Tuesday 6.07.11 @ 2pm
Rod Mach, principal of HiperLogic and VMWare expert, continues the conversation, explaining how cloud computing technologies like VMWare and VEEAM enable robust and secure offsite backup and disaster recovery options.
Mike: Good afternoon! Welcome to our webinar today. The title of the webinar is called The Backup and Disaster Recovery Options using Cloud Computing.
I'm Mike Klein. I'm going to kick off this webinar for everybody here and then I'll be handing it off to Rod Mach who's going to cover this topic in a lot more detail. Today's webinar is the first in a two week webinar series, or two webinars, on cloud computing for backup and disaster recovery.
This webinar series is designed to provide you a more detailed view on how cloud computing can change the rules for disaster recovery. Our series also will culminate in our cloud computing seminar, a live in person seminar, in downtown Detroit on June 21st. If you're interested in joining us for that you can email us at email@example.com or you can visit our website under resources and events and learn more about our live cloud seminar coming up on the 21st of June.
So with that, let me just kick this off and set a context for our conversation. The first place I want to start is where you're looking at the traditional disaster recovery trade-off. I like to think about the trade-off as a spectrum of solutions for disaster recovery where you may think on one end of the spectrum about tape backup and on the other end of the spectrum you may be looking at hot site disaster recovery or active-active environments.
When we plot this out, a good way that I find to explain this, especially if you have to explain disaster recovery to your CEO or CFO and you're asking for money to implement a particular solution, is to think of it in terms of dollars on the x-axis and recovery time objective being from very slow recovery time to very fast recovery time on the y-axis and plotting solutions across that spectrum.
So when you're looking at tape backup for example, it can be a very cost-effective solution, but on the other hand it can be error-prone and it can take days to recover if you've got to go through a traditional solution where you've got to recover and find your server, you've got to then restore your operating systems, restore all the patches you had on your operating systems, restore your applications, restore all the patches on your applications, and then finally put your data back on your server.
On the other end of the spectrum you've got hotsite DR where you're literally taking and replicating SANS or replicating your data between two sites so that you can fail over in a far quicker situation but that entire scenario requires in a traditional DR solution a traditional DR environment two fully replicated solutions.
So you're basically taking your primary production data center and replicating those servers and those applications in that same data center. Or setting up a second data center and then setting up communications between them. So it actually costs you more than two times the total cost or doubling the cost of your production environment to do production and disaster recovery.
I'm going to back up a slide here just to share with you, this is actually a true picture of the last hurricane that hit Houston. One of the things we talk about with disasters is in reality disaster recovery is your insurance policy. Hopefully, you'll never need to use it, but if you do need to use it, you need to make sure you test it, you're prepared in advance, you've got that insurance policy to recover from that disaster.
In this case, in Houston, as the workers got out there after the tornado blew through, they found data disks, they found paper files, and they actually found laptops strewn across the street. Certainly unexpected and if all your data was on your laptop, you've got a lot of problems when it comes to disaster recovery if you hadn't backed that up.
So let's move from the traditional environment where you've got this trade-off and talk about what changes in the cloud. One of the key advantages that comes with virtualization is the advantage to really lower your overall cost and lower your recovery time objectives, in a lot of ways shifting your curve to the left in that chart that I just showed on the previous screen.
When you get to virtualization you can now back up your entire server, which means you can capture your operating system, all of the operating system patches, you can capture your applications that have been loaded on your servers, you can capture all of the application patches and your data in your one backup. So now that would mean you would go through a step-by-step process of recovering all of your OS's and applications and patches and hope that you got those right, you know that you got those right because you actually have a snapshot of your entire virtual server in your backup process.
And as I mentioned what this does for you, it actually shifts the entire disaster recovery curve to the left. You can get much faster online backup now becomes much more cost-effective, hot site DR becomes much more cost-effective and you can actually shift it for the same dollars, shift and get a much faster recovery time if you have to in the overall solution that you're looking at. So what I'm going to do now is hand this conversation off to Rod Mach.
Please, as we're going through this process if you have questions either on the overview or specific questions on some of the slides that Rod is going to be talking about or the topics he's going to be talking about, please feel free to enter those in and we'll be going through those questions and answering your questions at the end of this presentation. I'm going to pass this off to Rod to take us through more details.
Rod: Thanks Mike. So today I'm going to be talking about my solution to the virtualization journey and how it can help you with your disaster in recovery and backup. Currently, in a physical world, there's quite a few problems with backup that have existed for a long time. One of the first is that there's no guaranteed recover ability. As you mentioned, you may be backing up your sites to tape or your servers to tape, but when it comes to actually testing those backups, there ís no guarantee that the tape are good or perhaps the backup that you took, that there's a problem with it and you really don't know until you're under the gun and in a disaster type situation if you're going to be able to recover from those backups that you've been taking, and have full confidence that you'll be able to restore them.
It's also complex to do restore and recovery in a physical world. Often you either need to have an identical piece of hardware to be able to restore on quickly as kind of a hot site or you're going to have to try to take that image that you had and restore it on a dissimilar piece of hardware and your recovery time is going to go up quite a bit from the fact that it may have been quite a bit of years between different architectures and definitely restoration from bare metal is quite complex in the physical world.
Another problem people are facing is the data deluge as data increases the backup window is also growing and the backup will actually go beyond the traditional evening backup cycle and start to push into business hours. And so that problem can also exist in the virtual world but there's certain technologies that will enable you to straighten that backup window.
Also, significant storage requirements include full backup on regular rotations and even in a virtual world you have these servers become file and these files are portable, but you have files inside of those files, you have configuration files associated with those virtual machine files and so you still need to pick a technology that is understanding of this new world.
So often people on their virtualization journey will try to adopt the legacy techniques or the techniques that they use in the physical world in the virtual world. Although traditionally put a backup base inside of each virtual machine and try to back it up similar to how they did in the past. The problems with these types of approaches is they don't scale.
In the past, you would have had one backup agent per server with one ethernet cord but we put 50 virtual servers on 50 agents with just a couple of physical NICs that bandwidth starts to cause problems. You also can't backup the image itself or if you do it an image backup you may not be obtaining the configuration files associated with that. So you may increase complexity or still make your life difficult even though you virtualized without using the proper technology.
We're going to go through how virtualization and cloud computing can help you recover easier or better than you have in a physical world. So the way a lot of people do their backup validation today is they just pray that when they need it, like your insurance policy, you hope that it's going to work like you had planned. Unfortunately, that often does not end up being the case. The problem with not verifying your backup is that when you actually go to restore you may find that there was maybe a bootloader missing. So you were successfully backing up corruption perhaps for weeks or months that you didn't discover until the server rebooted and then you're no longer able to restore from that backup and donít have a backup that goes far enough to restore.
So VEEAM, which is one of the technologies that Online Tech uses in their data plans to make sure their customers stay online, has a technology that takes advantage of the virtual environment. You can actually guarantee the recoverability of every backup every time for every virtual machine. So you can get a basic report that says every single day I know that not only is that virtual machine or my server and its data backed up, I can actually guarantee that it's going to boot and the services and applications that I need will start up and the data that I need to access is there. So you can guarantee automatically without human intervention that the backup is going to be recoverable in the event of a disaster.
The way that it basically functions at a high level is it takes your backup repository, and straight from that backup image without using any additional disk space, it boots in a fenced off virtual environment all of your virtual machines, all your servers and applications, and then interrogates them to make sure that they actually booted fully up, that the web server that you were depending on is acceptable, that everything that you wanted to do for any virtual machine application or any other conceivable thing you can think of has been validated and then emails a generated report.
That is quite an advantage in moving to a virtual environment that is not possible in the physical world that would give you immediate peace of mind in an automated fashion that you'll be able to recover in the unlikely event of a disaster that you know you'll be able to get back online.
Like Mike had said, one difficulty of the physical world is though many of the options that have low RTO are, in fact, quite expensive. In a virtual world you can help reduce that recovery time from hours and days down into just seconds with the correct technology in place. It can take advantage of the fact that those virtual machine servers are sitting there.
One of the options that they have is a thing called instant VM recovery. So with VEEAM it can actually boot from the backup straight from a backup image your virtual machine and only stream the necessary changes that are required to keep that machine running. And that gives you some time if a machine would crash or go down you can actually boot from the backup image, be up within 30 seconds, and that gives you some time to debug what is going on, what happened to your original server.
You can do these things basically in parallel. The nice thing about this fast RTO is if you have a 100G virtual machine or server, you don't have to wait for the time it takes to copy all of that data from the backup into a virtual machine. It actually does this magic straight from a backup image to get you back online fast. So if you have a critical virtual machine like an exchange server that needs to be up quick, you can definitely reduce your RTO using a backup solution like VEEAM.
The other item that happens in the physical world and even in a virtual world is that the backup window is ever increasing. In a virtual world it's quite easy to spin up, and clouds especially, you can create virtual machines within a few seconds and you might go from a site that was maybe constrained to twenty or fifty servers where you all of a sudden can create hundreds of servers. Then that backup window that you used to be able to accomplish overnight is now growing. So there's new technology in a virtual world that is able to detect only the changes that have occurred in your servers overnight and only back those up. Traditionally you may have to do a backup, a full, do a bunch of incrementals, do another full.
What VEEAM has invented is the technology that says Okay, we know the blocks that have changed, we're going to do a synthetic backup. Every day we can do a full backup. The way we generate that backup is just taking the blocks that have changed and looking at our original full backup and just recreate it. And then we create a reverse incremental so we can go back in time. So you basically can get the advantage of a full backup every single day at the speed of what a traditional incremental would be. So you can take advantage of this technology to shrink that backup window. If you're not able to meet your backup window overnight, you can use this technology.
Now the next event in this series is going to talk about how to use your SANS to also shrink that backup window. You can leverage the fact that your files exist on some type of shared storage that automatically shrink the backup window using VMWare and storage technology.
The problem with data explosion that we're facing is also reducing the backup storage requirements. If each one of your virtual servers takes 10 Gigabytes and you need to backup a hundred of those, you end up having to back up just a phenomenal amount of data. Now in reality, all of those bits of data are nearly identical. A windows server 2008 file in ten servers is all the same. So there's no real reason to have to back up all of those individual duplicate files.
So backup technologies like VEEAM that have de-duplication built-in can automatically recognize the fact that these three 10 GB servers are all really the same file, so I only need to store one copy of that file, and I'll just make pointers to where those files go in the other virtual machines. So you can take a 30 GB backup and reduce it to 10 GBs, that's a 200% reduction, without any intervention. That saves you not only disk space on your local servers, but if you use Online Tech to replicate between data centers or from your data center to their data center, it reduces the amount of bits that have to go over a WAN link to be able to back up.
In a virtual world, there are some other complexities that exist so you take some of the same problems in the physical world and you do have those in a virtual world, you need a tool that helps reduce some of that complexity. Taking image level backups, for example, you may still want to get individual files out of those backups. You may also want to back up configuration files to those virtual machines. So a technology like VEEAM that is purpose-built for virtualization, has a wizard-driven way to basically restore any of those individual files. So if you accidentally delete a configuration file, or you need to pull a file out of a particular virtual machine, or you want to restore the whole server within a few minutes, you can do that with a simple point and click wizard driven recovery. So it reduces human error, whether those are your engineers or Online Tech engineers, it reduces the human factor and basically automates all of these functions.
Another big differentiater in different technologies that you can use to restore or backup or recover virtual machines in a virtual environment is their support for Microsoft volume shadow copy services. So many applications, even those from native VMware, can backup, like for example, SQL Server or Active Directory, but when it comes to restoring, there's a certain thing that has to happen to restore those properly. For example, when you want to restore an exchange mailbox, if you didn't truncate the logs when you did the backup, it's not actually a supported backup by Microsoft. So by default, if you use VMware tool integration, with VMware backup, that does not occur.
VEEAM has written a special DSS driver that handles not only the backup, but the restore so things like active directory and exchange that can recover properly. So it's not enough to simply back something up, you need to make sure that the support is there to be able to restore that correctly in the application of mode that needs to happen. So VEEAM automatically handles all of this. It has a feature that actually, it's called U-AIR, which lets you restore from an individual active directory object out of that image backup, you can restore individual sequel table, you can restore an individual mailbox, you can do all that without having to put in an agent in the backup or do anything special. It's just automatically taking care of that for you out of the backup image that you backed up. So you're getting both the file level recovery, the configuration recovery, and the application item recovery, all with one solution, with one throat to choke for support.
Now as you start to create more virtual machines and you have files all over these different machines, you may discover that just finding what your customer comes to you and says, "I'll need to recover file XYZ that I deleted. Where is that, what backup image is on that, what server was it on?" The user may not have that knowledge but you're still going to be responsible to restore, or Online Tech is going to need restore that for you, with VEEAM it has, basically a kind of a built-in Google-like search, where you can, with just a few things that you know about the file, or that your customer can tell you about the file, it can pinpoint all of the locations where that could or is, where you be able to quickly restore. So if you are under pressure to get a file to your CEO within a few minutes, you're going to be able to recover that quickly, instead of saying well, let me get back to you when I figure out where that file is.
Now one advantage of VEEAM is they also offer replication within the product. So not only do you get backup, you can do replication. So this is not real-time synchronous replication, like Mike had mentioned that's expensive but if you can do within five minute RPO or recovery point objective, if you could lose up to five minutes of data in a disaster, you could use something like VEEAM to do what's called Near-CDP replication to replicate your entire site, all of the servers, all of their configurations to a DR site. For example from Ann Arbor Online Tech to Flint, you could basically back up your entire site through VEEAM and have just a few minutes of Recovery Point Objective and be back up online within a few minutes if you had to failover to another data center. So it's a two-in-one product and you get both of those with one solution.
As far as the technology, the technology Online Tech is using for VEEAM and VMware is best in class. They've won tons of awards, they're a leader in the solution, they're a VMworld partner, they've been around for a long time, they have over 15,000 customers in the state, it's a rock-solid product. They have 100% reliability with they're sure backup. They have low RTOs and RPOs. Compared to having to combine a whole bunch of different solutions to accomplish basically the same task is quite a bit of cost saving. They have the easiest restore for VMware and hyper-V. Definitely, like I said, they're a technology leader for VM backup the virtualization space is ever changing and legacy backup do have difficulty keeping up with the fast pace of changes of the environment. You get backup and replication in one solution.
So that's the presentation, I'll pass it back to Mike here, thanks for listening.
Mike: Great, nice job, Rod, thank you.
So a couple of questions that have come up. The first question that came up is can we get these slides? And I think we'll definitely email these out. So if you registered for this, we'll definitely email this out. If you're watching this after the fact and you want a copy of the slides, you can always email us at firstname.lastname@example.org