Data storage sticker shock: the need to recalculate data storage TCO.

Today, storage administrators must select, install, configure and manage dozens of independent storage, backup, archiving, and replication products as well as management tools to adequately store and protect data. With all of these disjointed and distributed pieces, enterprises face high levels of complexity, cost, and redundancy, prompting many to implement partial protection, and even eschew functions like disaster recovery.

However, there is an approach with the potential to drastically change the storage and data protection equation. Advances in disk technology/performance, component cost reductions, and emerging technologies in clustering/grid computing and virtualization can integrate some traditional data storage applications--i.e. backup, disaster recovery and hierarchical storage management--into a distributed system environment. With this approach, the need for separate dedicated hardware/software products no longer exists.

This article compares the Total Cost of Ownership (TCO) of existing storage products against the distributed "self-protecting" storage systems that can be created as a result of technology advances. In fact, self-protecting storage systems are already being deployed and can dramatically reduce capital equipment costs by as much as 50% and provide even greater administration cost savings.

[FIGURE 1 OMITTED]

Data Storage Sticker Shock: Over-Priced, Overly Complex, Overkill 

Data storage systems today can create issues in administrative complexity, data availability and cost.

Administrative Complexity arises from the requirement to install, upgrade, and manage dozens of storage-related hardware and software products. Weekend and nightly tape backups are time consuming, with frequent backup and tape drive failures. Managing the shipment of tapes to be trucked and retrieved from offsite storage facilities is a manually tedious, continual process. Disaster recovery is still too complex for most IT departments to install and support across multiple locations. And with servers constantly consumed with data, knowing what to archive and what to retain on the server is guesswork for IT administrators without sophisticated SRM software, which adds complexity.

Data Availability becomes an issue when it takes hours to restore a lost/deleted file from an onsite backup tape. And if that tape is stored offsite, users may not get their file recovered for days. With so many applications and servers to manage, it's not uncommon for a small but significant percentage of the primary storage to inadvertently never be configured for backup, making it impossible to recover that data. Site disasters can take weeks to recover from with traditional tape-based restores. When file systems run out of space, as they often do, applications stop running until storage administrators provide additional capacity through complex, manual SAN LUN allocation, FC switch zoning, host HBA configuration, and volume and filesystem expansion. Many remote replication hardware/software products suffer from the fundamental problem of accidental local file deletions, causing the immediate deletion of all replicated copies, both on site and off site.

Capital and operational costs escalate through the purchase and management of dozens of isolated and independent hardware components, software products, and management tools. Other costs are attributable to the inadvertent over-replication of data--for every megabyte file that is created, there exists 10-15 megabytes of that file residing on multiple full backup tape sets, HSM tapes, archive tapes, and replication disks. Backup servers, backup software licenses, jukeboxes and tape drives must be continually added to the protection arsenal to ensure complete backups within the limited weekend backup window. And disaster recovery solutions can more than double the capital and operational costs of the primary site alone.

[FIGURE 2 OMITTED]

These challenges become more acute for mid-sized businesses, as well as departments within larger enterprises--where data volumes can rival larger organizations, but budget and IT resources do not. A new self-protecting storage and data protection model can be considered, particularly for file-based, Network Attached Storage (NAS), to support hundreds of business-critical applications such as email, databases, document/content management and vertical applications.

The Birth of Self-Protecting Storage

So what is a self-protecting storage (SPS) system? A self-protecting storage system comprises three major components. The first two--NAS filers and relatively inexpensive, yet powerful servers with Serial ATA (SATA) storage--are not terribly innovative. However, the third component (intelligent software that virtualizes and aggregates the collective system potential) creates the equivalent of a grid computer designed specifically for storage and data protection.

Figure 1 represents an example of how an SPS system could be physically assembled to support the full storage and data protection needs of two geographically dispersed sites--including the full recovery of all data in either site in the event of a single site failure.

Each site of an SPS system will have one or more NAS Filers and an SPS Repository made up of SATA-based Servers.

The SPS software leverages virtualization, advanced data management techniques and grid computing concepts to pool resources and enable automated data movement based on a single, user-defined policy.

By tightly integrating the software with NAS systems, and connecting them with servers/SATA storage, the SPS can seamlessly and automatically:

* Execute backups, both locally/remotely

* Compress multiple file versions for space conservation

* Back up configuration information for rapid disaster recovery

* Provide two-tiered storage with automated migration of inactive data

* Continuously check data integrity, automatically detecting/repairing corrupted files

* Self-manage/heal around system faults, and easily discover and accommodate higher performance, higher capacity components, avoiding issues of media obsolescence 

* Define and manage all functions based on one, universal policy

TCO by the Numbers

To see the impact an SPS system can have on the TCO of storage and protection, we'll first review first year capital costs and then consider the operational costs savings of an SPS system. Finally, we'll describe the projected total savings across a 5-year period.

The SPS Capital Equipment Cost Advantage (Year 1)

The following table will itemize capital equipment cost of onsite and offsite data storage system and software to support 5 terabytes of user data. This will reflect first year capital costs alone.

In the first year, the capital equipment costs of an SPS solution totals to about half of the cost of traditional data storage and data protection tools and systems.

The SPS Operational Cost Advantage

A Total Cost of Ownership analysis would be incomplete without including operational cost savings. SPS systems are designed to be easier to install, configure, manage and maintain, eliminating the day-to-day operational tasks and their associated cost and complexity.

This means no more:

* Weekend full tape backups

* Physical tape rotation, offsite storage

* Weekend/nightly backup windows

* Synthetic full/reclamation backup processes and resources

* Tape library/drive failure/service management

* Slow restore times from onsite/offsite backup and archive tapes

* Having to restore files from backups that were accidentally deleted from both locations by standard replication products

* NAS "filesystem full" alerts

* LUN allocation, SAN management volume and filesystem expansion hassles

* Tape media obsolescence

* Selecting, testing, purchasing, deploying and managing multiple independent storage, protection and management tools

* Dealing with the complexities of the building and ongoing management of an effective site disaster recovery system with multiple independent products

Robert Reeder, the CIO/CTO of W.A. Wilde, a Direct Mail and Fulfillment company, is an early-adopter of SPS systems. "I found that the amount of storage that each of my administrators can adequately manage increased by 5X with SPS systems," Reeder said.

Assuming each administrator can manage 2TB of primary storage, onsite and offsite backups, and site disaster recovery, it would take 2.5 administrators to manage 5TB of data in our TCO example. With an SPS system, only half of a single administrator's time would be required to manage the storage and protection of that same 5TB of data. At a burdened cost of $80K per storage administrator X 2.5 administrators, this is a dramatic yearly cost savings of $160K.

5 Year TCO

The 5-year TCO model (see Table) assumes 5TB of user storage and protection with a 20% yearly growth rate. First year SPS capital costs are about half the traditional storage costs and they approach one-third the cost of traditional systems after the first year.

Operational costs for SPS systems remain at 20% the cost of traditional systems. After 5 years it would only require a single administrator to manage the storage and protection of data with an SPS System. With the traditional systems implementation, it could take more than 5 administrators to manage the same amount of data.

From an ROI perspective, even if the purchase of an initial SPS system is compared with an already deployed traditional system, the return on investment will be evident in less than 18 months. An SPS system can save almost $2M over a 5-year period in capital and administrative costs over traditional systems.

Summary

SPS systems represent a significant shift in the way administrators store, protect, and manage data. SPS systems will rapidly become commonplace, because the current isolated, independent storage and protection schemes continue to miss the mark in addressing backup complexity, disaster recovery cost and uninterrupted application access to data. Many organizations are already reaping the benefits of an SPS system in their environment, including reduced cost and complexity and increased availability to data.

Dave Therrien is the founder and CTO of ExaGrid Systems Inc. (Westborough, MA)

www.exagrid.com