Deploying server-side cache is an effective way to accelerate application workloads; whether they are hosted on bare metal servers or virtualized infrastructure. But with so many flash and SSD options to choose from – server-side, all flash arrays, hybrid storage systems, etc. IT decision makers may be concerned about which solution is best for their business. Furthermore, many organizations already have various forms of flash distributed throughout the data center. How can all of these resources be efficiently and effectively managed without resorting to multiple “panes of glass”?
The good news is that when the right type of server caching software intelligence is integrated into the environment, data centers can utilize and benefit from multiple forms of flash and SSD and manage them all from a centralized interface.
Capacity and Performance De-Coupling
Perhaps one of the biggest challenges for organizations today is scaling the performance of their virtualized application workloads without resorting to costly primary storage upgrades. In the past when storage performance couldn’t keep up with application I/O demands, storage administrators would either have to add more hard disk drive capacity to their systems and resort to techniques like short-stroking drives (only writing data to the innermost portion of the disk platter), adding flash modules to their arrays or doing a wholesale disk array refresh.
While these approaches for improving performance can buy organizations some time, ultimately they are not a cost-effective long-term way to scale storage for the environment. Server-side storage options like PCIe flash and drive form factor SSD can help augment storage I/O performance by placing very fast storage resources directly where the applications live – on the servers themselves. What’s more, when application aware caching intelligence is installed, businesses may continue using their existing primary storage assets. Under this model, server-side PCIe flash and SSD can effectively function as the performance tier, while conventional hard disk drive arrays can operate as the capacity or data storage tier.
But many organizations have other forms of memory based storage in their environment that they could take advantage of to solve storage I/O bottlenecks. There may be DRAM modules inside some of their hypervisor servers and/or there could be flash and SSD modules populated in their primary storage arrays. How can these assets also be leveraged and centrally managed when server-side caching is implemented?
There are some server-side caching technologies that enable organizations to utilize other forms of SSD in a couple of unique ways. First, some server-side caching solutions can be used to create a “multi-level cache”. For example, since DRAM provides storage performance I/O speeds that exceed flash or SSD, these resources can be configured to function as the primary storage performance tier for the most I/O demanding application workloads.
Another highly useful attribute of some server-side caching offerings is the ability to create a “remote cache”. This is a particularly important capability since it allows for the seamless movement of virtual machines (VMs) across hypervisors – otherwise known as server vMotion. For example, by mirroring read data from local server-side flash to a remote cache, like a flash or SSD device on an external storage array, newly migrated VMs can immediately reference their hot data sets from the remote cache. Without this feature, all application data would have to be referenced from hard disk drives on the storage array until another cache resource could be “re-warmed” with all the active hot data sets.
In fact, these additional capabilities can be seamlessly implemented by deploying a server-side cache acceleration software layer that can deliver a single point of management for all flash resources in the data center.
For more information on how server-side caching can compliment other forms of flash based storage, please join us for an informative webinar with Intel on Wednesday April 29th. Pre-registrants will receive an advanced copy of the Storage Switzerland article, “How To Choose The Right Application Caching Architecture”. Once you register, you will have access to over 60 on-demand versions of Storage Switzerland webinars and you will never have to re-register again for a future event.
Intel is a client of Storage Switzerland