Archive for the ‘Disaster Recovery’ Category

Image Based Disk to Disk Backup, the New Standard

Saturday, March 7th, 2009

The major problems with tape backup are:

  1. It is unreliable.  Tapes don’t get changed, they overheat, they are terribly hard and slow to verify.
  2. Restoration is slow because tape is slow and because you must rebuild the OS and restore files separately.  It is time consuming to test the tape as well.
  3. If you run out of room you typically throw away your investment and start over unless you can have someone change tapes and run a two tape backup which is usually very problematic.

These  issues lead to many minor issues such as never REALLY knowing if you have a good backup.  Over the last one to two years a new method of backup has become available to small to medium business.  Very low disk prices has made it affordable to leave unreliable and slow tapes behind and use disk drives to backup disk drives (disk to disk backup).  Disks offer several benefits over tape as a backup medium:

  1. Tape is a sequential access medium, meaning to read the last file on the tape you have to read ALL the other files.  Disk is random access, meaning to get to any file on the disk you just look it up in a directory and read it.
  2. It is far faster and easier to validate that a disk based backup is working
  3. Disks media is more reliable than tape media
  4. Disk allows the backup of another disk as an image, this means you not only get the files but you get the Operating System exactly as it is running

Image based backup means you keep an exact copy of the server or workstation you are backing up.  This image is so complete that you can actually mount it as a virtual drive and run it.  More importantly new technology allows the image to be put on dissimilar hardware and the software finds all the drivers that need replacing and takes care of getting that done.  Is is extremely smart about it and can put an image on from most hardware onto must any other hardware with no human intervention.  Rarely you will have to find a driver on the web and deliver it to the machine but it will tell you easily when this is needed.  To test the backup you can easily mount the image and see the files that are backed up.  With the image mounted you can copy a version of a file back to your real hard disk.  Finally incremental “backups” can be made by simply repeating the changes, bit by bit, that have changed on your real hard drive, onto your image making the amount of data to change far lower.  This means you can have several versions of files backed up by having several incrementals on the drive.

Image based, disk to disk backup is far superior to ANY tape backup I have ever seen.  It is affordable, reliable and fast.  That said there are many ways to set it up and it is very important that you use an experienced engineer to create your image backup scenario for the maximum reliablity and ease of use.  We went through about 20 different iterations before finding what we currently consider the right one.

Why “Backup” is the wrong question, you should be thinking “Recovery”

Tuesday, January 27th, 2009

Having written that Tape Backup is not recommended, I should now write about what we are recommending for a solution.  Prudent marketing might make me say……email me to find out, but that’s not in the spirit of a good blog and frankly it’s not that hard to figure out what I might be thinking about.  But before I do the “what” article I think some important background information is necessary.  It is important to understand what you are protecting against.   Backup by itself is really meaningless, what you need is recovery.  There are actually a lot of things to think about when you think about recovery.  Recovery can be categorized as needed in three cases.

  1. I have overwritten or deleted a file. I need a copy that was saved yesterday or last week or last month.
  2. I have just had a disk drive crash, a motherboard fail, a controller card fail, etc. etc.
  3. I have just lost access to my servers physically and electronically either through fire, flood, theft, earthquake, etc.

For each case there is a consequence and a probability of occurring.  In the first case the consequence is annoying and perhaps costly.  It is rarely if ever catastrophic, although it can be in the case of a legal brief that must be filed by a certain time.  It is, on the other hand, very likely that this will occur multiple times.  The second case will very likely happen to you at some point in your business life, perhaps multiple times.  The consequence is loss of the services supported by that hardware for some period of time and perhaps loss of a lot of important data.  The third case, which is the one talked about the most in the tape backup days, is the absolute least likely to occur but the most catastrophic because you will have lost access to all your servers and services.   When figuring out a restore plan you should think about Recovery Time Objective and Recovery Point Objective for each of your systems.

Recover Time Objective is the amount of time you are willing to wait in the event of a failure of a particular category on a particular system

Recovery Point Objective is the point in time to which you need to recover in the event of a failure of a particular category on a particular system.

The following examples are not suggested times these are examples of how the results of your analysis would be documented, your RPO and RTO could be wildly different.

  1. Email System,   Category 2: RPO - loss of no more than 1 hour of email.  RTO - 3 hours
  2. Electronic Medical Records System, Category 3: RPO - Loss of no more than 1 minute, RTO - 1 hour
  3. Word Processing Files, Category 2: RPO - Loss of a day is acceptable.  RTO - 24 hours

You may not need to get your email system back that quickly and you may easily be able to sustain the loss of a days email.  There are clients, on the other hand, who will set the objective at zero lost emails and recovery times of less than 30 seconds.  As you might expect the tighter the recovery point objective and the shorter the recovery time objective the more the solution is going to cost to secure and maintain.

Is there a suggested solution that is cost effective and gives good RTO and RPO?  Glad you asked.  See the article, Image Based Disk to Disk Backup, the New Standard (available by end of January 09)

The benefits of tape backup….it’s better than nothing!

Saturday, November 15th, 2008

Seriously there is still a place for tape backup but it’s not in most small to medium business networks.  Here are the problems:

1.  It is not random access so it’s difficult to retrieve single files or folders.

2.  It is subject to degradation easily based on heat and moisture.

3.  The tapes are expensive and wear out.

4.  You must find the same tape drive to read tapes in the event of system loss (theft/fire).

5.  You must restore data separately from the operating system making restoring a complete system time consuming and error prone.

6.  When your technician finds out there is a failure that requires restoring from tape, ask them if there is fear in the pit of their stomach.  Truthful caring one’s will tell you they are nervous.

Tape was the only way we had of reasonably backing up systems till just a year or two ago.  But there are now very inexpensive disk storage devices on the market and great software to take advantage of disk as a backup medium.  These two things make tapes a VERY bad choice for new backup in nearly all cases.   Next installment on the blog….What about off-site backup?.