost importance. --Ed].
About the Author
Mark Ferelli is an independent technology journalist and commentator specializing in computer storage. He has been an industry analyst, and was editor of Computer Technology Review for 17 years. His articles on storage technology number in the hundreds, and he has taught or moderated seminars at trade events nationwide.
Related links: (Open in a new window.)
www.asigra.com/
View printable version (opens in new window)
Back