Shannon Snowden, Senior Technical Architect at Zerto.
The concept of the software-defined data center (SDDC) rose to prominence during VMworld 2012, with VMware touting it as the next big leap forward in information technology management. The term refers to an IT facility where the networking, storage, CPU and security are virtualized and delivered as a service. Furthermore, the provisioning and operation of the entire infrastructure is completely automated by software. This integration and automation brings everything in the infrastructure together and allows for a high degree of flexibility. It also means the person managing the data center can more efficiently utilize resources while providing better service to the company. Finally, the concept works at the hypervisor level, which allows the company to better and more fully utilize its hardware.
The Logical Next Step
It’s important and exciting to note that the realization of this concept is already happening. More and more companies are setting up data centers or thinking of data centers in this way, which means that related spin-off concepts have begun to appear. For example, software-defined networking has been coined as an approach in which networking control is decoupled from hardware and controlled by a software application.
Because many IT organizations are comfortable with the concept – and implementation – of SDDC, the logical next step is the adaption of software-defined disaster recovery (SDDR.) This approach represents complete hardware abstraction, meaning disaster recovery is no longer tied to the hardware. It’s now part of the software, which makes the process of replicating and recovering significantly more efficient.
Maximum Efficiency and Flexibility
To be specific, implementing the concept of SDDR allows for simple disaster recovery implementation by extending the benefits of virtualization, such as flexibility and portability, to Disaster Recover (DR). The management of DR is also greatly improved, as SDDR incorporates replication and recovery of multiple sites and can be initiated and maintained from one central location. Individual virtual machines or entire virtualized data centers can now be moved to and from any physical location, either to avoid disasters or as part of scheduled data center relocations.
SDDR works at the hypervisor layer and does not depend on hardware. When one compares a hardware-based deployment to an SDDR deployment, the benefits really come into focus. For a hardware-based disaster recovery solution, a multi-site deployment is very difficult, expensive and typically requires on-site configuration by the hardware and software vendors. The build time can be anywhere from several days to several weeks, depending on the number of sites. Contrast that with the typical multi-site SDDR deployment that normally takes about an hour to install and configure, is done remotely and includes testing sample failovers and failbacks.
The SDDR concept allows for fast recovery – near-continuous data protection combined with non-disruptive DR testing allows for recovery time objectives of minutes instead of hours or days. There’s also no reliance on inefficient storage snapshots, whose associated delays in recovery can cause major performance issues. Per-VM replication provides efficient bandwidth usage with minimal network impact and the model enables maximum flexibility, where VMs can be protected and recovered to private, public or a hybrid clouds.
Does this sound familiar? If this sounds familiar, it should! This is essentially the same argument made for server virtualization versus physical servers that we have all witnessed as fact for years. SDDR just expands the portability benefits to encompass multiple geographically separated sites.
Impact on Cloud Service Providers
For all the talk of DR being something that could easily move to the cloud, reality shows that DR specifically has lagged behind other cloud services being offered by cloud service providers (CSPs). DR has proven to be much more complex to provide than other cloud services. There are three irrefutable requirements that have to be addressed in order to provide cloud-based DR services successfully that SDDR can meet and physical hardware cannot.
1. It must be hardware agnostic.
In a hardware solution, the provider has to connect disparate hardware at a deeper level in the infrastructure than is normally needed for other cloud-based services. While the CSP might match one customer’s hardware, it is highly unlikely to match very many customers at the hardware level. SDDR works at the virtualization level so that cloud service providers can offer disaster recovery services to companies that do not have the same hardware as the CSP, specifically because they are leveraging the benefits of virtualization.
2. It has to have integrated multi-tenant support.
CSPs have to build a multi-tenant environment– this means that they have many customer environments, but that each of those are completely separate and secure. For DR purposes, multi-tenancy is quite complicated to achieve and manage if the DR tools are not designed to be multi-tenant.
3. It has to have streamlined customer management capability.
SDDR attracts customers to cloud-based DR unlike anything that previously existed. The customer base tends to grow very rapidly and can crush an ill-prepared support department. Having a DR tool that is designed for ease of management with profile driven, self-service customer interaction tools is critical for CSPs to successfully offer DRaaS in a cost-effective manner. Providers are able to offer global management and service level reporting for customers, regardless of VM location.
Benefits to Providers and Consumers
SDDR is so effective that is has essentially created a new service offering for CSPs. The SDDR model allows for rapid, flexible growth and enables CSP to initiate their first service offering or to expand their existing service offering with the ability to meet aggressive service level agreements and compete for the top tier DR services.
With the advanced multi-tenant management capability, CSPs can be a fully managed DR as a service offering to some customers or more of a customer-initiated, self-service DR provider if that is what the customer prefers. In fact, SDDR is a great entry point for companies that possibly have never considered cloud-based services before because of the significant savings in capital and operational costs of not needing a dedicated DR site.
SDDR makes both providing and consuming DR as a service better because it removes the dependency on hardware, supports multiple sites, streamlines DR management and leverages virtualization to its fullest extent.
Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library.