Everyone needs some summer downtime. If you have not taken your summer downtime yet I hope you have plans to do it soon. However when it comes to business, the last thing we need on a vacation is a frantic phone call that your business has taken some downtime. While downtime in IT systems is unavoidable the question is about how much resilience is built into the system architecture. In fact availability is one of the hallmarks of a modern data center and should be considered in any data center modernization effort. Continuous Availability, unlike other loosely used definitions of high availability, is about delivering a zero RTO service level to ensure business critical applications are never down. A continuous availability technology is able to do this by making sure every IO is captured for instant replication across distance thereby creating active-active data centers that offer highest disaster resilience and also eliminate planned downtime. Let us look at some of the considerations for choosing the right continuous availability technology for the most demanding all-flash workloads.
Performance that Maximizes All-Flash Storage Availability
IT organizations around the world are rapidly adopting all-flash arrays to consolidate their business critical applications with uncompromising performance and data services with significantly better total economic value. A zero RTO availability technology needs to be the most efficient data mover to replicate at a speed that matches flash performance. Needless to say, a technology that uses array cycles for replication is eating into the flash performance. This is the reason why a dedicated availability solution like VPLEX has become an obvious choice for more than half of Global Fortune 500. At the same time you don’t need the availability technology to be duplicating the functionality that the all-flash array already has. Instead the replication technology should maximize the availability of the flash storage that is running more and more business critical workloads.
Future Ready Scale
As famous ice hockey player Wayne Gretzky once said “skate to where the puck is going to be, not where it has been”. Availability technology that is in the data path has to be ready to handle at least twice the IOPS and workloads that it is being purchased for. Clearly the hardware platform needs to have enough room to grow both in terms of scale out architecture as well as enough compute power that future software releases can take advantage of using increasingly parallel computing algorithms.
While linear scaling in performance is appreciated; the licensing costs for a growing environment quickly becomes unattractive. This is where customers should look for licensing models that are more favorable for growing environments. (more…)