Hybrid All-Flash Arrays drive down cost by mixing high performance flash with cost effective hard disk drives (HDD). Establishing predictable performance is the chief concern for Hybrid Array buyers. As long as data is being accessed from the flash tier, performance should be as good as an all- flash array. But when data has to be accessed from the HDD tier there can be a noticeable decrease in performance. In our on demand webinar, “How To Make Hybrid Perform Like All-Flash”, we discuss the steps that IT professionals can take to make sure their hybrid array provides predictable, high performance similar to an All-Flash Array.
Register for the on demand webinar “How To Make Hybrid Perform Like All-Flash"
There are four steps to getting all-flash performance from a Hybrid Array and in this webinar you will learn about each of them. The first step is to properly size the flash tier. Most All-Flash vendors are overly conservative about how large the flash tier should be. They try to get you into the smallest flash array possible so that the cost per GB will be more attractive. While cost per GB is important, so is consistent performance. The key is to size the flash tier so that is more than adequate for your active data set. You don’t want to be “penny wise and pound foolish” in this decision.
The next step is to make sure that the flash tier is optimized. This means a hybrid system with a robust data efficiency feature set. This data efficiency offering should include deduplication, compression, writable snapshots (clones) and thin provisioning. The goal is to make sure that every ounce of capacity is extracted from the flash investment.
The third step is to identify mission critical workloads and lock those workloads into cache. The storage system should allow you to pin to the flash tier both bare metal, non-virtualized systems, as well as specific virtual machines on a physical host. This ensures that these mission critical systems will always have predictable, high performance.
The final step may be the hardest, but it also may provide the highest return on the flash investment, optimizing application code. Application code is often most responsible for inefficient flash utilization. This inefficiency can include poorly designed indexes, unnecessary queries and extensive reports that dilute the cache with old data. Troubleshooting these problems often requires database analysis tools, but the investment in those tools can save thousands of dollars in additional flash purchases.
To properly implement these steps requires a little pre-planning and a hybrid storage system that has data efficiency and cache pinning features. In our webinar we will discuss the process and the hardware that can make it all possible.