For veterans of the IT services industry, DR has always been a popular topic of conversation with potential clients. Those that have been around long enough will certainly remember how many of those conversations progressed.
Typically, it went something like this.
Potential Client: We’ve determined that our current IT infrastructure DR plan puts our business at risk and we are interviewing service providers to assess potential solutions.
IT Services Vendor: What are your infrastructure RPO and RTO targets?
Potential Client: Our CIO wants us to maintain a RPO/RTO of 4 hours or less.
It wasn’t that long ago that everyone in this conversation would have understood that the quote from the service provider was going to be well beyond what the client intended to spend as part of the overall IT budget. This was typical for both small and large environments. Inevitably, the parties would work backwards by decreasing the expected deliverables for the solution until an acceptable price point could be reached. Sometimes the solution met so few of the organization’s requirements, that the conversation would be abandoned with no action.
Was the CIO delusional for requesting such aggressive (for the time) SLAs? Of course not. The importance of the data and the underlying applications and infrastructure was self-evident. The reality was that, not only were the options to meet those goals extremely expensive, there was very little guarantee that it would work as planned when it came down to crunch time.
The reason for the expense was that each production resource had to be duplicated, to a certain extent, at the remote site. This infrastructure would need to be purchased or leased, co-located, upgraded and required experienced technicians to maintain. All of this in hope that it would never have to be used in a live situation.
Fast forward to the present and with the evolution of virtualized workloads, resource pools, metered billing and any to any replication technology, those RPO/RTO targets are now achievable and at a fraction of the cost. The underlying services billing model that makes this a reality consists of a reserved billing storage component for data replication and burstable billing compute resources that can be deployed on demand and be billed per hour of use.
Reserved storage provides a target storage repository sized to handle all replicated workloads plus potential growth dependent on changes in the production environment. Reserved storage is billed on a per GB per month basis. The storage reservation quantity can be increased at any time to mirror changes in the production environment.
Burst compute refers to on-demand CPU and RAM which are necessary to operate the virtual workloads during production failover or testing. Because replication is accomplished without live workloads, the burst compute resources are available on demand and no charges are incurred until the workloads are powered on. CPU is metered on average GHz of CPU used per hour. RAM is metered as average gigabytes (GB) consumed per hour. These burst compute charges are tallied and billed monthly. When testing or failback is complete, the resources are returned and the burst charges are no longer accrued.
While cost is still top of mind for IT Directors and CIOs, the conversations around solutions for IT's data protection and DR needs are drastically different. Reserved storage plus burst compute pricing for DRaaS allows IT organizations to execute a robust disaster recovery plan without having to pay for live compute resources waiting for use. The major obstacles to a credible DR solution, even for small businesses, have been mitigated by technology advances and wide spread adoption of said advances.
Once the question of cost has been addressed, the discussion moves to more important issues. How do end users connect to the DR environment once failover is complete? Does the recovery site adhere to the same security standards as my production environment? How is failback accomplished? These are just a few of many important questions not related to cost.
In conclusion, the reserved plus burst model allows customers to apply the advantages of two pricing models where it makes the most sense thereby protecting critical data without the burden of barely used, monthly infrastructure costs at the service provider location. A comprehensive solution will also provide assisted initial setup, volume discounts for storage, simplified day-to-day operations via a self-service console, straightforward network configuration, the option for customer initiated failover, as well as detailed billing, monitoring and compliance reporting.