Printer Friendly

Darwin meets data storage: televaulting provisions storage as a utility.

In the first edition of "The Origin of the Species", Charles Darwin never used the term "evolution". Instead, Darwin focused on the concept of "divergence" wherein new species with specialized characteristics emerged from the same main source; they didn't evolve from a secondary species. To illustrate this concept, Darwin used the Tree of Life with multiple branches, and the tree representing the core species or product. When divergence occurred, the tree (not an individual branch) sprouts a new branch--creating a distinct, new species instead of an offshoot or evolution of an existing branch.



The principle of divergence is well established in the computer industry: printer technology diverged into laser, inkjet and workgroup "species," while personal computing technology diverged into desktop, laptop and handheld branches. In fact, data storage has its own tree of life (Figure 1), which illustrates the divergence of the storage industry, including the divergence of data protection software which has added distinct branches for tape backup, mirroring, replication, D2D and now televaulting technologies.

Distributed backup is a unique, diverged arena from the familiar data center-focused backup, with its own unique set of pain points challenging IT administrators. The fundamental problem for administrators is how to effectively manage the data protection of remote data islands scattered around the country or around the world. To date, the common solution has been to try to make traditional tape-based backup solutions work in these dispersed data islands, but the results are unacceptable.

Many remote sites lack dedicated IT staff to manage and enforce good backup practices and policies. Even if backups are run regularly, they often ignore data stored on individual PCs and laptops and focus only on server-based storage, leaving a large volume of critical corporate data at risk. In fact, one of the most critical backup pain points can more accurately be characterized as a data restoration pain point: offsite vaulting of backup tapes. This is a multi-faceted issue that begins with the problems associated with the insecurity of transporting the tapes to a remote vaulting site and proper inventory management of those tapes. Even on the rare occasions when this process is managed properly, it still leaves enterprises with a major problem--unacceptable long time-to-recovery windows. Having critical data unavailable for hours or days translates into direct negative economic impact on business operations, including customer service, end-user productivity and transaction processing.

Software licenses and agents pose another problem for backing up remote sites. Whenever a new PC is added to the network, a new (and typically costly) backup agent must be installed, and any backup software upgrade is a complex process. This software license model can pose a major expense to an organization, and if the backups are rarely performed at remote sites, the license fees are totally wasted. Finally, the old model of tape-based backup performed within a specific backup window is unrealistic in the current 24X7 global environment. A better solution is needed, which explains the technology divergence into distributed televaulting.

Televaulting is a new backup paradigm architected specifically for geographically dispersed environments. Moreover, televaulting is based on a utility service provisioning model to deliver distributed backup/restore with centralized management and a unique pricing model that enables the IT organization to transform backup from a cost center to a profit center. The prime benefit of televaulting is that it enables enterprises to operate in large, complex geographically dispersed network environments to effectively and affordably manage and protect data offsite from one centralized location. Televaulting also allows enterprises to assign actual backup/restore costs to appropriate users within the organization. In essence, the corporate IT group operates as a backup "service provider" to remote data islands--such as departments, branches, and satellite offices within the organization who become customers and are billed according to consumption. The licensing model for televaulting software is unprecedented for the backup market: the data-collection software is free. Users pay a one-time perpetual license fee based only on the actual compressed capacity of data that is centrally stored aggregately across all sites.

The basis of the enterprise televaulting data protection platform is an agentless software architecture that eliminates the cost and complexity of loading a software agent on every target server and desktop that needs to be protected. Instead, the enterprise simply requires two software components: DS-System software loaded on a standard server located in the data center, and one or more DS-Clients based anywhere in the world and connected via an IP WAN. No additional client software is required for the individual target machines that are backed up at the remote site. As shown in Figure 2, the DS-Client runs on an existing Windows or Linux server and collects data to be protected. The DS-System runs on a Windows, Solaris or Linux host used for online storage, management, monitoring, and billing processing.

The DS-Client software integrates with NT domains, Trusts and Novell NDS trees, and adopts the LAN's existing security settings, ensuring that all data residing on any systems connected to the network is fully protected. The televaulting backup model also addresses the critical data stored on laptops of mobile users with the DS Client Lite software.

Like any utility, the innovative televaulting licensing model is based on actual usage, in this case the amount of compressed data that is backed up from the remote DS-Client to the DS-System. To perform efficiently on high-latency network connections, the DS Client performs comprehensive file analysis, including byte-level delta processing and common file elimination. The residue backup data is then further compressed and encrypted prior to being sent offsite to the DS System. The net effect of the compression, delta processing, and common file elimination processes is approximately a four-fold reduction in the data transmission and storage, and effectively a similar cost reduction for the enterprise.

Utilizing televaulting to protect geographically distributed data has no impact on the data center backup operations. Once the backup data is transmitted to the DS System in the data center, it can be managed just like data from any other application server. The DS System integrates seamlessly with existing enterprise backup software and processes. Additionally, it supports SNMP integration with enterprise network management systems such as HP Open View and BMC Patrol.

Centralized Management

Perhaps the most serious issue with protecting remote data is assuring the frequency and quality of backup operations. In many cases, remote offices lack dedicated IT staff and good backup practices are rare or non-existent. Televaulting resolves this problem by giving IT administrators centralized tools to schedule, monitor, manage and perform remote restores for all DS Clients. The televaulting platform is based on a Backup Service Provider (BSP) model, enabling distributed backup services to be provisioned as a utility to maximize the efficiency of backup operations while minimizing the complexity and costs. The management tools include options that give administrators the ability to perform predictive analysis of a data set prior to backup operations, monitor SLA performance and control/enforce backup budget allocations/quotas.

To meet the requirements for true BSP quality of service, the televaulting platform incorporates two key functions: a comprehensive billing system and a self-healing tool.

The billing system is an essential televaulting tool that allows IT departments to easily charge back their backup services to departments or other corporate entities while enforcing specific budget and capacity quotas for each individual corporate customer. The televaulting billing system processes information collected by the DS-Client in order to bill the customer on a monthly usage basis, just like any other utility provider. Televaulting billing data is easily exported to existing third-party billing formats.

A core function of the billing system is integrated budgeting, allowing customers to allocate a specified backup budget based on storage capacity or financial limits. This budgeting feature helps to keeps costs in line by notifying the customer's management if and when budgets have been surpassed. Experience has also shown that end users are much more judicious about their personal data storage practices when they understand that their own department is being directly billed for their usage.

The Self-Healing tool is a transparent televaulting utility that enables an IT operation to ensure that it is meeting its QoS commitment to its corporate backup "customers." This tool runs seamlessly in the background, constantly scanning the DS-System for anomalies. It immediately notifies the administrator (and adds a record in the event log) when it encounters a corrupted or otherwise problematic file. The Self-Healing tool then automatically isolates the potentially corrupted file to a pre-determined quarantined location where it can be safely deleted. To complete the self-healing cycle, a notification trigger is sent to the DS-Client to resend a latest generation copy of the file on its next scheduled backup.

The unique Backup Lifecycle Management tool automatically migrates DS System's data from disk storage to tape-based storage. The policy-based tools keep all "critical" data on disk and only migrate data categorized as "important" to tape. The unique feature of the BLM tool is that it allows end users to set the policies controlling the migration of data from disk to tape, based on the understanding that end users are the best people to know the needs for their every day file access. The BLM tool is technology and file system agnostic and integrates with HSM file systems, such as Sun's SAMFS, ADIC's StorNext, Legato's DiskXtender, and other third-party ILM solutions.


Televaulting also addresses two of the major restrictions of conventional remote tape backup solutions: performance and capacity scalability. The DS System servers are based on an N+1 grid computing model. When the processing demands of a single DS-System server become too much, additional low-cost servers can be added to create a scalable grid. All backup processes are still managed as if there was only a single virtual DS System, but performance is scaled and load balanced across all of the available processors in the grid.

Another potential backup bottleneck is the file count limitations on a single mount point with any file system. With the DS System collecting backup data from multiple DS Clients around the world, it does not take long before the mount point barrier is reached, putting an artificial cap on the available capacity. The DS System eliminates this restriction with a proprietary automatic Multiple Mount Point capability that enables access to virtually unlimited capacity, with support for 2 billion directories in a single volume.


Televaulting is the first solution that effectively addresses the cost, performance, scalability, and management of data protection issues inherent in distributed backup. The innovative (service-oriented architecture) of Asigra Televaulting for Enterprises is the key to the distributed backup advantages of the televaulting software model. This new approach to distributed backup is the first technology that provisions backup as a utility, giving customers the advantages of a "pay as you go" application rather the cost penalties of the traditional licensing and agents model. By centralizing the administration of backups of dispersed data islands, Asigra Televaulting enables further reductions of the operational costs associated with distributed backup, and improves remote end-users' restore experience.

The backup market is poised to rapidly diverge away from the outdated tape-based data center model designed for single-sites. As part of this divergence, it makes good sense to deploy a correct backup strategy designed for multisites, as well as a correct pricing model. Distributed televaulting delivers the cost, performance and scalability benefits of a next-generation distributed backup application. In fact, deploying backup services as a utility is the ultimate advantage for IT administrators who now have the opportunity to turn what was formerly a complex, costly application to deliver into an IT profit center. The televaulting example of provisioning backup as a utility now serves as a model that IT managers can point to for other storage management applications, with the long-term objective of provisioning all storage resources as utilities.

RELATED ARTICLE: RAStor 4000 Intelligent Storage Array

In an effort to tap the entry-level storage market, Rasilient has introduced the RAStor 4000, a rack-mount RAID array in a 3U rack-mount form factor with the ability to scale to 3.75TB of raw storage capacity using 15 dual-ported SATA hard disk modules. The system offers both DAS and SAN functionality and provides iSCSI or FC host connectivity. Every component is hot swappable with no interruption in service or loss of data, adhering to the principle that the failure of any system component must never allow a system-wide failure and loss of data availability. Other features include a 72-hour battery-backed cache protection, full redundancy with remote serviceability for controllers, power supplies, fans and hard drives and passive mid-plane design to thwart system failure. The system supports 1.5 and 3.0 Gb/s SATA drives and offers multiple RAID configurations (RAID 0, 1, 5, and 10).

Installation and management of the array is simplified using the browser-based configuration and interface, but operating system support is limited to Windows 2003 Server and Linux.

Rasilient's RAStor 4-000 should adequately answer the needs of evolving data centers and provide OEMs and ISVs a powerful and flexible building block for storage applications.

Key Features:

* Zero Downtime Architecture

* Passive Mid-plane

* Dedicated SATA Channels

* Disk Roaming

* Active-Active Controllers & Transparent Fail-Over

* On-Line Capacity Expansion

* Robust Enclosure

* Mirrored Cache

* Multi-Path IO and Cache Coherency

* On-Line Firmware Update

* "Hardware" In-Box Clustering

* Software RAID

Eran Farajun is executive vice president at Asigra, Inc. (Toronto, Ontario, Canada)
COPYRIGHT 2004 West World Productions, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Storage Networking; Charles Darwin
Author:Farajun, Eran
Publication:Computer Technology Review
Geographic Code:1USA
Date:Oct 1, 2004
Previous Article:E-mail archiving: a key piece to a compliance strategy.
Next Article:The future of hard drives: diversification.

Related Articles
Celebrating Darwin Day. (International Humanism).
Charles Darwin: The Power of Place.
The Northern Territory's first Long Service Medal.
Storage in utility computing: 7 critical questions for IT.
Utility computing: slowly but surely ... it's coming.
Making the move to utility storage: overcoming the crisis of complexity.
Looking back.
USAF looks beyond backup to data recovery.
Inside the Beagle with Charles Darwin.
The state of utility computing: on-demand computing today.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters