Not too long ago, server-free backup was the best option for backing up large data sets. However, there are some legitimate alternatives available today. W. Curtis Preston discusses how server-free backup works, pros and cons of server-free backup, the server-free backup market, and today's alternatives to this complicated approach. His answers are also available below as an MP3 download.
Listen to the server-free backup FAQ
Table of contents:
>>What is server-free backup, and how does it work?
>>What type of environment is server-free backup best suited for?
>>What are the drawbacks of server-free backup?
>>What is the difference between server-free and LAN-free backup?
>>Who offers server-free backup products?
>>Are there any standards for server-free or LAN-free backup?
>> Are there any alternatives to server-free backup?
Server-free backup is an umbrella term for two different kinds of backup. The general distinction is that it is backup that is going to be done outside of the server that is being backed up -- hence the name. Some of the methods for doing server-free backup actually use another server, so some people don't consider them server free. But, they are still free of the server that is being backed up.
There are a number of different ways that this is accomplished, and they generally have an alternate copy of the data, such as a business continuous volume (BCV) or a split-mirror, or perhaps a snapshot, and then present that data to some other system or device that can copy the data to tape. The typical way that this is done is via a dedicated backup server. So, you have a server that's being backed up. It's using data that's on the storage area network (SAN), which is being controlled and split off and snapped by a backup server. Then, that extra copy of the data is presented to another server that transfers the data to tape.
There is another way, which is truly server free, which uses something called third-party copy, which is a SCSI command. Third-party copy can tell a device, such as a disk array that supports the command, to actually copy its data straight to a tape drive without having to go through a server.
The best way to do any backup is the simplest way to do that backup. Complexity brings, well, complexity. But, along with complexity comes cost and risk that things could go wrong. So, server-free backup is more complicated than other forms of backup. Therefore, it should only be used when it is really necessary.
The types of environments that tend to move toward server-free are environments that have a very large data set behind an individual server. Very large is relative, but we're definitely using the word terabyte in a plural sense at this point, and the load on that individual server is too much. So, you want to move that I/O load off the server and onto something else -- either a dedicated server or the disk hardware itself.
There is inevitably going to be a license involved. Server-free backup is a complex thing from a coding perspective, so that is going to come along with cost when you purchase the license. The dedicated server or disk array that you need to do this isn't free either. So, the big downsides are complexity and cost, but there is a risk involved as well.
For example, if you are doing server-free backup you may have logs in three different places. You might have logs on the client that's being backed up, the backup server that's transferring the data and the main backup server that's controlling the whole process. So you have logs in three places that you need to check, and when things work great they're great. But, when things don't work it can be rather complex to figure out why.
LAN-free backup is simply connecting the server that needs to be backed up to a SAN, so the data is transferred over Fibre Channel to a tape drive, VTL or disk.
In server-free backup, the data does not flow through the server being backed up. The server being backed up is still involved, because it has to do things like quiesce the database or quiesce a file system. But, once the frozen image has been created, the data movement no longer involves the original server.
I think you'll find that any backup product that's aiming at the enterprise, will offer server-free backup. If a vendor doesn't offer server-free backup they either aren't aiming at the enterprise or they aren't doing it very well. So, if you are the type of company that wants server-free backup, then you should look at an enterprise backup product.
Not so much. Each of the ISVs, the independent software vendors, have their own way of talking to the disk vendors. And each disk vendor has their way of talking to the ISVs. So, if you are an ISV that plans to support server-free backup, there's no single API for you to write to. So, generally, an ISV has to decide which of the OEMs they are going to support. Most start with the obvious two or three and then the others as the market dictates.
While its really important for the people that need it, the people that need it are definitely in the minority. The market share for server-free backup is relatively small compared to the general backup market. This kind of thing doesn't tend to drive standards.
If you asked this question five years ago, I would say no. It was the only way to do large server backups. However, the advent of continuous data protection (CDP) and near-CDP has created real alternatives to server-free backup. And, I think the large ISVs are agreeing with that. Many, if not most, of them have either acquired CDP technology or they are writing it themselves.
With a large server, I don't have enough time in the day or enough CPU cycles to back up this server in a reasonable amount of time. So, how are we going to do that? You have to take a step back. Really, what is a backup? Its just an alternate copy of your data for restores. CDP and near-CDP solve the problem in a different way. They don't do the full backup or the nightly backup in the traditional sense. So CDP and near-CDP essentially remove the problems associated with large server backup. Because you are backing up the data as its changing, CDP and near-CDP work really well when you are backing up a large data set.
With CDP, you are transferring data to the backup system literally every second so you are spreading out the load over the course of a whole day. With near-CDP you have the option of doing it that way or you can transfer it in small chunks -- say every hour you transfer the blocks that were changed during that time. Either way, you are moving a significantly smaller amount of data than you are during a traditional backup.
One could argue that CDP and near-CDP products are server-free backup in a general sense, because the work happens outside the server. Perhaps we should broaden the definition of server-free backup to include CDP and near-CDP. If someone is considering server-free backup, I would strongly urge them to consider all three of these technologies and select the one that is the most appropriate, the least expensive and the least complex while meeting their needs.
W. Curtis Preston (a.k.a. "Mr. Backup"), Executive Editor and Independent Backup Expert, has been singularly focused on data backup and recovery for more than 15 years. From starting as a backup admin at a $35 billion dollar credit card company to being one of the most sought-after consultants, writers and speakers in this space, it's hard to find someone more focused on recovering lost data. He is the webmaster of BackupCentral.com, the author of hundreds of articles, and the books "Backup and Recovery" and "Using SANs and NAS.".