StoreAge remote mirroring

• A dramatic reduction in TCO • A cost-effective, robust DR solution to protect not just your “Tier 1 data” – but based the lower TCO – any data on your SAN.

• A single, multi-vendor solution that will support all servers and storage devices in your SAN.

• Ability to access live remote data for test and backup purposes.

• Advanced protection against corruption and virus infection not found in typical mirroring.

“Are you now – or are you considering – mirroring data to a remote site?” If the answer to this question is “yes”, then read on. The next few pages may change the way you view remote mirroring!

Why is remote mirroring becoming such a hot topic? The events of 9/11, the blackout of most of the East Coast and numerous new regulatory requirements have convinced corporate management that they must mirror their data to a location that is not just a few miles away, but states away. It is no longer a “nice to have”, but now it is a “must have” architecture. The challenge is finding a solution that will protect not just your tier 1 data, but all of your SAN-based data, without breaking your IT budget.

→→ StoreAge multiMirror

multiMirror brings long distance remote mirroring to the masses.

The total cost of implementing a mirroring solution using multiMirror is an order of magnitude less expensive than the mirroring solutions that currently make up 90% of the market.

This technology allows users that have not been able to afford all the costs of mirroring data over long distances to finally acquire an affordable solution that will allow the level of data protection required for critical applications or regulatory compliance requirements.

Why does it cost so much less to implement a multiMirror solution? The best way to describe the cost savings is to start by analyzing how traditional remote mirroring is done.

Array-based Mirroring

Over 90% of currently installed remote mirroring is array-based. The two key cost contributors to these solutions are that they are proprietary to the array that supports them, and that they mirror every single I/O to the remote site. Examples of these technologies are SRDF (EMC), TruCopy (HDS) and PPRC (IBM).

Cost of Proprietary Solutions - All of these solutions require that the user purchase very similar, if not identical, high-end arrays for both the primary location and the remote “target” location. In today’s heterogeneous storage world, requiring the same storage on both ends of the mirror is a costly problem.

Limitations of Proprietary Solutions:

• Locks you into a specific manufacturer for the remote storage

• Prohibits the use of much lower cost storage at the remote site, such as new SATA arrays

• Causes inefficient use of storage resources by prohibiting the use of “other” existing storage

• Requires different mirroring software for every vendor’s storage array

Cost of mirroring ALL I/O’s – The most substantial cost of this approach to mirroring is the cost of the communication lines between the two sites. When you must send every I/O to the remote site, the throughput capability of the communication lines must be equal to the write I/O rate coming into the array.This requires very large, dedicated communication pipes that can cost as much on an annual basis as the cost of the two high-end arrays themselves. Additionally, the array at the target site must have a high throughput rate to be able to handle all the I/O’s transferred in a timely manner. If the I/O flow to the target site is interrupted in any way (e.g., communication failure, target array failure) the integrity of the mirror is in jeopardy. For this reason, many companies that employ this approach are forced to lease redundant communication lines, and use large cache buffers on the source array to store I/O’s that cannot be immediately transmitted.

Limitations of Mirroring all I/Os:

• Huge communication cost between sites

• Array of large I/O throughput required at the remote site

• Many users deploy redundant communication capabilities, doubling cost

• Large cache buffers required at source site

→→ Cost Effective multiMirror

The fundamental differences between multiMirror and array-based mirroring are 1) multiMirror is a network software solution (outside the proprietary array) and 2) multiMirror uses snapshot enhanced mirroring instead of requiring every I/O be sent to the remote site. These factors create huge cost saving opportunities.

How It Works – multiMirror uses snapshot technology to reduce mirror traffic and allow lower cost communication lines. multiMirror starts by taking a snapshot of the volume to be mirrored. This process freezes the volume and creates a small temporary volume where all new writes are directed. In the background, multiMirror transfers the frozen data to the remote site. When that process is complete, another snapshot is taken. This freezes the temporary volume that was collecting the write data and creates a new temporary volume, which becomes the new target for write data. The original temporary volume contains all the changes between the two snapshots. This data is now transferred to the remote site. This process is repeated on an ongoing basis. As new snapshots are created, change volumes from previous snapshots are applied back to the original volume, assuring the user of time consistent, updated copies at the remote site.

Heterogeneous Storage – As a network application, multiMirror is an array-independent software solution that supports all block-level storage devices in a SAN. This allows users to separate their software application from proprietary arrays, purchasing one mirror application for all their arrays. The target array can be from a different manufacturer than that of the source site array. Users now have complete independence when purchasing new storage and have the ability to re-deploy existing storage to support a remote mirroring application (or other StoreAge applications).

Snapshot Enhanced – Rather than send every I/O to the remote site, multiMirror creates a series of point-in-time snapshots and will only transmit the changes between snapshots to the remote site. This approach can substantially reduce the amount of data traffic between sites. If a block of data changes 1000 times between these snapshots, multiMirror will only need to send the latest block, saving the transmission of 999 blocks (in this example). The other advantage of sending only the latest changes between snapshots is that it eliminates the need to queue accumulated I/O’s before they can be transmitted. New I/O’s simply update the data while the previously changed blocks are being transmitted. This means users can use much lower bandwidth communication lines (less data, no I/O queuing) and can use a remote array with much less throughput capability. Actual bandwidth savings vary according to application, but can often result in a reduction of network traffic by a factor of TEN TIMES OR MORE over traditional approaches.

Additionally, snapshot-based multiMirror has a built-in, fully-integrated data protection engine — in that data corruption that normally would corrupt a target device in a typical mirror, can be recovered from in the StoreAge solution, by simply “rolling-back” to a prior point-in-time snapshot. Finally, because we transfer snapshot “deltas”, rather than individual I/Os, we do not run into data consistency problems that typically impact traditional mirroring architectures. The multiMirror target is always consistent.

Cost Saving Opportunities

• Heterogeneous storage capability allows users to deploy any manufacturers storage device

• Target site can utilize low-end, low-cost storage

• multiMirror can work effectively with low bandwidth communication lines

• User can use existing IP communication lines for transmitting mirror data

• StoreAge SVM incorporates built-in FC to IP protocol conversion, eliminating the need to purchase additional hardware when mirroring over IP

Additional Advantages of multiMirror

→→ No Need for Redundant Communication Lines – multiMirror allows the mirroring process to be paused. If a communication link goes down, the mirroring process is simply paused, updates continue at the source site, and when communication is restored the mirror continues where it left off. No data caching is required at the local site.

→→ Save Snapshots at Remote Site – multiMirror allows users to keep up to four of the snapshots created by multiMirror. This allows the user to actually use the data at the remote site by mounting these snapshots. Traditional solutions require the user to break the mirror before the data is useable.

→→ Live DR Test - Ability to test your DR capability while multiMirror is running. Unlike other products there is no need to stop the mirroring process to validate the remote site data.

→→ Live Backup at Remote Site – Users may mount one of the saved snapshots to a backup server and perform a backup to tape at the remote site. This offloads the backup task from production servers at the main data center and eliminates the need to physically transport tape cartridges to the remote site.

→→ Greater Recoverability – Multiple recoverable images of data are created and held both locally and remotely. Many disaster scenarios involve corruption of data before complete failure. In the case of array-based solutions, this corruption is immediately passed to the remote location, rendering the recovery data corrupt and useless. multiMirror uniquely prevents this problem by providing multiple recoverable images at both the local and remote sites.

→→ Live Testing of Mirrored Data – Many remote sites are also test sites. With multiMirror snapshots at the remote site, users can mount this data to test servers, for testing of patches, new versions of existing applications or testing new applications with live data.

_____________________________________________________________________________________________________

Home How to reach us! Request a Quote or Information Site Map News Forms Policies Search

Please contact us at 800-663-5523 or click here for Pricing, Inventory or to Request additional information on any product .