The rumored closing of COPAN Systems in early December 2009 raised a few alarm bells around the storage industry. However it was not COPAN Systems’ demise that was the main cause of concern as its impending doom has been rumored for some time. Rather it is question of whether of not the MAID technology that COPAN largely succeeded in making its signature feature will die along with COPAN. After all, if COPAN cannot make a go of MAID technology in this übergreen corporate environment, who can and under what circumstances?
It is still difficult to find a definitive formal statement that confirms COPAN’s demise. However its closure has been authoritatively blogged about by Bill Mottram, a managing partner at Veridictus Associates and calls that DCIG put into COPAN during business hours have gone directly to a voice mailbox with no return phone calls. Further, a look at the LinkedIn profiles of COPAN employees shows them listing “December 2009” as their date of last employment.
Still it is somewhat unexpected when COPAN, with its “right place, right time” MAID (Massive Array of Idle Disks) technology, declares it quits. After all, organizations are more socially responsible than ever, want an economical means to keep continually growing amounts of archival and backup data on disk and then deduplicate it. Plus going forward organizations will face even more pressure to more efficiently and effectively manage storage in their data centers going forward.
In December 2006, Congress passed Public Law 109-431 that instructed the Environmental Protection Agency (EPA) to prepare and report on the use of energy in data centers now and into the future. The resulting EPA report found that 50% of all data centers will be unable to buy any additional power by 2012 because of the lack of new available power.
So one would think that this combination of internal and external forces would push organizations to adopt MAID technology sooner rather than later. Yet here I am writing about COPAN’s demise. So what gives with COPAN? Two plausible theories have surfaced.
First, in the Mottram blog, he postulated that COPAN was strong in its implementation of MAID but failed to grow its ecosystem of application partners that could take advantage of MAID. Mottram says, “This made it difficult for end-users to understand how to apply MAID technology in such as way as to help them meet their data center challenges.”
Second, despite COPAN’s strong implementation of MAID, one of its little known idiosyncrasies was that it only allowed for 25% of the drives in its array to be powered on at any given time. While on the surface this sounds acceptable, this percentage was too low and resulted in unacceptable application wait times as applications waited for powered down drives to spin up.
COPAN for the most part used a first generation iteration of MAID that met the high capacity and low cost definition of Tier 3 data storage solution. In this form, COPAN was more than adequate for long term monthly or yearly full backups, long term archives or where data is known to only require infrequent access.
However, this implementation of MAID is not reflective of how most corporate applications operated. As a result, it was only suited for a small set of applications that could deal with COPAN’s “power-up” latency. This left COPAN in a predicament where it worked well in only a very few customer environments which contributed to its collapse.
So does COPAN’s failure mean either the end of MAID or negate its value proposition? It does not. But the lesson one needs to take away from COPAN’s demise is this: “Saving energy is vital but most organizations cannot sacrifice application performance to achieve this ideal.”
Clearly what needed to occur in COPAN’s case is that it needed to update its implementation of MAID on its disk libraries in order to meet customer demands. In COPAN’s case, it did not so it is gone. However MAID as a technology has evolved and matured such that second generation implementations of MAID in the form of MAID 2.0 now exist.
A prime example of this is the AutoMAID technology available from Nexsan Technologies. In its MAID 2.0 implementation, AutoMAID allows 100% of the drives in a Nexsan storage system to operate at full power. This eliminates the inordinate response time delays caused by the “power up” design inherent in MAID 1.0 implementations.
MAID 2.0 implementations also provide more levels of energy savings than just the “On” or “Off” power settings found in MAID 1.0 implementations. In the case of Nexsan’s AutoMAID, it offers three settings:
- Heads are unloaded which results in a 15 to 20% savings in energy while still providing sub-second recovery time for the 1st I/O and full speed for all subsequent I/Os.
- Heads are unloaded and drives slow to 4000 RPMs which resulting in 35 to 45% savings in energy with 15 second recovery time for the 1st I/O and full speed for all subsequent I/Os.
- The disk stops spinning which results in a 60 – 70% savings in energy and 30 – 45 second recovery time for the 1st I/O and full speed for all subsequent I/Os.
Maybe most important to organizations, they can set policies within AutoMAID on the Nexsan systems for specific disks so that disks do not spin down unexpectedly for specific applications. However because of the relatively rapid response times that these different settings within AutoMAID provide and its ability to control when disks spin down and back up, organizations can use it for nearly any application, not just archive and backup.
In fact, about the only instance where organizations may not want to turn on one of the three levels of AutoMAID’s energy saving features is a high performance application that runs 24 hours a day, 7 days a week which is usually only a small percentage (10% or less) of an organization’s total number of applications.
COPAN is rumored to be gone but that does not mean organizations should write-off MAID technology as well. MAID 2.0 is better architected to meet the needs of today’s application and business requirements without presenting the same application performance risks that MAID 1.0 presented. By providing policies and differing levels of energy savings such as what Nexsan’s AutoMAID provides, organizations can more confidently deploy it even for production applications as it can satisfy those performance needs while still enabling organizations to meet today’s and tomorrow’s concerns about social responsibility and power conservation.