An Introduction to Remote Backup
- Published on Friday, October 26, 2007
- Written by Bob Zeidman
ADVANTAGES OF REMOTE BACKUP
Remote backup offers the user three main advantages with regard to disaster recovery. They are off-site storage, full automation, and full service.
Remote backup allows the user to store critical data off-site. This is necessary to guard against catastrophic loss. In a fire, earthquake, flood, or other natural disaster, the user's backup tapes are as likely to be destroyed as the main computer system if they are both maintained at the same site. With an RBS, the main system may be destroyed, but the data is still available. The user will suffer damage, but his or her business will still be viable.
Remote backup is fully automated. While other backup methods, such as tape backup, can be automated to take place at scheduled times, it is still up to the user to perform many tasks. The user must maintain the tape drive in good working order. Since a tape drive is a mechanical device, it has a much higher failure rate than an all electronic device such as a modem.
The user must also check that data is being correctly written to the tapes and that the tapes are maintained in good condition so that the data is not lost. The tapes must be removed from the tape drive on a regular basis before they fill up. If a tape fills up, a scheduled backup will not occur, since there will be no room on the tape to write the new data. With remote backup, these are simply not issues.
Finally, perhaps the most important aspect of an RBS is that it transfers the responsibility for maintaining the files from the user to the staff of the RBS. When a computer user purchases backup equipment or backup software, the manufacturer is responsible that the equipment or software works correctly. The manufacturer is not responsible if the user loses important data. An RBS, on the other hand, will not survive if the customer loses data for any reason whatsoever. Therefore, an RBS must provide excellent service and make certain that all data is backed up correctly and can be retrieved without problems.
There are a number of important issues that must be considered when choosing a remote backup service. Each one is discussed below.
DATA INTEGRITY DURING TRANSFER
Data integrity must be assured while transmitting files to the remote file server. Since the data being backed up can be of any nature, even a single bit corruption cannot be tolerated. An RBS cannot assume to know the type of data being transmitted or the amount of loss that is tolerable. A single bit corruption in a high resolution computer image will not be noticed by the human eye. A single bit loss of a computer application program will cause it to function incorrectly if it functions at all.
DATA INTEGRITY AT THE FILE SERVER
Once data is transferred to the remote file server, its integrity must be maintained for as long as the customer is active. In other words, the customer must be able to retrieve any and all data at any time, regardless of when it was backed up. Certain steps can be taken by the RBS to increase reliability to a reasonable point.
An uninterruptible power supply (UPS) must be used at the server that is guaranteed to maintain operation for a specified amount of time during a power failure. This must be enough time to back the system up and bring it down in a controlled manner so that data is not lost. In addition, the system must be built from reliable components from reliable vendors. The system must be monitored 24 hours a day so that problems can be corrected immediately should they arise. Also, the file server itself must be regularly backed up and stored in a safe, secure location.
Transfer rates over typical phone lines are increasing quickly. It is now possible to transmit at 28,800 bps using state-of-the-art modems. However, in looking at the last nine years of technology, modem transfer rates have increased by a factor of 20 while data storage capacity has increased by a factor of nearly 1000. Let us examine the needs of a typical small business PC user. Figure 2 on page 53 shows the typical modem speed available in each of the last nine years, and the typical disk drive size shipped with a new personal computer. The graph assumes a disk usage factor of 50%. The disk usage factor refers to the average actual amount of storage used by a typical user - the graph assumes that a typical user will consume 50% of the disk space available. These numbers are used only to give a rough idea of the problem of data transfer faced by an RBS. The estimated time to backup a PC, without using any speedup methods, is shown to be close to 40 hours.
An RBS must offer fast communication lines to its customers, such as ISDN. But these new technologies will not have widespread use for years. In the meantime, methods must be used to accommodate current technologies for these users and their data backup needs. The following features are typically implemented to provide faster backup times.
The remote backup software should perform loss-less compression on each file before it is backed up to the server in order to reduce the transmission time. Loss-less compression means that no data is lost even though the file is smaller. For example, a file containing the word 'the' repeated 50 times could be compressed to a single file containing 'the(50)'. The compressed file is much smaller but could be expanded to give you exactly the original file.
The remote backup software should perform incremental backup of the user's files. Incremental backup is a method whereby a record is kept of the date and time when each file was last backed up. If a file that is scheduled for backup has not been modified since that time, it will not be backed up. In other words, only files which have been modified since the previous backup are actually transmitted to the remote server. In this way, data that has not changed does not need to be re-transmitted.
Critical File Backup
In an ideal situation, an RBS would replace a tape backup system and allow an entire computer disk to be backed up. As the graph in Figure 2 showed, however, it does not look like this will be possible until significantly faster transmission technologies are more widely available. Because of this, remote backup should be used in conjunction with a normal backup procedure. An RBS should encourage customers to only back up those files which are truly critical and cannot be replaced if they are lost. These files include financial data, customer databases, proprietary designs, graphics, text, etc. Critical files are defined as user-generated files which cannot be replaced or recreated from other means or from other files. For example, a software developer should backup only source code and files that contain compiler switches. A software developer should not backup object files or executable files, since these can be regenerated by the compiler from the source code.
An RBS typically does not encourage the backup of application programs or operating systems since these can be repurchased and reinstalled if they are lost. The cost of backing up these types of files is not cost-effective compared to the risk of losing them and then repurchasing them. Of course, since these programs do not change (until they are upgraded), the user will incur only a one-time charge for backing them up. In some particular high-risk situations, the user may want to backup these up and pay a fee to have the RBS maintain them.
Security is a very important issue these days as data is sent across public access networks and over unsecured phone lines. An RBS must be particularly concerned about security since users are, by definition, transmitting their proprietary and mission-critical data. Several steps should be taken by the RBS to insure that a user's data is secure from unauthorized access.
1) Walls between customer files on the server. Each user of the service should have a separate account on the remote file server that is inaccessible by any other user.
2) Encryption. The client software should perform encryption of the user's data before it is uploaded to the file server. This encryption should depend on a key known only to the user, and that is unreadable without this key.
3) Limited download capability. An RBS should allow easy uploading of files to the server, but should maintain very strict procedures for retrieving data from the server so that the data does not get into the wrong hands.
An RBS can allow the customer to perform completely unattended, scheduled backups. The software should automate the remote backup by allowing it to be scheduled at any time, without user interaction. The backup scheduling program should normally not affect the operation of the customer's computer - i.e. it should be 'asleep'.
At the scheduled backup time, the scheduler should 'wake up' and give the user some time to abort the backup in case the user is currently performing some critical task that cannot be interrupted. If the user does not abort, the scheduler should perform the remote backup and generate a report. It should then restore the state of the computer before it woke up, and go back to sleep.
The computer should be able to continue performing any application that may have been running before the remote backup.
Since remote backup occurs automatically, the user must have regular reports showing which files have been backed up and, for billing purposes, how many bytes were transferred and how long they were connected to the remote backup server. In addition, since communication lines sometimes have noise and other problems, the report should tell the customer whether any problems were encountered during the backup.
A remote backup service is a means of providing economical off-site backup. In order to be viable, the service must guarantee data integrity during transfer and also at the off-site storage facility. The data transfers must be fast to minimize the resource usage including computer time and connect time. The system must be secure so that unauthorized access to proprietary data cannot occur.
Finally, the system must be automatic so that it requires minimal effort by the user and is not seen as additional work, but rather as a reduction of work in an already busy schedule.
Bob Zeidman is the President and CEO of eVAULT Remote Backup Service, one of the originators of the concept.