QiNetix Aims to Move Mountains (of Tape)
Diogenes Analytical Laboratories Inc. estimates that nearly two-thirds of storage management efforts are directed to B/R. In terms of capital expenses, the acquisition costs of B/R software, tape libraries and tape devices are fairly obvious. The hidden budget-killer is the media itself, as those $60 to $80 tape cartridges add up quickly. Organizations that manage tens or hundreds of thousands of tapes are likely to spend more on media and media management than on everything else in their B/R budgets.
Assessing the costs of B/R
A high B/R job failure rate is typical, which is why it's a labor-intensive endeavor. Diogenes Analytical Labs estimates that nightly B/R job failure rates range from 5% to 20%. Storage administrators must perform "triage" on failed jobs each day to determine the cause of the failure, rectify the situation and determine when to re-initiate the job based on how risky it is to leave data unprotected. Moreover, tapes must be manually rotated in and out of the libraries, moved to off-site vaults and have scratch pools replenished.
Costs associated with media are more insidious. Our estimates show that most tapes are only 60% to 70% utilized, while 85% to 90% constitutes best practice. Low utilization is usually caused by two factors. First, improper grouping of jobs (i.e., jobs requiring unique media vs. those that don't) may cause overuse of unique media and yield unfilled tapes. Second, inaccurately grouping jobs with dissimilar expiration dates may result in media being retained for long periods to preserve only a fraction of the encapsulated data.
An effective B/R application should accomplish three goals:
- Reduce job failure.
- Reduce manual "triage" of failed jobs, or at least the impact of the failures.
- Improve media-use efficiency.
CommVault's QiNetix second-generation product is designed to improve B/R efficiency through reduced job failure and faster restore times. Because it's integrated with CommVault's Galaxy B/R product, it doesn't operate with other B/R products. Our tests were designed to determine how well QiNetix can solve the B/R imperatives vs. standalone B/R products.
Setting up the test
QiNetix is positioned as a foundation comprising five components:
- DataMigrator for hierarchical storage management (HSM)
- DataArchiver for B/R-related activities
- Quick Recovery for volume snapshots
- StorageManager for trending and reporting on primary storage
- QNet for chargeback and costing
Our tests focused on the reporting aspects of the product, the most useful element for improving storage operations. QiNetix is built around the CommServe console, which installs on a Windows server ideally dedicated to the product. The installation included a SQL Server database that stores configuration details. The CommServe console is similar to other storage resource management (SRM) architectures; it gathers information from the storage elements using data agents on managed servers. The data agents must be installed on any system that has connected storage to be backed up. Supported storage environments include internal storage, direct-attached storage (DAS), storage area networks (SANs) and network-attached storage (NAS).
Excellent interface: CommVault's QiNetix has an easy-to-use GUI with a Windows-like control panel that provides access to major system functions.
To conduct the tests, we engaged the Storage Networking Industry Association (SNIA) lab in Colorado Springs, CO. To exercise the product in as broad an environment as possible, we configured a sever farm that included Windows 2000 Server, Windows XP Professional, Sun Solaris 8 and SuSE Linux on an Intel platform. Storage devices included an IBM ESS F20 ("Shark"), a Network Appliance 800-series filer and 70GB of DAS. The tape environment was a Hewlett-Packard ESL9000 automated tape library with two LTO-2 tape drives. We installed the software from scratch; installation of the CommServe console, media agents and data agents went smoothly and took about two hours.
Understanding the logic of the product is paramount to using it successfully. We started by trying to set up and run jobs without studying the product, but quickly ran into problems. Our pre-release version didn't include a quick start guide, and all of the manuals are on CD. The lack of paper manuals is understandable, but a starter guide of some sort would be helpful as it's unrealistic to expect users to sit down and read the electronic manual before starting. It should be noted that CommVault plans to include such a guide with the product upon general release. However, the documentation is quite good and spending some quality time with it was beneficial.
Storage policies are the building blocks of QiNetix. Policies are created from the CommServe console and simplify the management of backup jobs. When a new system is added, the storage administrator simply "subscribes" the data agent on that system with the appropriate policy. We liked this feature because it ensures a more efficient use of media. Similar data is treated with a single policy and can therefore be scheduled, archived and expired simultaneously. Likewise, after all data is appropriately covered by a policy, the user can be certain the same data on a different system is equally protected. QiNetix generates an event warning if it encounters any problems.
Our only difficulty occurred because our configuration experienced host bus adapter problems. This resulted in stale configuration information residing in the console, which prevented us from running backup jobs. Manual editing was needed to correct the configuration, but an auto-update or discovery process would be preferable. Moreover, CommServe's error messages weren't very helpful, so we needed to contact CommVault support to resolve the issues. The support staff was knowledgeable and helpful. In light of this experience, it's important to note that CommVault recommends a site survey prior to installation, as well as professional-services assistance for new customers. We elected to have neither, and it's likely that our problems would have been resolved more quickly or avoided entirely had we done both.
The storage policy architecture also simplifies data migration and replication. Migrating older, infrequently accessed data to lower-cost storage maximizes utilization of premium disk. Moreover, this translates into reduced exposure from failed backup jobs because a copy of the data resides on a separate disk system until the data archive to tape completes successfully. This operation is essentially a disk-to-disk backup. Data migration and archiving have similar methodologies, thereby reducing the learning curve. The only difference between the two is that archive operations use tape as a target, whereas migration or replication uses SATA disk. Storage policies can be configured to automatically move data (via replication) to lower-cost disk and then archive it to the tape according to an aging policy. Although this may sound complicated, StorageManager makes the process quite simple.
We were impressed with the Java-based user interface. All commands are created and issued from the CommServe GUI. The console has a Windows-like control panel for configuring hardware, systems and other parameters (see Excellent interface). A variety of reports can be run manually or by schedule. Reports include media usage (e.g., element utilized) and backup job success/failure, giving the storage administrator a quick snapshot of both elements (see Quick stats from summary report). Unfortunately, the reports lack some of the graphing and exporting capabilities (to PowerPoint or Excel, for example) of other products. Nevertheless, vital data is available to ensure that media use is maximized and job failures are minimized. One criticism is that the terminology of the product is different from other B/R products and takes some effort to understand. But after assimilating the terminology, we found the product easy to use.
|Backup job summary report:|
Quick stats from summary report: QiNetix generates a summary report that provides a quick look at how backup operations fared.
Performance and features
Due to limitations inherent in a test environment, we couldn't scale the product to the hundreds or thousands of devices found in many large environments. However, the architecture of QiNetix follows the highly successful design of CommVault's Galaxy B/R product. That is, when a configuration becomes too cumbersome, it can be split into smaller, more manageable subgroups. Centralized management isn't sacrificed, as information from all sub-groups can be rolled up to a single master console.
QiNetix shouldn't be compared to general-purpose SRM products, nor should it be compared to B/R job-reporting tools. QiNetix actually falls somewhere between these two categories. Given the tight integration with Galaxy, IT organizations probably wouldn't consider a standalone implementation of QiNetix. QiNetix and Galaxy should be considered a B/R suite, and be compared to these types of suites.
Although QiNetix isn't as extensive as full-featured SRM products, it generates a tremendous amount of useful B/R information. For example, it generates reports about the speed and completion of backups, alerting administrators to jobs falling outside the backup window. It also provides reports about holes in the backup operation to identify vulnerabilities and risks of unprotected data. It provides this information in almost any configuration or combination of storage devices.
Rating the product
In rating the product, we focused on these aspects:
B/R is the low-hanging fruit for reducing storage costs and improving operations. Nevertheless, this "fruit" is surprisingly difficult to pick. In many cases, IT organizations don't have the tools necessary to identify and focus on the most important areas. Organizations may be well advised to take a fresh look at their B/R environment as a whole, and CommVault is worth checking out when they do.
- Storage Optimization Analytics: An Overview –Accenture