I'm attending a panelist discussion now on "How Virtualization can Enable and Improve DR for Any Sized Business". The panelists include Vizioncore's George Pradel, Director of Strategic Alliances. Others on the panel include Joel McKelvey from NetApp, Lynn Shourds of Double-Take Software, and Azmire Mohammed from VMware. The moderator of this session is Barb Goldworm, President and Chief Analyst at FOCUS.
Discussion highlights from this panel:
Key challenge as reported by FOCUS: 40% of businesses that were shut down for 3 days failed within 3 years. If your organization cannot recover from a disaster within 24-36 hours, you can expect to face the strong possibility of closing your doors permanently.
What's prevented DR: complexity and cost.
What's different with virtualization: lowered completed and lowered cost, by leveraging the encapsulated VMDK files.
Note that the DR stakes are higher than before. Virtualization is an enabler but it also increases DR exposure: if just one server goes down it has a bigger impact on the business.
Also reported by FOCUS, for the first time Disaster Recovery has surpassed consolidation as the top business driver pushing the adoption of virtual technology.
The Q&A to the panelists:
Q. PtoV DR is still difficult. When we tried to implement a project 18 months ago, we couldn't get over this hurdle and it killed the project. Is this still the case or are technologies available that can assist with this?
A. There are solutions which have been in the market for a long while which enable PtoV replication and conversion, including vReplicator and vConverter from Vizioncore. The key is to plan your DR in advance with all of the RPO/RTO levels that you require, and test and rehearse to prove your solution.
Q. With regards to SRM and replication, can you give me some idea of how these can interact? When it comes to VMware FT, are we losing transactions in flight or not?
A. VMware Fault Tolerance (FT) creates a shadow VM within a datacenter to provide a live standby VM available for mission-critical apps, tier 1 apps. SRM is the protection of multiple different datacenters, one-to-one and one-to-many. We also have customers using it in a ring configuration, and it's less about DR and more about load balancing. FT is the highest end - takes up storage and CPU capacity to deliver the higher recovery level. HA restarts the app, but FT is immediately available to pick up the workload. SRM is much more rigourous automation of restart of the entire site.
Double-Take is a file-based replication solution, which can be used to move data under the apps. vReplicator from Vizioncore is an image-based replication solution which is designed to move VMDK images around under the apps.
Q. For your DR site, is it recommended or mandatory to have a cluster? In-house vs outsourced: you stated it poses some challenges, can you specify what those are?
A. If the goal is to bring downtime to zero, then clustering is what you need. But this is relatively expensive and not what most organizations require. To enable efficient recovery with images, what's needed is the replicated image combined with some means to test and automate the restart of the image replicas at the secondary site. Clustering is really not required. For example, with vReplicator you can replicate the image, automate the test process, and restart the image to bring back the VM along with all of its applications and data very quickly and easily.
Inventorying your VMs and selecting those which must be protected in the environment is key to success of DR, and one of the new challenges faced with a virtual server environment. Everyone is familiar with VM sprawl and the associated problems in administering the virtual server environment. As applied to DR, it becomes more important than ever to keep track of which VMs are key to your business, make sure that these are the images which are included in your replication, and then test their recoverability on a regular basis to make sure that you're meeting the RPO/RTO objectives that your organization requires.
Q. There are a wide variety of tools available to apply to DR. Can you say a few words about how these tools compare and which should be applied to which parts of the job?
A. Backup and replication are merging, which enables a full range of DR benefits. These include being able to selectively replicate the VMs that you need to recover more quickly and combining that with use of restore from backup for the remaining VMs. Also, being able to relocate backup to your secondary DR site by backing up the replica VMs makes backup more cost-effective and easier to manage around centralized data from a number of sites. This is really not very different from traditional, except that the image includes more types of data so your protection levels are much higher covering all of the types of data that you need to fully recover.
This session was well attended, indicating the high amount of interest in the topic of applying virtualization to Disaster Recovery.