DAS or NAS? Making the right data-storage decision for deployed systems

Story

September 09, 2022

Steven Petric

Curtiss-Wright

For many years, military-system designers relied solely on direct attached storage (DAS) devices when they needed to deploy data storage on military platforms. These devices are embedded within or directly attached to the computers, such as single-board computers (SBC) on a platform. When a DAS device is used, only the specific SBC it’s attached to can access the stored data-at-rest (DAR). With Ethernet becoming commonplace on modern platforms, network attached storage (NAS) devices (also known as network file servers [NFS]) have emerged as an important alternative for deployed data storage. With NAS, stored DAR can be made available to any client devices on the Ethernet network. NAS and DAS are often used in combination to meet all program, platform, and application requirements. However, in those cases where it makes sense to use only a DAS or a NAS device, it’s useful to understand how they differ.

Due to the relative simplicity of a DAS, it can be much easier to deploy than a NAS device. Since a DAS device is installed within a computer, or attaches directly to it, broader connectivity requirements are mitigated. However, each DAS device on the platform must be individually deployed, upgraded, and maintained, which can increase total cost of ownership (TCO). A NAS device, in contrast, must be connected to the Ethernet network on a platform, which can initially increase deployment complexity and cost. While NAS devices are typically more expensive to purchase and deploy than DAS devices, they do provide more capabilities, such as centralized data storage, that can offset the time and cost required to purchase, upgrade, and maintain multiple storage devices.

Consider the use of data-storage devices on larger platforms where the stored data may be classified at different levels – top secret, secret, or sensitive but unclassified. These different data levels will need to be kept separate and access requirements will be different for each level. DAS offers the simplest approach for keeping data levels separate, because a DAS device is attached to only one SBC. If an Ethernet network is required, the simplest approach for dealing with multiple classification levels is to create physically separate networks (or enclaves) for each data level. Each enclave will require a separate NAS that services its own network with no connection between them. If information flow between enclaves is desired, a cross-domain solution is required to ensure the appropriate data flow and access standards.

A DAS device offers simplified integration compared to a NAS, since it can be installed directly onto an SBC. The DAS device requires no separate power supply, housing, or structure other than a printed circuit board and possible conduction-cooled frame. If data transport, rather than data transfer, is required, a DAS device that supports removable data cartridges such as an RMC [removable memory cartridge] card will be preferable. After the mission, the collected data can be transported back to the base station for more thorough analysis. A DAS device would not work for such an application, since it is not readily removable.

Using a NAS device can simplify software upgrades for users. Prior to each mission the current OS [operating system] and APP [application program] for each client can be loaded to the network via a removable data cartridge. If the OS or APP requires updating or upgrading, the new software versions can be distributed to all clients on the network upon booting, without having to remove the client or the NAS from the platform. An additional benefit of a NAS-based network boot approach is that the OS and APP data can be protected with encryption, such as National Security Agency (NSA)-approved CsFC two-layer encryption.

Network booting can also reduce SWaP [size, weight, and power] concerns since multiple DAS devices can be replaced with a single NAS device. That said, using a single NAS device for network booting may slightly increase total startup time for the system compared to booting with multiple, dedicated DAS devices. A local DAS device delivers the OS and APP more quickly since only one client must be serviced, while the NAS device must service all the network boot clients.

[Figure 1 | Shown: the XMC-554C 3U OpenVPX storage card, an example of a rugged DAS solution; in addition, the DTS1 Data Transport System DAS is a turnkey rugged network file server.]