Hosted Recovery Solutions
There’s a common saying among datacenter professionals that no matter how much it costs to have a disaster recovery program, the cost of not having one is far greater.
This was often used to convince skeptical CEOs of the need to build duplicate datacenter infrastructure despite its high cost and low utility. Lately, however, the cost side of disaster recovery has dropped precipitously as modern hosted solutions allow enterprises to avoid much or all of the up-front expenses while still providing reliability and availability levels equal to, or even better than, in-house facilities.
There are, of course, numerous ways to implement hosted disaster recovery services, all of which can deliver varying levels of utility based on user needs and existing infrastructure. In a broad sense, however, hosted disaster recovery platforms fall into two camps: Disaster Recovery plan as a Service (DRaaS) and the lesser-known Disaster Recovery on Cloud.
Quite often, these two deployment options are treated as one-and-the-same. But there are some key differences that could have a significant impact on costs, levels of service and other performance factors.
Disaster Recovery on Cloud (DRC)
DRC is best described as the OpEx version of hosted disaster recovery. It calls for a fully virtual enterprise environment ported to the cloud under an Infrastructure as a Service (IaaS) model. Its chief advantage is that it allows the enterprise to avoid the substantial capital costs of building bricks-and-mortar data facilities that are called into service only in the event of a serious emergency.
As a leased infrastructure solution, DRC gives the enterprise enormous flexibility when it comes to scaling DR capabilities up or down depending on data and infrastructure needs. With proper management and data visibility tools, enterprises pay only for the resources they actually use, a far more efficient use compared to traditional facilities, which are usually over-provisioned to ensure data and application availability. As well, DRC allows the management of physical-layer assets to be offloaded to the host provider, which not only lowers costs but ensures a robust upgrade path for both critical and non-critical systems. As well, leased infrastructure unburdens the enterprise of many licensing fees, power and cooling expenses, and labor costs.
Because DRC alleviates much of the complexity surrounding disaster recovery, enterprises should see more robust performance compared to traditional infrastructure. Essentially, the DRC environment acts as an extension of existing virtual infrastructure, with restoration of server images enabled through physical-to-virtual and/or virtual-to-virtual technologies. This, in turn, allows for much quicker re-establishment of normal operations than if, say, entire server farms, storage arrays and network configurations had to be rebuilt from scratch.
One popular misconception about DR in general is that it only becomes active during a recovery. But this is only the case for enterprises that view DR as an afterthought. A properly managed DR infrastructure incorporates a wide variety of daily, weekly, monthly and semi-annual procedures involving backup, replication, system maintenance, integration with primary infrastructure and so on. As well recovery drills and mock scenarios should have a place in every DR program. In a DRC environment, many of these tasks can be automated, with little or no impact on the day-to-day operations of the enterprise. When not in use, DRC-based virtual machines can be switched to an off-state, helping to reduce operational costs.
The true test of any DR system, of course, is how well it functions in an emergency. The two key metrics for any disaster recovery platform are the Recovery Point Objective (RPO) − the amount of data loss deemed acceptable before a scheduled backup takes place − and the Recovery Time Objective (RTO) − the acceptable level of downtime following a failure. In both instances, virtual hosted DR on Cloud platforms are matching or exceeding the top traditional DR systems.
DRC works best for enterprises that are already outfitted with a virtual disaster recovery architecture but are straining to maintain the physical infrastructure needed to host it. As well, highly regulated industries like banking, finance and insurance may have difficulty maintaining a third-party system due to stringent discovery and compliance requirements. Media, web-facing companies and manufacturing are probably more suited to DRC solutions. And, of course, enterprises that are already experimenting with cloud architectures will most likely find that DRC makes an excellent addition to storage or even application/service environments.
Disaster Recovery as a Service (DRaaS)
While DRC is an infrastructure solution, Disaster Recovery as a Service provides both the infrastructure and the backup and recovery platform itself. As such, it is a solution geared more toward organizations looking to build a low-cost, flexible recovery platform from scratch.
The best way to view the difference between DRC and DRaaS is that DRC is like a laptop fresh from the factory without any bundled software or even an operating system, while DRaaS is the same laptop with all the software and other resources needed to perform its given tasks.
In short, DRaaS is a tool-based end-to-end DR solution encompassing everything from infrastructure functions like server and network provisioning to database installation, management and monitoring. This tool-driven approach allows features like failover drills, compliance reporting, remote replication and recovery automation to be configured to suit customer needs.
As well, DRaaS provides a wide variety of data protection and recovery options that would be prohibitively expensive using traditional private recovery platforms. In this way, DRaaS provides the exact level of service required for enterprise needs, without having to over-provision either hardware or software in order to maintain the highest possible reliability and availability ratings.
In general, DRaaS is available on three tiers of service. As a hot backup solution, DRaaS offers mirrored servers in stand-by mode that are ready to kick in as soon as an outage occurs. This provides minimal RTO and RPO using synchronous replication, and is usually reserved for the most critical enterprise applications to allow the enterprise to maintain functionality of vital resources while full recovery proceeds for the remainder of the data ecosystem.
In warm backup mode, data is kept up-to-date using either synchronous or asynchronous replication based on the RPO needs of the enterprise. Servers are kept in warm idle mode and can be brought on-line within a few minutes, providing a high level of recoverability for most applications. Most enterprises find the warm mode to be adequate for their needs, but again it depends on the critical nature of the application and the service requirements of users.
Consequently, then, DRaaS can also provide a cold backup site, in which data is replicated periodically and recovery can take hours or even days. Cold service usually requires hardware to be brought online or repurposed from test and development systems, which means it provides the lowest level of technology business continuity. The advantage, however, is that cold service is available at extremely low cost and is thus reserved for lowest-priority data and applications.
It would be a mistake, however, to view the hot, warm, or cold tiers of DRaaS in terms of superiority or inferiority. Each level provides unique cost and capability benefits for particular data loads. And in fact, many enterprises find that a combination of tiers provides the most robust recovery solution at the lowest cost because they can be used to tailor a recovery platform according to the often changing nature of modern enterprise environments.
DRaaS is also associated with a broad set of services designed to remove much of the burden of building and maintaining reliable recovery infrastructure from the enterprise. These include consulting services to provide risk analysis and evaluate the business impact of various DR configurations, as well as cost-benefit analyses to align available services to enterprise needs. Disaster recovery requires broad synchronization between its various functions to maintain adequate RTO/RPO, not to mention regular testing and drills to maintain system readiness.
As well, DRaaS offers a full set of replication services that assume responsibility for much of the day-to-day backup responsibilities that ensure a speedy and effective recovery when the need arises. The most common approaches involve either host- or storage-based replication, coupled with deduplication, compression and encryption licensed from various third-party providers
One thing is clear: the old way of backing up and recovering data is proving too expensive and too cumbersome for the agile, dynamic data environments to today. The cost of buying, provisioning and maintaining vast infrastructure can no longer be justified in an era when profit margins are tight and the need to do more with less is growing.
And now that hosted services can provide a low-cost infrastructure-only solution for legacy platforms or full disaster recovery for those without a program at all, the time is finally right to investigate a third-party solution.
It’s a safe bet your competitors already are.
By Mandar Kulkarni,
Mandar is Vice President – Solutions Engineering and Private Cloud Practice at Netmagic IT Services Pvt. Ltd. He is responsible for planning, directing and leading the Netmagic Operations. He heads the team of service and support professionals who deliver quality and timely services to Netmagic’s 1000+ customers in their datacenters and across multiple locations.