Disaster Recovery is always a serious topic in every data centre and protecting a corporate Enterprise is paramount in today’s world of safeguarding business intelligence. The person sitting in that $3000 Executive high-backed chair has come to expect leaning back in it every morning, coffee in one hand and 0700 reports in the other. In over 25 years in this industry I am yet to meet anyone who has looked forward to informing the CEO that their morning reports are delayed or not coming.
There are many discussion points when we talk about disaster recovery; from recovery plans and objectives, infrastructure consolidation, non-disruptive testing, and so on. I intend though, to approach the discussion on a component of replication strategy itself; specifically, hypervisor-based replication versus traditional storage layer replication. Even with the latter I’ve chosen to limit its discussion to reduce the complexity and dryness of this article.
VMware’s vCenter Site Recovery Manager works with vSphere Replication to provide Recovery Point Objectives from 15 minutes to 24 hours. vSphere Replication is a hypervisor-based replication appliance which is easy to use and configure and provides significantly reduced network overhead when compared with storage layer replication strategies, such as an asynchronous SAN replication solution.
Hypervisor-based replication also allows us to consolidate hardware letting distributed data centres and infrastructure act as business continuity solutions for each other. This eliminates the need for warm stand-by data centres. By consolidating the production and fail-over sites we eliminate hardware overhead and wasted clock-cycles as racks of servers no longer sit idle waiting for a moment that, with any luck, will never come.
Storage replication strategies have evolved significantly over the years, with advances in file-based and block-level storage replication technologies. Regardless though, storage-level replication continues to use significant bandwidth, compared to hypervisor-based block-level replication and still relies on kernel and filter drivers which in turn increase IO operations and CPU cycles.
Replicating at the hypervisor lets IT Operations manage disaster recovery for each virtual machine. Do we really need to replicate our test environment for disaster recovery purposes? Not likely. With storage-level replication that usually isn’t an option.; it’s all or nothing. With little more than a few mouse clicks, data centre administrators can quickly select virtual machines to be replicated, ignoring test or lab environments by discretion.
One of the reasons I recommend VMware as a virtualization technology is its broad developer base. Not a day passes without someone releasing a VMware tool, companion application or appliance that makes data centre administration easier.
Zerto is a prime example with their business continuity and disaster recovery solution. The Zerto Virtual Manager, combined with Virtualization Replication Appliances utilize block-level replication on the hypervisor. WAN compression, throttling, journal-based point-in-time recovery name but a few of the many capabilities of this feature rich offering.
If you find yourself tasked with designing a disaster recovery solution for your corporation my advice to you is reduce your efforts in researching and following traditional disaster recovery models and instead look towards today’s newer technologies such as hypervisor-based replication strategies.
Research resources:
• VMware Site Recovery Manger
• Microsoft Hyper-V Replica
• Zerto BC/DR