FOSDEM is the biggest free and non-commercial event organized by and for the community. Its goal is to provide Free and Open Source developers a place to meet. No registration necessary.


Interview: Gratien D'haese

Gratien D'haese will give a talk about "Linux Disaster Recovery as a Service (with rear)" at FOSDEM 2011.

Could you briefly introduce yourself?

Gratien D'haese is living in the Antwerp region (Belgium) and has his own company specialized in IT (Unix/Linux) Consultancy. Gratien has been working with Unix since 1988 and followed Linux since its birth and started using it through Linux Nighthawk 0.99 release (March 1993). Still remember, well back in time (nineties), that within corporations Linux was a no-go. Well, it turned out they were wrong and today Linux (and Open Source) are actively used within all companies world-wide, because it's open and cheaper in license cost. That doesn't mean it is cheaper in support, but it became easier to switch from services or support companies.

Gratien is donating to the Open Source with 2 projects:

  • Make CD-ROM Recovery (mkcdrec) was initiated in 2000 and was a Linux disaster recovery (DR) program basically using CDR, DVDR or ISO images as targets. It has been widely used, but recently we decided to stop adding new features and bring it into maintenance mode which means security and bug fixes will still be applied for many years to go.
  • Relax and Recover (rear) was initiated in 2006 together with Schlomo Schapiro (from Germany) and is basically a rewrite from scratch of mkcdrec into a Linux disaster recovery framework. We noticed that since 2009 rear became quite popular and lots of big governments and corporations started to deploy this piece of software as their main DR solution.

What will your talk be about, exactly?

Gratien will talk about Linux Disaster Recovery principles and what is currently available in the Open Source market. The pros and cons will be discussed briefly and how to choose the best fitting solution to you. We will concentrate on one possible solution (Relax and Recover which is abbreviated as rear) to present the audience with some practical examples of how it could be done in reality.

When we lose data, a good backup policy can be of a great value, but when we lose a complete system, restoring the backup of the data is not enough. In the first place we need to replace the (broken) hardware with hopefully similar hardware and then start re-installation of the operating system together with additional application software. Configuring the operating system and applications can take hours or in the worst case days. Once the backup (client) software has been installed on the system all the user data can be restored. Again, this takes plenty of time. Of course, when time is of essence we better foresee an efficient way to get your system up and running again with all the data, including user data, installed like before. For this reason we better rely on a special-purpose program that primarily focuses on disaster recovery.

The disaster recovery plan should describe the recovery process and flow of activities. The flow of activities can be large, such as load the operating system, install application software, restore data, make configuration changes, post recovery checks, tests and open the service back to the users. Would it not be nice if we could reduce the above list of steps? Well, this is the purpose of this talk; introducing Relax and Recover (rear) which takes care of the "load operating system, install application software, restore data, make configuration changes, and post recovery checks" steps in one go!

What do you hope to accomplish by giving this talk? What do you expect?

In 2007 we were able to present Relax and Recover on FOSDEM in a Lightning Talk session which was limited to 15 minutes, but still we received some very good feedback which has been incorporated in the meantime. Therefore, now that we have the opportunity to go in-depth, wider and more precise in what Linux disaster recovery is all about we hope to convince people to have a look at DR software and start using it as a basic system administrator building block. There are still too many people thinking that having a good backup is enough; the contrary is true.

Personally, we enjoy talking to the audience and hope to return something to the open source community as we retrieved and used it all the time for decades. Nothing more, nothing less. However, we are looking forward to FOSDEM to meet people with different opinions and ideas about open source projects and in particular about rear. Without the community we are nowhere and rear wouldn't exist in the first place. We do it for the fun and enjoy the atmosphere of this fantastic conference. It is a kind of open source market place where people meet and exchange ideas and above all, where we are able to learn a lot!

What are the unique selling points of Relax and Recover compared to its competitors?

Relax and Recover (rear) is a true open source Linux Disaster Recovery (DR) product licensed under the GPL license. In the DR level products we have two camps: closed and open source. I never used the closed source disaster recovery products and therefore I cannot say much about them. In the open source category we have three pillars, being cloning, disaster recovery additionally above backup/restore software and the true disaster recovery products. Rear fits in the open source category and is a true DR solution. For DR purposes we can only give you some good advice for free: “test, test and test before deciding”.

Rear has two well-known competitors, being Make CD-ROM Recovery (mkcdrec) and Mondo Rescue. We already mentioned that mkcdrec (main contributor is also Gratien) will phase out gradually in favor of rear. So, there is only one competitor left and that is Mondo Rescue. We believe that rear's strength is its framework which is simple in design and extremely easy to adapt, and adding new components is just a matter of writing a (hopefully) simple script and putting it in the right directory structure (see the concept paper for architectural details). And above all, rear is completely written in bash which is the language known by all Unix/Linux system administrators.

How many developers are working on Rear?

We have two main contributors since 2006 and that is Schlomo and Gratien. We have currently two heavy contributors of patches and new features working at the Belgian Federal Police (Dag and Jeroen). And over the years, numerous people have donated patches and other positive feedback. We tend to keep the list of contributors in our AUTHORS and CHANGES files.

Which new features can we expect in the near future?

Cloning functionality towards P2P (physical to physical), P2V (physical to virtual) and V2P (virtual to physical) have just been released in the 1.9.* series. The basic functionality that rear offers is great for one system (or a few systems), but when we're dealing with hundreds of systems then it will become complex to manage all the recovery images. Therefore, we propose an extra service called “disaster recovery as a service” and an additional tool named “rear-server”. That is something we would like to introduce in the 2.* series range.

Rear-server will gather information of each individual system on a central server, including the disaster recovery images if desired. The rear-server can also act as a boot server (via PXE) to boot the recovery image made earlier of a particular system. On top of the central depot of all the recovery images and log files we will build a web interface to browse through the list easily.

The rear-server would be built with standard components already available in Linux, such as Postfix, PHP, Apache and MySQL. From the client side, our Linux system where we created the image on with the command “rear mkbackup” or “rear mkrescue”, the information will be sent to the rear-server via SMTP.

Imagine how easy it would become to create a recovery image from a Linux system in your network with just a click on a button via a web browser. Starting the recovery would become a simple task which could be executed by an operator without any intervention by an administrator. No need to search for the Disaster Recovery Procedure anymore, or running to a vault to find back the recovery image DVD. Life would be such much easier.

If you feel you can contribute by using it, by writing patches or by adding new features, do not hesitate and contact us – see the rear website for more details.

Creative Commons License
This interview is licensed under a Creative Commons Attribution 2.0 Belgium License.