By Larry O'Connor, founder/CEO, Other World Computing
The other day I watched my son go to iTunes, stream a video to his computer and thought how much more powerful our systems are today than 10 or so years ago.
Back then I recalled going to a computer store where a friend made a major upgrade to his system. He bought a large, expensive MPEG-2 card that would allow him to play DVDs smoothly, flawlessly.
Today, the processors have so much power and capability they do all of that and more while barely breaking a sweat.
The same can often be true for applications such as adding high-capacity storage when reliability, availability and performance are the key selection criteria. Now you can protect the content with either hardware or software RAID (redundant array of independent disks).
RAID has become a blanket term for data storage that improves fault tolerance and input/output (I/O) performance. It has also become the best approach to improve redundancy and parallelism which results in higher performance and availability of data in the event of a drive failure.
These features, when added to the newest high-speed communications channels such as Thunderbolt 2, deliver a robust and high performance solution that protects the content and still meets the needs of today's powerful CPUs.
Everyone who has ever lost business-critical documents or big segments of content will tell you that RAID is an absolute storage must have.
RAID is a proven data storage virtualization technology that combines multiple disk drives into a single logical unit. The result is complete data redundancy and/or improved performance.
Simply stated, data is distributed across the drives based on RAID levels (0-5) selected, with the higher levels improving data protection and recovery while performance is slightly sacrificed.
That much everyone agrees.
What they don't agree on is whether software RAID can replace hardware RAID in many applications that need access to large volumes of digital data and content quickly.
All of today's operating systems have some form of software RAID built into them – OS X, Windows 7/10 and Linux. While it is relatively simple to move from older to newer versions of the operating system with the Software RAID, it doesn't provide cross-platform support so moving from one OS to another isn't possible.
So if it's already there, maybe it makes sense to take advantage of today's software RAID solutions ... or perhaps not.
But you owe it to yourself (and your data) to understand the two before making a selection.
After all, it used to be said "No one ever got fired from by choosing IBM," but times have changed.
Similarly, selecting hardware RAID "because" requires a complete understanding of your environment, your needs.
Hardware Pros, Cons
Because all of the work is done on a separate controller card, there is no impact on the server if you have an older processor or decidedly heavy workload. That duplication also makes hardware RAID more expensive than software RAID.
Hardware RAID has made significant advances in recent years including:
- The ability to hot-swap drives in the event of a failure
- A menu-driven set-up wizard so production can often begin right out of the box
- Simplified drive replacement when disks fail
More expensive than standard controllers, RAID cards are proprietary from one manufacturer to another and at times, one model to another.
Because of the lack of specification (hardware and software) transparency and standardization it is very difficult to compare competitive hardware RAID offerings. Monitoring the health and performance of the array is system dependent and usually cannot be moved from one RAID controller to another.
Each hardware RAID manufacturer has specific maintenance routines for shaping, splitting and other activities. Workarounds and deviations are extremely difficult.
In the event the controller card fails, you must locate a fully compatible RAID card and on-disk Meta data is nearly impossible to recover. This is especially problematic if all of your data sits on a RAID system that was discontinued years ago.
Software Pros, Cons
Since all OSes today have some level of Software RAID pre-installed, server-attached disks can be configured into a RAID array and reconfigured if necessary or desired.
Apple was one of the first firms to provide the software RAID kernel for external storage subsystems and today, solutions are available with other operating systems.
The RAID management toolkit is OS-specific – regardless of the computer system and RAID subsystem that is being used. This standardization is consistent, enabling you to monitor the health of all RAID units in your operation in the same manner.
With RAID cards, each is specific to the card and can't be moved from one to another.
The difference then becomes one of evaluating third-party software RAID solutions that can provide the greatest degree of flexibility, performance and ease of use/maintenance for your operating system, your applications.
All of the RAID processing is done by the system's CPU, which today is extremely powerful enabling the system to handle a large number of processor-intensive activities simultaneously.
This is not to say that RAID operations can't slow the system down; but faster CPUs make the performance lag almost imperceptible as with the improvements that minimized the need for hardware MPEG-2 processing.
Minimum software RAID overhead is carried out when you have configured the external storage as RAID 0 or RAID 1. As you increase RAID level higher (2-5), more parity calculations are carried out thus having a minor impact on performance.
Since software RAID is OS specific, you can't share the storage solution across mixed operating systems. However, multiple systems – independent of unit year or model – can share the software RAID system.
Because of the configuration flexibility that software RAID provides, some users find it daunting.
With older hardware, RAID drives must be installed in a specific order; but with software raid, you simply install them any way you want and the arrangement can be changed without losing data.
Users can even move them to individual enclosures or install one or more in your computer. The RAID configuration and data will still be intact and available.
As long as the system or enclosure supports JBOD/independent drive access, you can access your data.
On the downside for software RAID, you do have to know the tools for the specific OS or OSes your systems operate on. This includes ensuring that the storage subsystem is mirrored.
While hardware RAID producers continue to emphasize that their solutions' added cost is justified because it eliminates performance degradation, most software RAID users who have converted say they don't experience any difference between the two approaches.
While disk drives are increasingly reliable, they do fail and often at the most inopportune or worst time.
As a result, some software RAID solution providers also offer drive health monitoring and email notification if there are any issues.
With today's vastly improved, more powerful CPUs, the performance differences between hardware and software RAID options have been virtually eliminated.
Now users can determine which approach will best meet their flexibility, reliability and performance requirements today and tomorrow.
