Storage tiering cuts expensive storage and file access hassle

By Khoo Boo Leong | Jun 23, 2009

Regardless of the economic climate, the cost of managing storage and the amount of storage enterprises have to manage continue to balloon. But the reality is that data decays in value over time. Yet, much of these data is retained -- often on expensive primary storage -- due to business and compliance policies or plain neglect.

Not surprising, the cost of storing data, backing it up, restoring it if some data is lost, and replicating it to separate sites for disaster recovery has overwhelmed IT budgets.

"The economy is definitely impacting IT budgets, with 47% of Fortune 1000 organizations interviewed planning to decrease their storage spending," said Robert Stevenson, managing director of Storage Research at TheInfoPro. "At the same time, these companies are actively looking for ways to boost storage utilization, which hovers at about 50 percent.”

Tiering benefits

To free up expensive disk space and reduce storage and backup costs, many of these companies are considering file virtualization systems and storage tiering solutions.

The latter automatically moves non-critical data from expensive storage resources to lower-cost alternatives based on pre-defined policies. It also shrinks backup and recovery windows by reducing the amount of redundant data backed up on a regular basis.

In this regard, F5 Networks’ ARX file virtualization system with Data Management OS version 5.0 and Data Manager version 2.5 “allows organizations to determine the value of the data based on easily observable characteristics,” said Kirby Wadsworth, vice-president of Global Marketing at the application delivery networking vendor.

“For instance, you could say mp3 files are not valuable. We can institute a policy to put these music files in the least expensive storage. We are not backing them up or replicating them to a disaster recovery site.”

Similarly, files that have aged out over time or files that have not been accessed for a period of time can be moved to less expensive storage and deduplicated. At the same time, administrators can set policies to replicate all files or specific groups of files to a specific storage tier, the cloud or another site. They can implement policies for users, groups of users or, volume or specific drives.

"There are operational and economic benefits to storage tiering, but the challenge for many companies is to find a way to implement tiering in an easy-to-manage and non-invasive manner," said Dave Russell, Research vice-president at Gartner. "Solutions such as file virtualization help overcome these challenges."

ARX in action

The ARX essentially works like F5’s BigIP application delivery controller. It enforces a business policy of, for instance, moving music files to the cloud, as a real-time, in-line proxy.

“Every time someone stores a music file to a network-attached storage, we intercept that action and send the file to the right storage tier based on the business policy,” said Wadsworth. “This is exactly what BigIP does. The main difference is that the ADC handles network traffic while the ARX handles storage traffic.”

With F5’s ARX, the user need not know where the data is stored. It breaks network drives such as the J: or H: drives so users store their files into multiple physical tiers. “The H: network drive could be some NetApp storage, some Windows storage server and some on Data Domain,” Wadsworth said. “The end-user wouldn’t know because everything is just H: drive.

“For instance, you still see your music files on the file manager even though they have been moved to the Data Domain box. The only impact for the user is that it probably takes an extra couple of heartbeats to get them back because they have been compressed on the Data Domain box.”

A policy can be set so that if a user updates or writes to a file that has been moved to tier-2 storage, ARX instantaneously moves that back to tier-1 storage.

F5’s experience with customers shows that typically, storage tiering will move 80% of the files that a company has off tier-1 storage because either these files have not been accessed over a period of time or they are not defined by the company as critical to be kept on tier-1 storage.

“Some companies are even moving their tier-3 storage to the cloud with storage-as-a-service providers. Still, to the end-user, they all still seem like they are on the network drive,” Wadsworth said.

What-if scenarios

In addition to providing detailed information such as file age, types of files, and which resources are being consumed, Data Manager 2.5 can project potential cost savings of storage tiering policies based on actual customer data.

“Data Manager reports tell you how many files you have and how many are on the EMC or NetApp box and how many of these have not been accessed for x number of months,” Wadsworth pointed out. “You can do what-if scenarios and estimate potential savings from, for example, moving files more than 6 months old to less expensive storage.”

Going beyond control based on a file’s obvious attributes like filename extension, age, size and owner, storage tiering can potentially allow enterprises to make business decisions based on the content of the file via indexing and metatags. “Then, organizations can adhere to health and financial industry regulation enforcement,” said Wadsworth.

“We’ve done that recognizing traffic flow through the BigIP. With a packet, we can look at it and detect it is a credit card record, for instance, so if you try to access that, we will enforce access policies. On the file side, the content is so massive that we cannot do that on-the-fly without indexing.”

So, it helps that F5 Networks partners with third-party vendors like Microsoft for their enterprise search and text analytics expertise. After all, indexing itself is a resource-intensive and computer-intensive exercise.

Tier-3 live

Meanwhile, ARX has gained significance amid the tussle between EMC and NetApp to acquire Data Domain because it complements virtual tape library and deduplication solutions. “We have been successful with the ARX, convincing people to use the Data Domain box as a live tier-3 storage device, not just as a tape library,” said Wadsworth.

“A live storage array requires high performance in processing on-the-fly,” he added. “The ARX moves files that are older onto the Data Domain box. That eliminates the performance issues because the user has no expectations of high performance for files that have not been touched for six months.

“The big deal about Data Domain is that it is the number one choice for customers for deduplication, the technology that matters most to storage managers and storage architects.

“It is a simple and obvious way to reduce data growth. So, by implementing this tiered storage approach with deduplication, enterprises get quick payback and lower TCO.”