Vulnerabilities / Threats // Insider Threats
6/24/2013
12:29 PM
George Crump
George Crump
Commentary
50%
50%

Should You Use Software-Defined Flash Storage?

"Software definition" leverages the processor's power to do things that used to require dedicated hardware. So, what does that mean for flash?

Have you noticed how everyone seems to be adding the words "software-defined" to their market term? In some cases this is just marketing gone wild. In other cases it is an appropriate description of the technology. We have software-defined storage, software-defined networking and the software-defined data center. I expect to hear to a CEO describe their entire organization as "software defined" any day now.

All this software definition is changing how technologies are developed. An excellent example is flash-based storage.

What Is Software Definition?

The use (and overuse) of the term software-defined is really about leveraging the power of the Intel CPU to do things in software that we used to do with dedicated hardware. Doing so should reduce costs and increase flexibility. As I described in my last column, "How Software-Defined Storage Impacts SSD," it can even help with adoption of hardware, like flash media and appliances.

Software-Defined Flash

The value of software definition is now working its way into the flash community and is the subject of much debate. The flash controller performs a variety of management functions that extend the life of the media and improve performance. Typically this flash intelligence is embedded as silicon or a field programmable gate array (FPGA), but now some flash vendors are starting to perform these management functions using the CPU of the host or storage system that the device is being installed on.

[ For more on software-defined storage, see Software-Defined Storage Vs. Traditional Storage Systems. ]

The upside of this method should be more rapid deployment of capabilities and functionality. It should also lead to lower development cost for the vendor. In theory this should lead to less-expensive products that are more feature rich. We have not, for the most part, seen this theory translate to reality. Most flash systems and devices that count on host CPUs to perform management functions are the same price as hardware-based flash systems and typically are at feature parity.

The other potential advantage of software-based flash management is that as you upgrade the CPU in the host you could see better flash performance. How much of this is reality is dependent on your application and whether you will actually carry your SSD investment between generations of servers as you upgrade them. In our experience this is not the case.

The downside of counting on the host CPU to execute the flash management functions is that it could impact flash and/or host performance when under stress. While the vendors that perform their flash management in hardware make a big deal about this potential problem, we have not seen a situation where it is apparent. As most IT professionals know, there is plenty of excess CPU resources in the typical data center today. Burning a few of those cycles on flash management routines should not be a major problem for most.

Should You Use Software-Defined Flash?

This is, at least right now, one of those situations where "it depends" doesn't apply. More appropriately, at least for right now, in most data centers it doesn't matter. You need to judge the flash device or storage system on its total merits. Does it deliver the performance you need, with a feature set you can live with at a price you can afford? While the advantages of hardware-based flash management or software-based management may become more apparent in the future, at least for today they are a wash.

Comment  | 
Print  | 
More Insights
Register for Dark Reading Newsletters
White Papers
Cartoon
Current Issue
Flash Poll
Video
Slideshows
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
CVE-2013-7421
Published: 2015-03-02
The Crypto API in the Linux kernel before 3.18.5 allows local users to load arbitrary kernel modules via a bind system call for an AF_ALG socket with a module name in the salg_name field, a different vulnerability than CVE-2014-9644.

CVE-2014-8160
Published: 2015-03-02
net/netfilter/nf_conntrack_proto_generic.c in the Linux kernel before 3.18 generates incorrect conntrack entries during handling of certain iptables rule sets for the SCTP, DCCP, GRE, and UDP-Lite protocols, which allows remote attackers to bypass intended access restrictions via packets with disall...

CVE-2014-9644
Published: 2015-03-02
The Crypto API in the Linux kernel before 3.18.5 allows local users to load arbitrary kernel modules via a bind system call for an AF_ALG socket with a parenthesized module template expression in the salg_name field, as demonstrated by the vfat(aes) expression, a different vulnerability than CVE-201...

CVE-2015-0239
Published: 2015-03-02
The em_sysenter function in arch/x86/kvm/emulate.c in the Linux kernel before 3.18.5, when the guest OS lacks SYSENTER MSR initialization, allows guest OS users to gain guest OS privileges or cause a denial of service (guest OS crash) by triggering use of a 16-bit code segment for emulation of a SYS...

CVE-2014-8921
Published: 2015-03-01
The IBM Notes Traveler Companion application 1.0 and 1.1 before 201411010515 for Window Phone, as distributed in IBM Notes Traveler 9.0.1, does not properly restrict the number of executions of the automatic configuration option, which makes it easier for remote attackers to capture credentials by c...

Dark Reading Radio
Archived Dark Reading Radio
How can security professionals better engage with their peers, both in person and online? In this Dark Reading Radio show, we will talk to leaders at some of the security industrys professional organizations about how security pros can get more involved with their colleagues in the same industry, with their peers in other industries, and with the IT security community as a whole.