What is the best way to take advantage of virtualization for disaster recovery and business continuity?

From a virtualization standpoint, along with DR, FaconStor NSS is going to be your best bet.

If you're interested in local recovery (servers restored in 10 minutes) and disaster recovery (entire site running in 30 minutes), a combination of FalconStor NSS and FalconStor CDP is a great solution.

Storage virtualization is the first step. This allows for local recovery (snapshots or continuous data protection) and remote recovery (replications or remote copies).

Server virtualization layers on top of the storage. Virtual servers are easy to bring online in a remote recovery site environment because you do not have to maintain one-to-one hardware. In addition to virtual-to-virtual recovery, you can do physical-to-virtual recovery by staging the drivers. Both of these lower the BC/DR costs.

I use Compellent Storage Center and Microsoft Hyper-V. As with the previous poster, I can recover a server or a full site in minutes. Email me if you would like to discuss the technical details.

Brian, I agree with J Wolfgang Goerlich, storage & server virtualization, BRD both physical and virtual. One other option is application redundancy in the cloud. One product that we have entertained, but not used, is from a company called Geminare which touts "real-time server replication with automatic failover, immediate remote operational capability with complete data protection"

An online/cloud component may be critical for companies that can't afford to (or simply don't) use offsite data centers to keep secure copies of data out of harm's way, but they tend to get very expensive fast, especially if you have a lot of endpoints. Besides which, if you suffer the kind of disaster that makes offline storage critical - a flood, fire, extended loss of power, etc - then you're stuck waiting weeks to get DVDs of data and have to cross your fingers that they work because theres no telling.

An alternative that has just come out in the last year or so are dedicated backup appliances that can cover and restore local and remote data (preferably without reconfiguration when anything moves) and can be used locally when needed for seeds and fast major disaster restores but kept remotely in a secure offsite location for daily incremental backup and trivial file restores.

For enterprises, appliances can be useful for keeping franchises safe while still autonomous, or as replication lifejackets for real-time, open-file data that has to survive for the company to survive. What's nice for them in the cost-cutting climate and for SMBs who are always cost-conscious, appliances are almost always cheaper than online services and subscriptions when there are a lot of endpoints even within the first year or two of service.

For more information on the subject Bitpipe has a white paper that my company sponsored as it highlights some of these issues fairly well: http://searchsmbstorage.bitpipe.com/detail/RES/1254317937_438.html

Virtualization is a great DR/BC alternative and one I think still hasn't really come into its own. There are several great products out there that can let you have a "warm" site with virtualized servers while continuously getting updates of delta changes applied to them. Those VM's don't even need to be up and running, just need to have the actual physical VM server up and running.

Platespin is a great tool that can migrate servers in all 4 directions (P2V, V2P, V2V, P2P) and then keep them sync'd with those delta changes. The VM's at the DR site don't even need to be running at the time (crazy I know). http://www.platespin.com/products/migrate/

But ultimately, using virtualization allows you to have a DR site with the minimum number of servers actually running, and then bring up the VM's quickly and have you back up and running in a very short time. The obvious DR plan, archtecture, network bandwidth, etc need to all be in place to make it all work. But once it is, it can be a great alternative to a traditional P2P DR strategy and it will really help keep costs down.

Brian, There are numerous virtualization products on the market. My experience has been with VMware. The key to a successful implementation is detailed planning (both of server infrastructure and DR/BC strategy) and proactive growth of the infrastructure once installed.

There are 2 major components here, server and desktop virtualization. I will focus on the server side for now. To fully utilize some of the features of VMware, namely high availability (HA) and distributed resource scheduling (DRS) a shared disk component (SAN) is required. There are a number of vendors and products available (Fiber Channel, iSCSI), with varied feature sets and cost levels. The buzz words floating around now involve thin-provisioning, de-duplication, virtual storage, snapshots, replication, cloning, etc. You will need to review the vendors, products and features to determine what your needs are and what works best for you. The emergence of Cloud Computing needs to be considered when reviewing recovery and capacity growth options.

In addition to the SAN, you will need to determine server hardware vendor, datacenter configuration, will you have a backup datacenter or a hosting provider, strategy for live replication between datacenters, identifying most critical applications and processes that need to be recovered first, snapshot and backup scheduling if appropriate, disaster recovery procedures (Site Recovery Manager {SRM}) for a scripted recovery and Disaster Recovery Testing, Lun sizing and cluster configuration.

Managing and monitoring the virtual Infrastructure as well as fine tuning it for optimal performance need to be reviewed. vSphere, the latest offering from VMware has enhanced troubleshooting and monitoring tools. Linked clones, vc heatbeat, vmkusage, capacity i/o, application virtualization are additional topics to think about. There are also additional management and monitoring tools available. Good luck.

Virtualization removes hardware dependency of OS. Hence it allows you to move an OS to dissimilar hardware without any OS changes and repair. This eliminates the need of similar hardware’s at DR site. Virtualization also enables to pack more applications into single server hardware; hence it helps to reduce number of hardware required at DR site.

Since VM are set of files, it can be copied, moved easily to any destination and can be easily started at new destination easily which reduces RTO.

VMware supports VM monitoring and VM restarts which can be used as poor man’s HA or replacement for costly failover clustering.

If you can afford a SAN to SAN replication to DR site, VMware SRM solution offers excellent DR automation. This simplifies DR procedure and reduces RPO& RTO.

if SAN replication is not an option you could backup VM's daily on to your backup tape and ship it to DR site. In the event of disaster you can restore VM from tape media and become operational with reasonable time at very low cost.

revert to me with detail on requirements if you need any further assistance in designing the solution.

Two things I would add...

Continuous Data Protection and Storage Virtualization

With most existing practices, backups of data usually occur during the wee hours when there is little if any activity on the network. The reasons for this are somewhat obvious. This is the time when most administrators can move the relatively massive amounts of data without being concerned about user performance. The practice results in what is known to us all as the ‘backup window’. What many administrators are finding is that the window size usually increases over time because more and more amounts of data need to be stored. On the other end of the spectrum there is often the pressure to reduce the window size because of the need to be up for more hours of operation. This two way crunch has resulted in sleepless nights for many, particularly when backups fail. Additionally, many do not even know how ‘restorable’ the images are. In a disaster recovery type of scenario, the pressure is then on and many find that key application and data base environments cannot be restored to complete consistency. Many times valid images are weeks or at times months old. Sometimes data is lost forever, sometimes it is restored, but in a dated fashion. The recent T-Mobile incident is a good example of this. It also brings home the point that not all disasters have to be fires, earthquakes & tornadoes. Sometimes it can just be a simple upgrade gone wrong.

Continuous Data Protection (CDP) and storage virtualization can work to virtually eliminate the need for backup windows and provide for the ability to have immediate restore of testable images that are ‘known good & restorable’ prior to any such disaster. (Be sure to TEST our DR plan!) Storage virtualization works to create a large massive repository out of the many arrays and disks that you may already have. Additionally, as storage is added – it is in turn virtualized so that it can be leveraged by any and all systems and applications within the environment. Perfect for the VM scenario that Jason highlights above. Just remember the standby VM is worthless without an intact and consistent data store. CDP works to provide the ability for full real time backups during production hours without the system & network performance impact that traditional methods are prone to. CDP journaling can be saved at multiple times during the day saving only the offset of data required to maintain data image consistency. These images can then be moved to tape, VTL (tape emulation on disk), or disk. Because the backup occurs from the journal, not the primary store, backups can occur more often and not affect front end system performance. This can work for not only file based but for block based (data base) environments as well. The technology allows for the ability to additionally journal on a transaction basis (commonly used in banking and financial) so that complete consistency can be delivered not only up to the minute but up to the actual transaction level. Additionally, as CDP is used over time, the amount of backup data required for consistency is reduced as typically only a small portion of what is backed up typically changes. The end result is a much more consistent data environment that is known good and far less expensive to operate over time.
(Additionally, your administrators can sleep easier at night rather than white knuckled with a cup of joe ; - )

Reach out any time if you find that you require assistance or guidance.

Edwin C. Koehler Jr.
Data Storage Solutions Architect
Storage Engine Inc.
Phone: 732.747.6995 ext. 267
Mobile: 585.645.4399
Email: edwin.koehler@storageengine.com
Web: www.storageengine.com

Hi Brian
Lets consider there are about 6 - 7 BC applications running on MS or Linux Clusters on Physical boxes with the SAN as an Backend with the FC Storage & if the IT fails the Busiless loss is very high as you have mentioned.
In this senerio you can think of Virtulization (Servers Consolidation) of these physical boxes.
The benifits are very high.
Then at DR site you can deploy the applications using Virtulization with the SAN as an Backend with the FC Storage.
For DR you will have to go for multiple options like VMware SRM & EMC Recoverpoint.
This will help you setinmg up DR site Synchronous or Asynchronous.
I have not mentioned any comments on budget , since you are talking about the DR & BC , so I assuem you are aware of the budgets for running BC & DR.

Most businesses recognise the importance of ensuring high availability and quick DR of their mission-critical applications and data to avoid financial losses after a system failure or disaster. Only a few have properly balanced technology costs with the needs of their organisation.

Most businesses, not necessarily all, but a large majority don't know if their critical business applications and data are properly protected. They have applied what they feel are the appropriate levels of technology, but often they have over-invested, possibly meeting the needs of one or two critical applications (although they don't know that for sure), with overkill for all other applications.

Building out an overall recovery strategy can't just be a roll of the dice. It's no longer enough to have an "In case of emergency" DR plan that just sits on the shelf. Now there is also a need to address the "small disasters" that typically impact daily business operations - namely lost or deleted data, application and hardware failures - any of which can result in hours or days of downtime and lost productivity.