In Part 3, we’re going to give you a real-world example to demonstrate how environment reviews can help you meet SLA requirements.
Recently, the board of directors dictated that a particular application has an SLA requiring no more than a 12 hour RPO and an 8 hour RTO. An environment review discovered that backups are conducted on this application every night at midnight and take 3 hours to complete. According to the maintenance contract on the hardware, it could take up to 4 hours to get replacement hardware onsite. Further, it’s estimated that it will take an additional hour to rack, 2 hours to install operating systems, 2 hours to install the application, and 6 hours to recover the data from tape.
Here is the worst case scenario for this application:
27 Hour RPO (nightly backups occur daily at midnight and take 3 hours to complete)
15 Hour RTO (4+1+2+2+6 hours from above)
In the above example, the SLA will not be met in the case of an unexpected outage. After many meetings and architectural reviews, it was determined that the following be done to ensure this application meets the SLA requirements.
- Change the backup methodology to take 1 point-in-time copy of the data and copy that data to tape every 6 hours (snapshot, then backup to tape). This requires additional tape drives and media for the environment to accommodate the extra workload as well as additional licensing for the snapshot feature on the existing disk subsystem. These changes will require additional administrative time to configure, script, and monitor. These steps should reduce this application’s RPO to 10 hours (6 hour backup intervals, 4 hours to copy to tape).
- Purchase server hardware to be onsite with the operating system and application staged. This will reduce the RTO time to 6 hours (recovering 9 hours from the RTO – 4 hours to get hardware onsite, 1 hour to rack, 2 hours for operating system install and 2 hours for application install). Additional hardware cost, opportunity cost of the computer room, and administration of the new hardware were considered.
Performing the above bulleted items brought the RPO to 10 hours and the RPO to 6 hours. These changes met the requirements of the SLA but not without additional cost. There are many ways to meet an SLA and the case study above should not be taken as a recommendation or solution for every situation.
Each application in the environment should go through its own discovery process and independent recommendations should be to meet the defined its defined SLA(s). This can be a daunting task and it might be good to start with the top ten most important applications and work down from there. Gathering this information will play a key role in determining strategic backup and storage strategies for an organization.
Please feel free to ask one of our data protection specialist with any questions you may have with your environment.