Do you flash, or know someone that does? Did you know most IT professionals are flashing today? Ok, no, get your mind out of the gutter. I’m not talking about streaking down the football field and exposing yourself to strangers, but that other type of flash. The type that helped us move away from large cassette players to tiny iPods.
When flash first entered the market, it was a high cost novelty item that allowed users to store and transfer limited amounts of data. As with all other major technology shifts, flash continued to gain in popularity and into the commodity item it is today.
Simultaneously, applications have become smarter and the explosion of data mining and online transactions has increased our need to become more productive at pushing data through our IT centers. IT managers are challenged with finding less expensive ways to meet these ever growing performance demands. Gone are the days when managers thought of storage as $/GB. Today, we think of high end storage as $/IOP. In return, IT managers are turning to Flash instead of paying for hundreds of spinning disks.
As with everything in the realm of storage, not all applications and performance needs are created equal. Flash is still relatively expensive when thinking of $/GB, so IT decision makers want to ensure they are investing in the right tier of flash technology for their specific needs. There are a ton of specialized manufacturers out there with various ways to optimize performance challenges. They all state that their way is the best, when in reality each serves a very specific purpose and can meet performance challenges around IOPS and latency at very different cost structures. I’d to take a moment to go through these different tiers of Flash.
Server Cache provides the lowest latency at <100 micro-seconds, increasing the total performance of all applications residing on the server. This reduced latency is needed for applications that are doing major reads and writes and need tremendous throughput like OLTP, Analytics and Reporting applications.
Server Level Clustered Storage
Utilizing server level flash storage provides a small hit to latency at around 200 micro-seconds, but now IT can guarantee the performance of specific applications and data. While server cache is good for applications with heavy read/write to the same data, there are some applications where performance is needed for data that is not touched that often and wouldn’t reside in cache.
All Flash Arrays
All flash arrays come in third providing <millisecond latency and reduce throughput as IO needs to utilize another layer of networking to access the data. The benefits of this extra layer are access to all of the benefits of shared storage and SAN technologies: high availability, snapshots, storage level replication and a larger footprint to store data.
Finally, tiered storage augmented with flash technology can provide around millisecond latency and is also the least expensive method for providing added performance to an IT infrastructure. Providing the best of all worlds, performance where you need it and capacity where you don’t. Unless you have an application with extremely high IO needs requiring a very short latency, a tiered storage array can provide a solution such that your data is working at the right performance level and at the right price for the business.
As IT continues to meet the growing demands of businesses become more reliant on technology, Flash is there to help anywhere you need it.