MAID Storage Technology Blog

MAID built for Pesistent data

17 Jul 2008 09:27:07 am GMT -6

MAID is for Persistent Storage…not Transactional Storage: The powering of disk drives is now being evaluated in the use cases that denote transactional storage models. I was recently asked to review and comment on a study that was conducted at the Penn State University. (You can find the paper [url=http://csl.cse.psu.edu/publications/ispass03.pdf]here[/url]). The research is titled, “Interplay of Entergy and Performance for Disk Arrays Running Transaction Processing Workloads”. I am presented with many research papers on storage, processing, networks…basically anything that pertains to “digital” bits….this is just in my nature. I highlighted this research for a couple of reasons. The first reason….I have been asked about this research paper by colleagues, partners, analysts, researchers….”how does this relate to COPAN’s MAID technology”. The second reason…I have found the research to be complementary to COPAN’s deployment of the Enterprise MAID technology. The research paper is targeted towards the power management of disk drives in a transactional environment. This is the utilization of the “idle” and “standby” modes of disk drives while in RAID configuration. The goal was to determine the effective operations and the total energy consumed by disk drives under a transactional workload. The power consumption baseline was that of disk drives that were noted as “always spinning” as compared to “power managed” disk drives. The other driving characteristic is the frequency in which the data is created, updated and modified. To paraphrase the summary of the research, you can get look at the details at your leisure, the research concluded that there were no or negative power consumption savings of disk drives while under a “transactional” workload. These results are in alignment with the research that was performed for COPAN. The use case for “power managed” disks drives are the backup/recovery, replication and archive. This is known as “persistent data”. The rational behind the use case of Enterprise MAID is need for scale, reliability, cost-effective storage that is “occasionally” and “in-frequently” accessed upon any request. I am not sure who had this quote first, but I am going to use it here…” I love it when a plan comes together”....

Hybrid disk drives...part I

30 May 2008 02:41:58 pm GMT -6

This blog entry will be one of many on the topic of [url=http://en.wikipedia.org/wiki/Hybrid_drive]"hybrid" disk drives[/url]. There will inevitably be the questions about utilizing flash memory as an option in storage systems surrounding… the cost ($/GB), the reliability (AFR, MTBF), the configuration settings (partitioning), not to mention the overall performance (IOPS, MB/s). I am sure more information will become available from the disk drive vendors as they get closer to delivery of [url=http://www.seagate.com/www/en-us/products/laptops/momentus/momentus_5400_psd_hybrid/]"hybrid" disk drives[/url] (although the current information is for laptop usages it is a start). How will the flash memory be used to solve [b][i]storage issues[/i][/b]? - should be the first question to be addressed. The use of the flash memory could be used as a temporary landing zone, I/O consolidation, resting area for "hot" data, etc. We will not know the answer until all the details until the disk vendors disclose the use and configuration of the flash memory option. The first aspect of the flash memory within disk drives should be the data access models. There are 4 basic access models of which there can be various derivatives of each. [list=1][*]Sequential Writes [*]Random Writes [*]Sequential Reads [*]Random Reads[/list]In most sequential performance operations, the use of a cache can be useful for the collating of large and small write requests. The key here will be the ability to destage the data from the flash to the disk, as this will be the performance bottleneck. The sequential read model will need the ability to manage the "read-ahead" buffers. This may be the most beneficial feature to the use of the flash memory. The last model, random read, may not see any significant improvements for performance. The random read is limited by the mechanical latency within the disk drives. Stay tuned…the actual implementation of the flash memory option within a disk drive should be interesting......

What does disk reliability and MAID have in common?

21 May 2008 09:41:22 am GMT -6

What does disk reliability and MAID have in common? Last week I had the opportunity to attend the IDEMA (International Disk Drive Equipment and Materials Association) symposium regarding, “ Hard Disk and Solid State Drive Reliability – The Pursuit of Excellence“ . The symposium was represented by many of the industry “gurus” and contributors to the HDD and ODD community. I could not do the symposium justice by trying to summarize all the information in one blog entry. For those of you who are members of IDEMA, which COPAN Systems has now joined, you can get the information from the website ([url=http://www.idema.org]www.idema.org[/url] ). There were a few topics which were directly related to MAID technology beyond just the “power management” of disk drives. The first was the duty-cycle of the disk drives as it applies to the environment from which it operates. During the conference I observed the use of several new terms, such as, “Fitness”, “Usage Consideration”, and “User Experience”. Each of these terms determines a set of parameters that can now be applied to the different data storage environments. Perhaps you would agree that there are “different data needs” which require “different storage systems”? The second point that was directly related to MAID technology was that disk drive reliability was trending upwards. In a report from a system vendor, it was stated that the disk drive AFRs had decreased 50% over the period from 2003 to 2008. COPAN has been shipping the Enhanced MAID Technology since August of 2004. During this time COPAN Systems has demonstrated that disk drives that are managed for data protection and archive, but that reduce duty-cycle of the drive while constantly monitoring environmental conditions, have resulted in less Annual Failure Rates (AFR). COPAN Systems has observed an 80% (or 1/5 less failures) decrease in disk drive failures at our customer sites. This reduction in failure rate has resulted in 5x longer service life of the disk drives in COPAN Systems' Enhanced MAID enclosures. COPAN Systems utilizes our Power Managed RAID™ for the management of duty-cycles of the disks; while operating a background monitoring of disks drive thresholds with our Disk Aero™ Software. The result of COPAN Systems' software and hardware has created a disk platform that can now be utilized and implemented in data protection and archive environments. Maybe there is more to Enhanced MAID than just power savings of “turning drives on and off”…..

What do you mean by GREEN?

20 Mar 2008 12:37:58 pm GMT -6

No matter how you say Green ... [i][b]Midori ? Japanese, Vert - French, Gr?n - German or 绿色 - Chinese[/b][/i], datacenter managers are looking to save green...that would be the green associated with capital expenses and operating costs. The real [i]"Green"[/i] is found in savings associated with recouping and offsetting traditional data center costs and it's measured in...Yens, Pounds, Euros, Marks, and Dollars... Datacenter managers are looking how to conserve power consumption and plan for storage capacity growth. I travelled the equivalent of 10 times around the world last year talking to datacenter managers about their data growth and their power consumption situation. During the past year there were a few common themes regarding power consumption during my visits: [list=1][*][i][b]How can I continue at my storage growth rate and save on datacenter floor space and power consumption?[/b][/i] [*][i][b]Now that I have maxed out my power consumption in my current datacenter, how do I add storage capacity without building another datacenter?[/b][/i] [*][i][b]With next year's budget approved for another datacenter, how to plan my storage capacity growth for the next 5+ years?[/b][/i][/list] In each of the cases that listed above, a datacenter manager can use COPAN Enterprise MAID technology. The use of MAID technology can be found in all COPAN product offerings...the Revolution 300T, the Revolution 300A and the Revolution 300M. To better understand how COPAN's MAID technology can address the datacenter storage capacity growth, we begin with the understanding of the COPAN Enterprise MAID attributes. First, COPAN delivers the highest disk storage density in the industry. There are 896 disk drives in a single cabinet covering just 10 square feet of datacenter. The result is an 89.6TB per square foot with the use of 1 TB disk drives. Next is the use of the Power Managed? RAID of the disk drives in each of the COPAN products. Power Managed RAID matches the use of power (watts) with the access to the data. This feature allows the system to operate in a range of power from "standby" (2,402 watts) to "fully operational" (6,368 watts) mode. When the COPAN products are operating in "standby" mode, the power density is 2.68 Watts/TB. The "fully operational" mode renders a 7.11 Watts/TB. When combining the highest density of disk drives with Power Managed RAID , a datacenter manager now has the ability to save 6x the floor space and 4x the power utilization. As a datacenter manager having the requirement to "do more with less", the adoption of Enterprise MAID is the answer for todays and tomorrows datacenter storage growth. I do not think I will travel less in 2008, but knowing that there exists an answer for the datacenter managers , certainly has made customer visits successful....

The Advantages of "Post Processing" in Data Storage

25 Jan 2008 11:34:06 am GMT -6

To date there has been much written and spoken about the MAID technology. The notable benefits of MAID technology are power savings, disk drive package density, extended reliability, and economics. Another feature and benefit of COPAN Systems' Enterprise MAID technology is the ability to apply "post processing" to long-term, persistent data. The first objective of any MAID platform is to provide reliable access and archive of data onto a disk-based platform. Once the data has been accumulated onto the MAID platform, the data is available for "post processing". Archived data can now be "post-processed" to apply additional storage services: such as replication, migration, deduplication, and encryption. These policy-based storage services can be interleaved and scheduled based upon the particular customer's storage requirements for data retention , data access and data fault tolerance. The Enterprise MAID platform maximizes the processing, decreases the power consumption and schedules the data movement. The use of a MAID platform eliminates unnecessary movement of information within a datacenter? while providing the intelligence to replicate, compress, classify and encrypt more efficiently. Data with long-term retention policies, pose additional challenges in their requirements for replication, migration, security, compression, compaction and classification. Technologies available today which retain data in an active or deep archive utilize CD, DVD and tape technologies. What does an IT manager do if the some of all of the data needs to be deduplicated, encrypted, replicated and/or dispositioned? The inaccessibility of data (CD and tape is offsite or in a closet), does not allow for the processing of the data for a fault tolerance, security, access, and disposition. The use of a MAID platform provides the ability to recall the data for "post-processing". The scheduling of these storage services in a "post processing" manner is a value-added feature of an Enterprise MAID platform. An IT manager can now combine the often disparate tasks of data protection, encryption, distribution, de-duplication, and dispositioning - often through a single pane of glass - on the MAID platform.This greatly simplifies the challenge of data storage management. The Enterprise MAID technology implemented within the COPAN Systems' products provides an ideal "post processing" environment. An example of a storage service utilizing "post processing" is data deduplication. The combination of COPAN Systems' MAID platform with a data-deduplication "post processing" engine - creates an ideal storage subsystem - one that optimizes storage capacity for persistent data storage and provides maximum post-processing performance. The ability to schedule power and processing for data deduplication operations, greatly simplifies the management of archived data onto the Enterprise MAID platform. Since the data resides on the Enterprise MAID platform, scheduled "storage services" on the data can be processed to fall within the specified parameters for storage system power utilization, providing both the most efficient utilization and greatest performance from the storage platform. Data deduplication "post processing" is just one example of managing power and processing on a MAID platform. Utilizing this system architecture, IT managers can easily account for access, capacity management, security and fault tolerance of their information. Knowing that the data is accumulated and protected on an Enterprise MAID platform, which schedules "post processing" services, creates a storage platform for IT managers to retain long term data with confidence, ease and cost effectiveness.