Dave Brown

Software-Defined Storage and How it Enables Continuity of Operations for Data Survivability

Did you know that software-defined storage (SDS), previously known in the industry as storage virtualization, enables continuity of operations (COOP) from the data availability and resiliency perspectives?

Well, luckily for you, if you didn’t you can read a bit further and find out more.

Through the capabilities of COTS-based SDS products that allow for heterogeneous mirroring and replicating of data to physically diverse locations, what had once been the weak point of a fixed or mobile data center, can now be equally as strong and hardened. This SDS capability strengthens the communications capability of the warfighter, just like server virtualization does for the server and application portions of a data center.

Data storage has long been, and to a very large degree still is, very homogenous. Once an environment is built upon a particular vendor’s brand and product line, it becomes very difficult operationally to add or replace that vendor’s product because there is no ‘standard’ when it comes to data storage communications like there is for example with a data packet / frame traveling through a multi-vendor ethernet switching environment. Of course there are common protocols used amongst storage vendors like fibre channel and iSCSI, although those protocols don’t extend to bridge any inter-vendor communication for the higher level values in data storage of synchronous mirroring, cache coherency or asynchronous replication. Beware there are some very new SDS products coming to market from the same vendors who used to poo-poo SDS, look to established and proven SDS products for betting the mission on.

SDS is a technology that originated, like most things in computing and storage, from the mainframe environment, only to be ‘reborn’ in the Open Systems world. The value of SDS is even greater than that of server virtualization, where the ‘win’ there is with the reduced acquisition and life cycle costs of servers required for a given mission, thus resulting in less power, cooling costs, less cost of space and weight. Also known as SWAP, which is particularly important for mobile or airborne payloads, and it also takes fewer staff to maintain and operate it.

In the virtualized server world you will still have the same costs for software licensing, even a little more because of the hypervisor licensing, with the savings coming from the reduced hardware spend. When you turnover your servers, you still retain your investment in all your software, which is perpetually licensed and not tied to a particular server.

Storage on the other hand is not that way. Storage has traditionally been, and continues to be, tied to the ‘stack’ of hardware. When the operational unit gets rid of their storage array for the next model because they either ran out of controller resource (CPU, cache, interface quantity/speed/type), disk resource (FC, SAS, SATA, SSD), the product went ‘end of sale’ and support costs started escalating, or ‘end of support,’ then all the investment in the storage service software is typically lost. When they buy the next array, they buy not only a new set of commodity storage hardware, they are buying the same storage services software they bought on the last array. Each storage array typically has several optional software features that are either ‘flat rated’ options or are ‘per feature per TB of capacity.’ With today’s ever constrained defense budget ($716 billion requested for FY2019), if there was a way to retain that investment, lowering the storage software life cycle cost to the unit, and even further lower the storage hardware lifecycle cost by leveraging an Open Architecture hardware model where commodity AMD or Intel based servers can be used, why not do it?

For example, incorporating SDS into the modern military’s data centers provides the power to leverage all the storage resource that is already in place. Except, SDS can make that resource more intelligent, higher performing, allows you to say ‘no thank you’ to renewing software feature support and allows for much more rapid adoption of the next generation of commodity hardware, even slowing the need to buy a replacement for an existing array (SAN). SDS abstraction of the storage resource into entities of virtual disks (full-featured ‘micro SANs’ independent of each other), just like a server hypervisor abstracts to virtual machines (virtual CPUs, memory and network interfaces), allows for discrete functions to be applied to those disks, like Synchronous Mirroring (full block-level complete with Cache Coherency), asynchronous remote replication, data migration, virtual disk pooling and continuous data protection.

COOP or Business Continuity in the Commercial World

The commercial world uses the term business continuity to signify that even through one or more issues exist in an environment, enough of the overall system is still up that the business continues. For instance, the military’s use of the term COOP is really very similar, that the mission of the given military branch and unit continues, even though issues to systems may come from normal hardware failures, or more drastic causes like enemy attacks.

How SDS assists in enabling COOP is through a capability called synchronous mirroring. Synchronous mirroring is when data can be mirrored between different places in a data center, between different mobile communications vans in a given area, and/or even across greater distances, provided the bandwidth is great enough and the latency is low enough, leveraging networks like the GIG-BE. SDS’s ‘equalizer effect’ enables disparate storage arrays (SANs) to now all have the same intelligence and communication language. See Figure 1, where the SDS Nodes can be spread over distance for battle hardening and the unit’s data can be mirrored between multiple locations so the mission continues. As the number of hosts scale, the SDS nodes can scale along with them.

Figure 1

Remote Replication

SDS’s equalizer effect further enables heterogeneous storage arrays (SANs) to now have the ability to asynchronously communicate over much greater (thousands of miles distances). This capability allows for information dissemination from the command down to the unit, command to command, unit to unit, etc.

When these asynchronous communication protocols are built on top of the TCP/IP stack, then sending even the most sensitive data is possible as the data can be sent through a Type-1 encryption unit before leaving the secured environment. See Figure 2, where the SDS Nodes can be spread over much greater distances, well beyond the range of even the most devastating weapons, so the mission, even COG, continues.

Figure 2

Virtual Disk Pooling & Data Migration

Virtual disk pooling & data migration are enablers for the warfighter unit to mix and match storage hardware for the mission. The functionality also allows for adapting as the mission changes, all the while still maintaining data availability to the host servers (whether virtual and/or physical) and their applications. Pooling underlying storage resources from different vendors, models and types (solid state or spinning) allows for using the best and most cost effective technology at the time and keeps from being tied to only what’s on a particular ‘menu.’ Data migration makes changing to that ‘next thing’ seamless to the command structure. In Figure 3, through SDS’s creation of the constant, a Virtual Disk, the ‘real disk’ on the back end can be pooled and migrated in and out as needed.

Figure 3

Continuous Data Protection

One of the mainstays of data protection can be seen as tape backup, complemented by snapshotting, to minimize the window of vulnerability to ‘roll back’ to another copy of your data. Whereas tape backup was typically a once a day incremental and weekly full, snapshots could be done several times a day, reducing the vulnerability window to several hours vs. 24 hours. Furthering the data protection evolution is continuous data protection (CDP), or sometimes known as continuous protection and recovery (CPR). With CDP/CPR, now the ‘rollback’ window can be as small as one second, almost eliminating any window of vulnerability. SDS makes this possible across any back end storage. And through Figure 4, you can see that it illustrates that via SDS, a virtual disk can be ‘rolled back’, or ‘rolled forward’ to any point, potentially even with one second granularity.

 

Figure 4

Conclusions

The conclusion is that through SDS, the warfighter can greatly increase her/his ‘data armaments,’ shielding themselves not only from the physical or cyber-attacks of another nation, but also from the financial attacks of traditional storage architectures that limit one’s ability to provide the underpinnings of a resilient storage architecture and strong force projection. The capability and time proven products exist today to field a storage environment that is one of ‘where information lives,’ vs. one of ‘where information dies.’


Watch a Demo

Interested in seeing more of what SDS has to offer? Schedule a 15-minute live demo with one of our Solutions Architects today!

Get a Live Demo

Talk with a solution advisor about how DataCore SDS can make your storage infrastructure modern, performant, and flexible.

Request Live Demo

Related Posts
 
Sander Puerto
Stretched Clusters: Shared Storage Considerations to Be Adaptable and Future-Ready
A stretched cluster, sometimes called a metro-cluster, is a deployment model in which two or more host servers are part of the same logical cluster…
 
Alfons Michels
Secondary Storage Doesn’t Have to Be Costly or Difficult to Manage
What Is Secondary Storage? Secondary storage complements the storage used for your major day-to-day operations (typically called primary storage). Primary storage is fast and highly…
 
Augie Gonzalez
Diverse Storage Systems – Putting Their Differences to Good Use
IT practitioners are taught since infancy the benefits of standardizing hardware. The “homogeneous doctrine” promises that sticking with one manufacturer and one model will make…