It would seem that a solid-state drive caching appliance that sits in front of other storage systems would be the best way to add solid-state storage performance to our environment. What are the pros and cons of implementing solid-state this way?
Solid-state drive (SSD) cache appliances are an excellent bridge to an eventual solid-state storage infrastructure, especially for network-attached storage environments. But this technology has some drawbacks, most notably the lack of appliances that are able to cache block I/O, since most appliances specifically accelerate NFS.
SSD cache appliances are positioned logically between the requesting applications and the storage. They typically have a bank of SSDs and some form of network optimization such as IP offload or storage protocol optimization.
Caching appliances act as an I/O "shock absorber" for the storage system by offloading a high percentage of inbound read requests destined for the storage system they sit in front of. As mentioned, most caching appliances on the market only accelerate NFS traffic at this time. If you're using CIFS, iSCSI or Fibre Channel, a caching appliance might not be an option.
How the application will perform in the event of a cache miss is another concern. A cache miss means the data wasn't in the SSD tier of the cache appliance and must be fetched from the hard disk drives (HDDs) of the storage device. The problem is that the performance impact of this event is very hard to test or evaluate, and the ramifications of a miss can range from a nonissue to a major problem.
If you have certain applications where occasional slow HDD performance is unacceptable, then you should look for a caching offering that will allow you to lock certain portions of data into the cache, commonly called cache pinning, or you should evaluate an all-flash array or SSD appliance.
About the expert:
George Crump is a longtime contributor to TechTarget, as well as president and founder of Storage Switzerland LLC, an IT analyst firm focused on the storage and virtualization segments. Before founding Storage Switzerland, George was chief technology officer in charge of technology testing, integration and product selection at one of the largest data storage integrators in the U.S.
Dig deeper on Data management tools
Related Q&A from George Crump
To determine the amount of flash to use with your hyper-converged system, take into account whether it will be used as a pool across all servers or ...continue reading
Hyper-converged architectures make provisioning storage for VMs simpler by integrating it into configuration options.continue reading
George Crump of Storage Switzerland offers insight on finding the best choice for backing up Active Directory in this Expert Answer.continue reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.