Tape Media Storage is Green!
No, we are not just talking about the color of the Ultrium cartridges.
Tape media has always been an integral part of the backup, recovery and archival strategy of most companies. Storage professionals find that moving inactive data to tape can help trim power and cooling costs in the data center. Tape drives run cooler compared to disk-based systems: Tape costs $0.45/TB to power and cool versus $96/TB for SATA disks—that’s 99% less than disk as per HP estimates. Another proof point being as per a report from the Clipper Group if electrical rates remain the same, the cost to acquire, power and cool disk systems for five years is almost eight times that of automated tape systems. If electrical rates continue to rise, the potential savings will be even greater. With lower heat output, tape drives can pack more capacity per cartridge especially with LTO 4 technology. This saves space without causing cooling costs to escalate.
As enterprises continue to deploy more servers and storage devices to keep pace with their exploding data storage needs, power consumption requirements and costs are also heating up to unprecedented levels. It can cost more to cool a data center than it does to lease the floor space that houses it. One industry expert estimates that the average utility cost for a 100,000 square-foot data center can be well over $6 million per year. For some companies, it would mean they now must compensate by moving to new facilities, putting off hardware purchases, or leaving floor space unoccupied to avoid paying corresponding utility costs.
Energy impact of storage systems
It is interesting to note some of the findings from IDC’s whitepaper on the diverse and exploding universe that studied the energy impact of storage systems. Inside a typical data center, of the 46% of the energy consumed by the IT load, servers guzzle 25%. At 13%, the storage subsystem is the second culprit followed by 8% for networking equipment. [Refer Box: Energy impact on storage systems]. So clearly, power consumption by storage is significant as it contributes 13% of the data center energy load and additionally puts pressure on cooling.
Sivasankaran L, Director Storage Practice, Sun Microsystems India Pvt Ltd, said, “IT budgets are shrinking but storage capacities are doubling every year. About 80% of enterprise data is accessed infrequently and it usually occupies the most expensive storage space, which also increases the back-up window. A business can use the same chassis for FC and SATA disks or can have them separately and very old data can be pushed onto tapes and archived. Tiering of storage can help achieve this.”
What impacts power consumption?
Many factors affect the power consumption by storage systems. First, there has been a rapid increase in the storage needs an ongoing predicted six-fold rise in digital content from 2007 to 2011; from 282 exabytes of storage to 1,773 exabytes during the same period. Shailesh Agarwal, Country Manager-Business Systems, STG – IBM India/South Asia, said, “How you configure RAID in the storage subsystem has a significant impact on the energy and cooling costs.” Agarwal is referring to the fact the different RAID levels have a significant impact on the way energy is consumed by storage systems. For example, consider JBODS (where a number of discs are used to scale capacity). If x is the cost of power/cooling then the result is x multiplied by number of JBODS and cost of each JBODS. If RAID 1 were used for example for mirroring, then one would need 2 disks [2 x power costs]. So the power and cooling equation becomes 2x. The cost keeps spiraling as you keep adding as you go up RAID 5 or RAID 6 to achieve reliability and high availability of data. An individual drive uses 5-15 watts of power depending on its capacity, rotation speed, form factor and operating state. Additionally despite capacity, costs reductions exceeding Moore’s law, RAID is not free—extra disks add to capital expenditure and operating expenses for power and cooling.
Niraj Mandal, Country Manager – India, Tandberg Data (Asia) Pte Ltd., said, “A RAID array is spread across the disc system and each of the discs keeps rotating and leading to power consumption and cooling costs. In the data center it becomes critical as it operates 24x7x365 multiplying storage power and cooling costs. Additionally a disc can consume about 350 watts of power in a NAS box, whereas tape takes 17 watts. Moreover, inside the data center the storage power and cooling costs keep multiplying.”
The cost of tape vs. disk data backup
|“Replacing tape with disk in data centers would take away a lot of floor space and require dramatically increased power and cooling and thereby energy consumption. It will also cost a lot of money” - Sivasankaran L|
Director, Storage Practice, Sun Microsystems India Pvt Ltd
|“A RAID array is spread across the disc system and each of the discs keeps rotating, leading to power consumption and cooling costs. In a data center, it becomes critical as the array operates 24x7x365″ - Niraj Mandal|
Country Manager – India, Tandberg Data (Asia) Pte Ltd
|“The capacity of tape drives doubled in each generation of LTO. From 2002 to 2008, we have seen an 8x increase in throughput and capacity in just six years” - Shailesh Agarwal|
Country Manager-Business Systems,
STG – IBM India/South Asia
According to Agarwal, “We are still seeing inactive data sitting inside a data center on networked storage and it is growing quickly whereas it should have been moved onto a tape library or drive. Since this inactive data is on rotating disk, it guzzles lot of power and generates enormous amount of heat, which pushes up the cost of storage. It’s a universal truth that has come out in the open that since tapes does not have any constantly rotating component unlike disk, it saves a lot of power and cooling costs and there is no risk of disk failure. Tape continues to be the most energy-efficient storage technology for long-term data retention–requiring neither power nor cooling–while maintaining data integrity for a long time.”
Mandal added, “If a business wants 10 TB SAN storage they may have to cough up $2160.39 per TB. Cost of data on tape especially on LTO4 would $64.81 per TB. Additionally the running cost of tape would be much lower than disc.”
Improvements in tape technology
The products in question are tape drives based on Linear Tape Open (LTO) 4 standard. In 2002 when LTO 1-based tape drives were launched it could hold a maximum of 2 GB of data. Over the period, the capacity kept pace with standards and today a LTO 4 drive can give a storage capacity of 800 GB per cartridge. Agarwal said, “The capacity of tape drives doubled in each generation of LTO tapes. From 2002 to 2008, we have seen an 8x increase in throughput and capacity in just six years. Additionally you can pack 24 TB of backed up data in a single 4U space (in the case of auto loaders)—that can save costly data center floor space.” For example, LTO-4 Ultrium tapes hold an astonishing 1.6 TB on a single cartridge, with built-in encryption and Write Once Read Many (WORM) options. It can transfers data to tape at 240 MB/second, which means that it only takes 4.2 seconds to transfer one GB of data. With tape automation that offers faster backup and retrieval, tape storage has kept pace with disk storage technology. When it comes to disks they are not portable when compared to tapes and the cost of the former is much higher than that of the latter. Today LTO-4 tapes coming in with around 1 TB of data, which can be stored on a tape [cartridge] and that too at almost half the price when compared to the disk.
Additionally in 2008 we saw more intelligent tape based solutions and tape drives shipping with advanced features such as lower bit-error rates, higher-bit densities, greater capacities and native fiber channel interfaces. Due to the higher archival requirements by Indian enterprises tape automation demand has shot up as there has been a need for faster data retrieval.
Sivasankaran continued many claimed that disk drives are less expensive than tape systems but the truth is tape continues to cost significantly less than disk. The cost of disk systems varies widely. Dual controllers are more expensive than a single controller is. Advanced features, such as management software, can add to the cost. Tape technology is environment friendly where as disk drives continue to spin and need electricity to power and cool the devices whether the device is accessed or not. Tape drives, on the other hand, use little power when not reading or writing tape cartridges. Tape cartridges require no power at all, when residing in an automated library. However, for most data centers the cost to power and cool devices is a growing concern. If electrical costs continue to rise, then storing older versions of backups or archival data on tape can help to keep energy costs in line. The cost to power and cool devices cannot be ignored. Electrical costs must be part of the purchasing costs and tape is up to 25 times less costly to power than disk. He added, “Data centers are continuing to buy tape as a critical component of the storage hierarchy. Replacing tapes with disk in data centers would take away lot of floor space—require dramatically increased power and cooling energy consumption and will cost a lot of money. It is very expensive to replace tape with disk in medium-to-large data centers therefore data centers will continue to use tape to store data.”
Deduplication and virtualization
Data deduplication another way that can reduce storage needs by eliminating redundant data. Only one unique instance of the data is actually retained on storage media, such as disk or tape. Redundant data is replaced with a pointer to the unique data copy. Many of the IBM customers are asking for deduplication solution so that they can free more storage. Agarwal said, “Using deduplication, businesses can increase the capacity to store backup data by 1:20 and it that can help them further in reducing the backup window and protect the investment they have made in tape.”
Businesses are not using tape just because it is green. They are using it because of more strategic importance as in the case of Syndicate Bank or TVS Motors as saving on power and cooling costs of tape incurred just happened. Mandal added that there are many customers who have invested in LTO4 tape library [for data center] are Skoda Motors, BSNL, ARD, TIFR and SBI amongst a few. While the objective of their investment in LTO4 was to reduce the storage costs, power and cooling cost savings has become an additional rider.”
“Enterprise and mid-sized organizations are faced with data centers that are reaching a breaking point of complexity and manageability, while at the same time experiencing explosive demands for storage for new data, transactions, e-mail and back-up files. To address this challenge we have ProtecTIER de-duplication gateway offers in-line data de-duplication that is integrated with server and storage infrastructure to help organizations significantly reduce the amount and cost of physical storage required in data centers,” said Agarwal.
Businesses using tape library or tape drives are not new as they have been using it as an important strategic business tool to reduce storage complexity and associated overheads in managing it. Now they have an additional rider of earning carbon credit by going green as they can save significant power and cooling costs by using tape technologies.