Continuous Availability: More On What You Need to Know (Part 2)

According to the Forrester Research study that I mentioned in my previous blog post, there still is a lot of confusion in the marketplace about continuous availability (CA), including what it is and how it works in an organization to improve service availability while reducing costs.  That is why I am posting some more questions and answers on continuous availability.  These Q&As are representative of the dialogues I have been having with IT executives in the marketplace:

Question:    How is a continuously available application less expensive that a non-CA application?

Mission critical and business critical (5 x 9s and 3 x 9s) application are typically deployed with High Availability (HA), which is same-site redundancy, either with spares in a compute farm or clustered servers deployed (usually) in active/passive mode.  Most mission and business critical applications also have complete or nearly complete redundant processing capabilities in a disaster recovery site.  These architecture redundancies within a site, along with redundant capabilities in the Disaster Recovery (DR) site, result in a lot of idle assets.   In some cases the redundancy is a high as 200% of the basic need! 

Savings are realized through combining HA and DR practices and by eliminating or nearly eliminating idle assets.  Additional cost savings can be made with fractional computing and infrastructure.  Quite often we see reductions in server counts by as much as 60% in these environments.  With a reduction in server counts, acquisition cost is reduced, license costs reduced, along with the power, rent, and personnel to maintain the equipment.  Additional cost savings are seen by redeploying DR personnel to other BC or availability roles and freeing up application resources from DR testing and run book maintenance.

Question:    So, does CA mode pay for itself?

For applications with HA protection and DR coverage, CA pays for itself in the cost savings realized from a converged HA and DR environment.  Often triple server costs (HA fail-over and DR fail-over) can be reduced to a fraction of the cost for acquisition, power, floor space, licensing, maintenance, and administration.  Cost is eliminated for DR solution maintenance, testing and administration.  

Question:    Why is CA better than HA and DR?

Continuous Availability offers organizations the opportunity to consolidate disparate Disaster Recovery and High Availability programs.  Consider that there are several ways to provide server High Availability — for example, clustered passive servers or extra servers in a web or application farm.  On the DR side, the options include: array based replication; host replication; database replication; backup image replication; fail-over managers; etc.  Each individual technique is probably simple and effective, but when you add them all up, the environment is suddenly complex.  Also consider that most mission- and business-critical applicationss have both DR and HA solutions.  Just by combining HA and DR solutions, complexity is reduced and cost savings can be realized.

From a data loss and availability perspective, Continuous Availability solutions can eliminate data loss and simplify the availability schemes.

Question:   If DR and HA are combined, do recovery point objectives (RPOs) and recovery time objectives (RTOs) go away?

Continuous Availability is a relatively new discipline, and it will take a while for regulators and auditors to catch up with technology capabilities.  In the US, FFEIC (banking) regulators require certain regulated entities to either test their DR solution or regularly use their recovery solution.  In financial institutions that have CA solutions deployed, regulators have required that the regulated entity demonstrate site-loss scenario testing

EMC recommends that organizations be prepared to educate their auditors on the converged HA and DR solution and show how it can eliminate recovery.   We also recommend that organizations develop and test site-offline and site-online run books. EMC often assists clients in the creation of these run books, which can provide a low impact, documented and rehearsed procedure to take a site offline and bring it back online. The run books should be used for site maintenance and are often used in lieu of traditional DR testing

Question:    If CA eliminates the effort, expense and risk of disaster recovery testing, will I have to transfer that effort and expense into testing the Continuous Availability solution, to assure it will work as planned?

Unlike DR, which is based on an outage event and recovery event model, Continuous Availability is Active/Active. There is no event or recovery that requires testing. However, over time, changes in applications, data sets, and connected infrastructure can change the optimum configuration parameters for the VPLEX, negatively impacting system performance. For this reason, EMC recommends that Continuous Availability customers also leverage EMC VPLEX Health Check services to periodically test the configuration and performance health of their system. These Health Checks help customers identify existing as well as potential configuration, process and other issues, and provide customers with actionable recommendations to improve performance and return on investment. The Health Check services are non-disruptive and are typically a fraction of the cost associated with planning and performing disaster recovery testing.