In the above server-centric backup scenarios, the user or system administrator is primarily responsible for backing up the system. As a result, the following data management inefficiencies often occur:
- Inconsistent timing of backups
- Neglecting to do backups
- Different types of backup media and software across heterogeneous platforms
- Failure to move copies of backup data to a remote location for safe keeping (backup tape is often stored adjacent to server)
- Inefficient tracking of backup copies
- Inefficient labeling of backup copies
- Mediocre media reliability
- Mediocre backup device reliability
- Lack of a comprehensive disaster recovery plan
- Low LAN bandwidth availability
Further, cost and the flexibility contribute to the decision against using a server-centric backup and recovery scheme. This is because only one server can connect to one storage device, leading to bandwidth problems and operating system bottlenecks. By moving storage to a separate intelligent network, servers can focus on their primary function - serving applications.
A new technology has emerged for server data storage that has the potential to greatly improve the backup and recovery process. This technology is called Storage Area Network (SAN). A SAN is a "pumped-up" LAN that lives on the back end of a set of servers, connected by an ultra high-speed network (called the fabric). This fabric consists of fibre channel hubs, bridges, and/or switches designed to enable many-to-many access between storage and servers.
A SAN-based storage infrastructure uses fibre channel technology, which has the potential to provide higher levels of business continuity while giving users greater peace of mind. Today, some organizations are faced with backup windows creeping into operational windows. A fibre channel "fabric" allows users can take advantage of the higher bandwidth available on SANs (up to 100 megabytes per second), translating into shorter backup windows for customers. Fibre channel and asynchronous transfer mode (ATM) also provide greater distances between servers and storage devices, enabling an open path to remote electronic vaulting. Users can now "move the data, not the tapes".
"Server-less" Backups via SAN
This year, expect several vendors to debut direct disk-to-tape "server-less" backups via SAN technology. Specialized software drivers will enable intelligent and direct disk-to-tape data transfer, without server overhead and loss of bandwidth. This proven software technology is already imbedded in mainframe-based virtual storage solutions, including StorageTek's Virtual Storage Manager (VSM) and IBM's Magstar Virtual Tape Server (VTS) via back-end ESCON channels.
Offloading the server backup workload encourages more frequent backups. Additionally, offloading backup cycles from the CPUs inspires more point-in-time backups, with frequency based on the criticality of the application. This strategy decreases response time and increases productivity in event of a failure. For example, a brokerage firm may run an hourly backup schedule, whereas a retail operation may choose backups on an eight-hour interval.
On the horizon is an improvement on server-less backups. "SnapShot" style backups are done today on RAID 6 devices where the data pointers [What is a pointer?] are captured and virtual backups occur instantaneously. Log-structured file systems allow the virtual data to be "frozen in time" and streamed out to awaiting tape devices. Copying occurs concurrently while the application continues writing new data to the same logical device.
A common storage platform simplifies backups because you can use a single backup software suite to manage all of your SAN data with an integrated backup and recovery policy. This is performed from a central location for ease of operations, simple audit trail and data security.
SANs Equal Economy of Scale and Mainframe-Class of Service
Why not leverage the savings gained from consolidating numerous small tape systems and individual backup software packages? The expenses saved can be applied to justify the more robust SAN-based solution. Take advantage of the economy of scale that a SAN holds. You can redirect these savings to high-end, high-speed, high-reliability backup systems. Reduce or eliminate data loss from less reliable, unmanaged tapes in the bargain.
It does not take long to make a case for the benefits of SAN in your business continuity plans. The first measurement to consider is "cost-of-delay." Cost-of-delay can be calculated by using factors such as the cost of individual devices plus media, including productivity loss from defective media and equipment failure, and data loss due to failure to perform regular, consistent backups.
Another cost-saving factor is the benefit of hands-free backup and recovery when using tape libraries to automate the storage and retrieval of data. This automation yields the added benefits of physical and logical security of the valuable data. Prudent business continuity strategy enlists mainframe-class serviceability, reliability and availability, as well as a higher level of customer services structure for 24 x 7 operations.
Enter High Availability and Continuous Availability with SANs
The market definition of high availability (HA) can be elusive, ranging from the ability to withstand a single drive failure (RAID) all the way to clustering/automatic failover and remote mirror/hot-sites. It boils down to the customers' data accessibility requirements - what data is needed, and how quickly. High availability is simplified with SANs. Dissimilar server platforms can share common SAN storage, therefore, a system outage doesn't mean that your storage is inaccessible. Some examples of HA applications would be email message servers, document management and check processing servers for the insurance and finance industry, or plant processing servers.
Some applications require continuous availability and non-stop operations. This means defining duplex operation centers separated by sufficient distance to protect against natural disasters and utility outages. Anywhere from one mile to over one hundred miles separation is needed. The SAN-based storage is then mirrored, with redundant fabric and paths provided by Fibre Channel bridges and switches. Servers and networks must also be mirrored. Either server can take over the full load in case of a disaster or outage. I discovered this while serving in the U.S. Air Force, too many years ago to mention. Even in those days, the Air Force employed similar continuous availability schemes in its early warning systems. This was necessary, but the costs to maintain three discrete sites with fully redundant systems were huge. These days this level of continuous operations is within reach of most companies via SAN technology.
Can this silver lining have a cloud?
SANs are an emerging technology, and not yet a panacea to solve all woes. However SANs do raise the bar of improved business continuity strategy. It also simplifies many of the tasks of business continuity planners.
What are the negative factors associated with SANs? For one thing, hot-site service providers are just beginning to embrace SAN systems and fibre channel. This issue must be resolved with your hot- site vendor prior to implementing a SAN. Careful negotiations will ensure that you have a place to go with your data in the event of a disaster.
The initial cost of replacing legacy storage with SAN-based storage must be amortized and justified. All factors should be considered in a return-on-investment (ROI) formula, even those pesky "soft" dollars like lost data and lost productivity. Any company that has experienced an outage can derive an estimate of the cost of downtime. Use your business impact analysis (BIA) to help justify the cost of SAN. Don't forget, total cost of ownership must be considered in your justification. Nevertheless, you should consider investing today in SANs to augment your business continuity strategy.
The Advantages of SANs
As we have shown, the new SAN environments enable:
- Reduction in total cost of ownership for an organization's client/server environment
- Economies of scale and granularity similar to mainframe-class storage protection and recovery
- Configuration flexibility - data can be backed up in a traditionally remote environment using servers or can be controlled via a server-less backup mode. Both of these approaches offer a hands-free environment for data protection.
- High availability (HA) and continuous RAID data protection via server clustering with disk array attachments and open architecture schemes
- Ease of business justification using BIA results for high-impact and critical server applications. BIA-based priorities can be used to configure order-of-recovery within SAN elements.
Storage area networks may not be the solution to every IT problem, but they can have an immediate and positive impact on one of the most critical elements - backup and recovery - through increased manageability, flexibility and scalability.
Fred Aylstock, CBCP, Consulting Practice Manager, StorageTek Consulting Services Mr. Aylstock works for StorageTek as a consulting practice manager in Jacksonville, Florida, where he is responsible for marketing worldwide D/R and backup/recovery consulting services.
Ken DeOre, CBCP As a former marketing practice manager for StorageTek in Atlanta, Georgia, Mr. DeOre had the responsibility for marketing and business development for worldwide business continuity consulting services.
Rob Latimer, Senior Practice Manager, StorageTek Consulting Services. Mr. Latimer works for StorageTek as a senior practice manager in San Diego, California, where he is responsible for marketing worldwide SAN storage consulting services.