Advanced Snapshots Whitepaper

Transcription

1 Advanced Snapshots Whitepaper WHITEPAPER

2 Introduction: The Need for Snapshots In any storage system, the one feature that is of paramount importance to the user is data protection. While the idea that data will not be lost due to a system crash or a design bug should be a given, the administrator and users also expect that a storage system will protect them from inadvertent deletions, unwanted modifications, malicious agents, and other data modifications that were not approved. A wide spectrum of data storage solutions has been proposed by the storage industry, and implemented to various degrees. Perhaps the most common method of data protection is through scheduling data backups at regular intervals, either on the storage system itself, or to tape media. This method is virtually foolproof in terms of the data integrity it provides, but is burdened with the major disadvantages of requiring massive storage capacity, and causing substantial service disruptions during the backup period. Because server backups have the potential to disrupt operation or drain bandwidth and/or performance, they are often performed overnight or other periods of slow / low server traffic, which can require additional resources outside of normal working hours. Moreover, as a result of these disadvantages, it is typically impossible to schedule backups more frequently than about once a day, causing a potentially large window of vulnerability if a restore must be made. In response to the drawbacks involved with simple reliance on daily (or nightly) tape or drive backups, data snapshots are an industry-wide standard for protecting data with increasingly fine granularity. Whether they are captured according to a schedule or under the direction of an administrator, a data snapshot involves the storage system taking point-in-time images, or snapshots, of the contents of a volume in a storage system. In a way, each snapshot is similar to an incremental backup of the data within the system. What distinguishes snapshots is the manner in which they back up data, and their positive impact on space utilization and system performance. WHITEPAPER 2 Generally speaking, snapshots look and behave like complete backups

3 they can be mounted as volumes, read simultaneously without affecting the volume that they represent, rolled back onto the volume if required, and deleted to create free space. Some snapshot implementations also permit writable snapshots, in which snapshots may be modified to create branches of volumes. The ease and flexibility of performing these operations has been instrumental in speeding up the adoption of snapshot technology in the IT industry. As snapshot technology has evolved and grown more mature and stable, several vendors have discovered ways of exploiting snapshots to make other storage-related tasks faster and more convenient. Two prime examples of this are the concepts of Data Tiering and Remote Replication. These features are crucial to any storage system, but in a system that supports snapshots, their implementation becomes faster, simpler, and consequently, more reliable. This paper will serve to detail in a concise manner the primary ways in which snapshots can be incorporated into a storage system to enhance data security and contribute to increased system performance, flexibility. Implementing Snapshots in a Storage System To understand the value that snapshots provide, it is important to also have an understanding of how they differ from traditional methods of data backup to disk or tape media. Themajor distinction between snapshots and full-system disk backups is this: in a system that has active snapshots, data at a particular address is not backed up until it is modified. In other words, snapshots take advantage of the fact that a backup is identical to the data everywhere except where the data has changed, so that the snapshot only needs to protect data where it is being changed. WHITEPAPER 3 Let us assume that a certain portion of the disk has been overwritten with data, and a snapshot is then taken of that portion. When a subsequent write request attempts to modify data that was written before the snapshot was taken, this new data must not be allowed to overwrite the old data. There are two ways in which this old data is protected, and as such there are also two kinds of snapshots that reflect this. These two different types of snapshots, Copy-On-

4 Write and Redirect-on-Write, are explained below. COPY-ON-WRITE SNAPSHOTS In a Copy-On-Write (COW) implementation, snapshots protect the old data by copying it to a separate snapshot space when there is a new write that needs to overwrite it, as shown in the figure below: FIGURE 1: COPY-ON-WRITE SNAPSHOTS After the old data has been copied to the new location, the data in the old location is overwritten with the new data. Until the next snapshot is taken, all subsequent writes that arrive for the same location overwrite the new data, without touching the old data in the snapshot space. In copy-on-write snapshots, therefore, every first write that tries to overwrite an old one also spawns an additional read (of the old data) and a write (of the old data to the snapshot space). WHITEPAPER 4 The snapshot space may be organized in different ways depending on the implementation. Regardless of the implementation, it is necessary to maintain a table that keeps track of the logical addresses of data that have been copied to the snapshot space. Often, this table is just a hash (by logical address) of the location of old data versus that of new data. This table needs to be consulted

5 when a snapshot is mounted and read, and preferably should be stored in the system s main memory to allow for rapid reads from the snapshot. Since the quantity of old data that needs to be protected may be arbitrarily large, these tables can grow large and unwieldy if they are maintained on a per-sector level. For this reason, most implementations choose not to track data on a per-sector level; instead, a larger granularity is chosen, such as 64kB. In this way, data is always copied out in 64kB portions, regardless of the size of the new write request. Although vendors may refer to this granularity variously as the chunk size, physical extent, snap granularity, copy size, and so on, the implementation is the same. REDIRECT-ON-WRITE (ROW) SNAPSHOTS In contrast, Redirect-on-Write (ROW) snapshots use a different method of protecting data that needs to be overwritten by new writes after a snapshot has been taken. ROW snapshots preserve the old data in its old location, and instead, redirect the new write to a new location. This procedure is illustrated in the figure below. FIGURE 2: REDIRECT-ON-WRITE SNAPSHOTS WHITEPAPER 5

6 In this way, the physical address of the data is changed for the current data. All subsequent reads and writes of data for the volume are performed at the new location. Snapshot reads continue to be performed from the old location. As with Copy-on-Write snapshots, systems that implement Redirect-On-Write snapshots also need to maintain tables that contain a map of where old data and new data are stored. However, this mapping needs to be updated and referred to for volume reads in Redirect-on-Write, as opposed to merely for snapshot reads in Copy-on-Write. As mentioned earlier, it is therefore very important to maintain these tables in main memory for Redirect-on-Write snapshot systems. And in order to keep the amount of memory manageably small, the chunk size for these systems is usually more than one sector, for example 64kB. Because write requests arriving to the disk may be smaller than the chunk size, and since a chunk s worth of data always must be valid when it is written to the new location, Redirect-on-Write snapshots sometimes require a read-modifywrite cycle to keep data valid at the chunk granularity. If the chunk size is 64kB, and a new 8kB write arrives, 64kB of old data is read, the 8kB change is applied, and the updated 64kB is written to the new location. Therefore, the first write after a redirect-on-write snapshot may sometimes spawn an additional read. A COMPARISON OF COW AND ROW SNAPSHOTS While Copy-on-Write snapshots are currently encountered most frequently in the market and have a strong history, Redirect-on-Write snapshots are gaining in popularity. This increased popularity is due in part to the greatly improved performance that ROW snapshots provide in comparison to COW snapshots. The additional I/O load for first-time rewrites in ROW snapshots is, at most, one extra read; when a full chunk is written, this extra read is also avoided. WHITEPAPER 6 For COW snapshots, however, each new write necessarily requires an extra read and an extra write. Since the read takes place at the same location that the new write is targeting for writing, there is an additional translational latency associated with the I/O, which imposes a considerable performance penalty.

7 On the other hand, COW snapshots provide one particular degree of flexibility that ROW snapshots do not. Since the area in which snapshots are stored is clearly demarcated for COW snapshots, the administrator has the option of storing snapshot information on lower cost storage, since it may be considered stale, or less important in certain kinds of storage systems,. For example, it may be possible to store current data on a RAID-1 disk, and snapshot data on RAID- 5 disks because of the disparity in the data s importance. The administrator could also choose to store current data on fast SAS disks, and snapshot data on slower (but cheaper) SATA disks. This flexibility is not available with ROW snapshots, because the volume is scattered between the original location and the snapshot space. The following table compares a few of the other relative benefits of ROW and COW snapshots: Copy-on-Write Snapshots Flexibility to store volume and data separately Better handling of depleted snapshots space - volume is not affected Predictable amount of space recovered by deleting snapshots Redirect-on-Write Snapshots Superior performance, constant regardless of the number of active snapshots Superior space utilization, since uninitialized locations can be tracked and not redirected Deletions and rollbacks are nearly instantaneous TABLE 1: BENEFITS COMPARISON, ROW VS. COW SNAPSHOTS WHITEPAPER 7 Because of the different value and benefits that these two types of snapshots offer, choosing between COW and ROW snapshots may be difficult. In complex systems, an optimal solution could in fact be to designate certain volumes to use COW snapshots, and others to use ROW snapshots.

8 Operations that can be Performed on Snapshots As mentioned in the introduction of this paper, the beauty of snapshots lies in their versatility as a complete form of backup and their ability to make other storage-related tasks faster and more convenient. This section will detail some of the operations that can be performed on snapshots, so we may have a better idea of just how storage vendors have used snapshots to make data protection in their software and appliances even more robust and flexible. CREATION OF SNAPSHOTS Snapshots may be created in one of two ways. An administrator may decide to create a snapshot schedule, which automatically takes snapshots at predefined intervals of time. When there are a large number of volumes that need to be managed, this is often the only practical option. Most storage management software packages available today offer tremendous flexibility in creating schedules for snapshot creation. MOUNTING SNAPSHOT VOLUMES In order to read from a snapshot volume, it is necessary to mount a snapshot as a read-only volume. For example, if a user has accidentally deleted a file, it may be required to mount a previous snapshot and recover the file. Sometimes, it may be necessary to mount multiple snapshots simultaneously, compare the contents of a particular file, and choose the one that is closest to the point of loss to copy back to the original volume. SNAPSHOT DELETION WHITEPAPER 8 In order to recover storage capacity, administrators must routinely delete old snapshots. This is often part of a snapshot schedule, and as a working rule, the oldest snapshot is deleted prior to the creation of a new snapshot. Sometimes, administrators may choose a different strategy for snapshot deletion. For example, if a volume has not changed substantially between two earlier snapshots, but has changed dramatically between two later ones, the administrator may prefer to delete one from the first pair rather than one of

9 the second. SNAPSHOT ROLLBACK When data has become irretrievably corrupt, or has been destroyed entirely, the administrator has no choice but to perform a rollback from a snapshot onto the volume. In a rollback operation, the administrator has the ability to essentially select a snapshot from a list of those currently available and rewrite it to the target, so as to replace the entire contents of the volume. The most important criteria for rollback is that it must take as little time as possible. The term RTO, or Recovery Time Objective, is commonly used in the storage industry to describe this criteria. In business vital servers and data, the optimal RTO is one that is as near to zero as possible, since the amount of time needed for a rollback is equal to the amount of time that service is interrupted. Writable Snapshots The more advanced snapshot implementations available in the market today allow snapshot volumes to be mounted as both read-only as well as read-write volumes. When a snapshot is capable of being mounted as read-write, it is called a writable snapshot. Administrators often make use of writable snapshots in order to test out changes on a running system without affecting the active data. An administrator may wish to find out how much performance is improved if a defragmentation is performed on a particular volume, without actually executing the operation unless it yields a desired threshold benefit. Additionally, an administrator may want to find out if the application of an OS service pack causes any existing services to stop working. WHITEPAPER 9 For test situations such as these, where experiments must be performed without interruption of service, writable snapshots would seem to be the best solution. In these situations, an administrator would typically create a snapshot reserved for the purposes of the experiment, and leave the actual volume available for use. The snapshot would be mounted as a writable snapshot,

10 and the experiment would be performed as needed. After completion of the experiment, the administrator may decide to delete the writable snapshot, or conversely to discard the volume and convert the writable snapshot into the active volume contents. This example is just one way of how vendors and their end-users are take advantage of the power of snapshot technology to enhance system performance, availability, and potential. Snapshots and Replication Assisted Features Snapshots are an interesting and innovative extension to storage systems, and as we have seen from the examples above, they can be leveraged to enable various new system features and simplify more mature ones. This is certainly the case when snapshots are applied to replication, making systems more powerful, and enhancing their performance. To get an idea of the performance boost that snapshots offer, and how this benefits replication practices, it would be helpful to next look briefly at the different types of replication on the market. After that, we will look more closely at how snapshots are applied to replication to make replication even more effective. SYNCHRONOUS REPLICATION Synchronous Replication allows data to be stored on a pair (or more) of storage appliances, with one being designated the primary, typically the application server which is receiving the original I/Os, and secondary appliances to be used as duplicates of the primary. When an I/O is recorded, it is written simultaneously to both the primary and the secondary to ensure parity. The writing of the I/O to the primary is in essence blocked until the secondary is updated. In this way, volumes are protected from failure at the granularity of an I/O, since the primary and secondary sites are always perfectly synchronized with each other, as the name suggests. ASYNCHRONOUS REPLICATION WHITEPAPER 10 As can be expected, the speed of communication between the primary and

11 the secondary has a tremendous impact on system speed, since I/Os cannot be recorded until the write to the secondary is complete. Such an arrangement also requires a great deal of network traffic, and would not be recommended over long distances because of the crippling performance impact this would create for the system in. Asynchronous replication is often adopted in response to the high bandwidth requirements and distance restrictions of synchronous replication particularly in cases where primary and secondary storage are separated by considerable geographical distance. Asynchronous replication can minimize the performance impact and bandwidth requirements of synchronous replication by forgoing the need to write to the primary and secondary simultaneously. Because of the absence of a hold on recording the application write, application performance is improved, as I/Os can be recorded faster. In many cases, reduced bandwidth costs can be achieved by leveraging this data lag period. Furthermore, replication can be executed over greater distances because the time/write constraint has been eliminated. The major drawback of asynchronous replication, is that the sequencing of application I/O writes, or what is known as the write order fidelity problem, can create data consistency problems, as explained below. Volume replication is an increasingly important and sought-after feature in storage systems that protect valuable data. One of the most important criteria that govern the replication algorithms underpinning replication technology is the consistency of data at any replication point. Since applications like Oracle using a volume may write data in a certain order, failure to adhere to that order may render data unusable if the replication process is performed at an inconsistent point. This issue is known as write-order fidelity, and is a major consideration in replication systems. Snapshot-Assisted Replication in Detail WHITEPAPER 11 When snapshots are used in conjunction with replication, write-order fidelity is easy to achieve. When a snapshot is taken, an administrator can be certain that all data that arrived before the snapshot was taken is part of the snapshot, and

12 all data that arrived subsequently, is not. APPLICATION AGENTS AND SNAPSHOTS Applications like Microsoft SQL Server or Oracle that are running on a server and subject to snapshots can take advantage of agents (components of these applications, sometimes called writers ) to assist them to pause (or quiesce ) I/Os momentarily so that the snapshot service may take an accurate picture of application data. An application-aware agent can take advantage of Microsoft Volume Shadow Copy Service (VSS), which is used on the client side to initiate the snapshot as part of the OS. Working with VSS it can notify the application using the volume of the imminent snapshot, allowing it to quiesce its data in a consistent state to facilitate the snapshot. These agents (or writers ) interact with Microsoft VSS to prepare the application for an imminent snapshot: When applications and services are running, the writer responds to signals provided by the Volume Shadow Copy Service interface to allow applications to prepare and quiesce their data stores for shadow copy creation and to ensure that no writes occur on the volume while the shadow copy is being created. (During quiescence, applications make data on the disk consistent. For example, an application might flush its buffers to disk or write out in-memory data to disk.) The snapshot can now be replicated to a remote location with the assurance of maintaining write-order fidelity and application-level consistency. This type of replication does not interrupt or slow down I/O on the volume, because it reads from a snapshot volume, leaving the data volume untouched. The fact that an administrator has very fine control over snapshot creation and deletion means that it is easy for her to integrate snapshots as part of a replication scheme in a consistent and convenient manner. WHITEPAPER 12 Snapshot-assisted replication allows replication of snapshots in chronological order on a remote machine. Snapshots can be organized by applicationbased consistency groups, fin order to provide snapshot consistency across

13 multiple volumes. During failover, the server automatic rolls back to the latest consistent snapshot. A periodic/continuous replication where the secondary grows snapshot at a time with the primary, and some applying replication scheduling to support efficient use of network bandwidth when this data is replicated to remote volumes. FIGURE 3: TYPICAL SAR ARCHITECTURE SNAPSHOTS AND DISK BACKUP Disk-to-disk (D2D) backup moves backed up data to a separate disk based appliance, commonly over a network. D2D is gaining popularity because disks are faster than tape devices and therefore reduce the back up window. The back up window can also be dramatically reduced with D2D backups, and even eliminated altogether by utilizing Snapshot technologies available in StorTrends. When snapshots are employed to make data back up more efficient, backup windows that lasted hours can now be reduced to minutes or even seconds. SNAPSHOTS AND DATA TIERING WHITEPAPER 13 Data Tiering, sometimes also referred to as Information Lifecycle Management (ILM), is an increasingly important function of storage systems. Since

14 information systems are usually used continuously over several years, it becomes necessary to differentiate old data from new, and give priority to new data over old data. An administrator may want to move old data to cheaper, lower-performance media, and move frequently used new data to faster media. Snapshots have the natural advantage of having time information integrated into them, thus making them an ideal choice around which to build a data tiering implementation. Snapshots can be leveraged in data tiering implementations to timestamp data and migrate it accordingly, and some data tiering features of storage systems also use snapshots to provide granular block level tracking and management, and automatic data classification by storage type, RPM, size, RAID level, and so on. In this way we can see several examples of how snapshots are leveraged in storage systems to enhance backups and data tiering, making them more efficient and effective, increasing data continuity, and contributing to bandwidth and cost reduction. StorTrends Advanced Snapshot Technology Since its inception, StorTrends from American Megatrends has consistently featured industry-leading support for snapshots, with an exhaustive array of innovative features, and an implementation that optimizes performance and storage capacity utilization. THE POWER OF StorTrends ADVANCED SNAPSHOT TECHNOLOGY The advanced snapshot technology found in StorTrends arrays enables both Read-Only and Write-Only snapshots at the block level. This module allows for rapid creation and deletion of a snapshot with near-zero degradation, permitting fast back-ups- with the assurance of a complete and secure backup. Advanced Snapshot technology is focused on performance, enabling users to mount a snapshot as a volume, read from a snapshot simultaneously, instantaneously roll back to a snapshot and delete it. WHITEPAPER 14

15 THE ADVANTAGES OF StorTrends ADVANCED SNAPSHOT TECHNOLOGY The Advanced Snapshot technology featured in StorTrends software also has a number of unique features. Small and medium-sized businesses (SMBs) have access to and functions that were previously available only to enterprise-level users. Some of the key innovative features of StorTrends Advanced Snapshots technology include: Space and I/O efficient Redirect On Write Snapshots R/O and R/W SnapshotsMinimal impact on I/O performance (both sequential and random I/Os) Maintains original snapshots even when writable snapshots are modified Zero degradation in creating additional snapshots Multi-Level scheduling of snapshots VSS Aware Snapshots Rollback terabytes of data in a second. Delta snap reads: snapshot is taken of only changed (delta) data portion StorTrends allows all snapshots, as well as the data volume, to be mounted and accessed simultaneously. There is minimal performance degradation despite this capability. StorTrends itx aims to give administrators maximum flexibility when recovering data from snapshots. WHITEPAPER 15 Scheduling of snapshots in StorTrends itx can be done at the following intervals: 5 minutes, hour, day, week, and month. Management of snapshots is through CLI or the integrated ManageTrends web-based interface, giving an administrator control over the creation, deletion, mounting (read-only and writable) and rollback through either of these two interfaces.

16 FIGURE 4: THE ManageTrends GUI, PART OF StorTrends itx StorTrends itx also provides full support for both COW and ROW snapshots, thus providing an additional degree of flexibility to the administrator. It is possible, therefore, to choose the right implementation for the right application, taking advantage of the benefits of both COW and ROW technologies. In fact, StorTrends also supports 8,048 snapshots. While current hardware places a limit on the number of snapshots available to a system, this number scales with the amount of hardware resources available, such as processing power and memory, and will continue to increase into the future. The high performance snapshots in StorTrends are based on thin provisioning, so that the administrator does not need to reserve space for snapshot during RAID array creation. The physical space for snapshots is allocated on demand by employing thin provisioning. WHITEPAPER 16 With StorTrends itx, the time taken to create a snapshotis less than 100 milliseconds. This provides tremendous flexibility to the user, because the administrator is now free to create schedules with extremely fine granularities.. In addition, because the time window is so small, StorTrends implementation of snapshots is fully compatible with initiator-side methodologies like Microsoft Virtual Shadow Service (VSS), which allow snapshots to be fully application-aware.

17 StorTrends itx also provides the industry s most fully featured writable snapshot implementation. In this implementation, any snapshot can have an associated writable snapshot, which can be mounted read-write. The writable snapshot may be created, deleted and rolled back with exactly the same feature set as read-only snapshots. Performance of writable snapshots is as good as the performance of read-only snapshots, and minimally degrades the performance of the main volume. The StorTrends snapshot implementation allows administrators to delete any snapshot at random. Snapshot deletion is fully schedulable, and takes less than 100 milliseconds, providing the administrator with maximum flexibility in scheduling deletions. Additionally, freeing of space due to snapshot deletion is immediate, and so additional space becomes available at once when a snapshot is deleted. In order to make the transition from a server that is down to a server that is up again as smooth as possible, StorTrends also provides maximum flexibility for rollbacks. It is possible to roll back from any arbitrary snapshot, regardless of how far back in time it is, without having to roll back intermediate snapshots successively. Additionally, the time taken for a rollback is typically less than 100 milliseconds, and often substantially less. If the rollback has occurred from a ROW snapshot, the entire rollback is completed in that time, and there is no data migration. The volume is available for I/Os as soon as the rollback is signaled as completed. SNAPSHOT AGENTS WHITEPAPER 17 StorTrends itx supports the creation of SAN snapshots using Microsoft VSS snapshot technology, and through application-specific agents for application servers like Microsoft Exchange Server, SQL, and Oracle. As described in an earlier section of this paper, agents are specific modules that work only for the application servers for which they are written. The Microsoft VSS framework provides the application servers (writers), requestors (backup software) and hardware provider (StorTrends) a mechanism to talk to each other to perform a

18 backup operation through snapshots. Microsoft VSS-based snapshots can also support agent-less, LAN-free backup. SNAPSHOT-ASSISTED REPLICATION This snapshot-assisted replication technology featured in StorTrends itx allows replication of snapshots in chronological order on a remote machine. Moreover, snapshots in StorTrends itx can be organized by application-based consistency groups, so that in fail-over to a secondary appliance, StorTrends itx will automatically rollback to the latest consistent snapshot. AMI supports various means of synchronous and asynchronous replication, and provides full support for snapshot-assisted replication. As mentioned earlier, StorTrends allows an administrator very fine control of snapshot creation and deletion, making it easier for him to integrate snapshots and replication in a consistent and convenient manner. FIGURE 5: SNAPSHOT-ASSISTED REPLICATION IN StorTrends itx DATA TIERING WHITEPAPER 18 Data Tiering in StorTrends itx takes maximum advantage of its fullfeatured snapshots to timestamp data and migrate it accordingly. The Data Tiering module in StorTrends itx features granular block level tracking and

19 management, and automatic data classification by storage type, RPM, size, RAID level, etc. This permits the administrator to allocate data based on these classifications, as well as its usefulness and popularity, to either faster, more expensive storage, or storage that is less powerful, yet more affordable, for efficient storage resource allocation. PERFORMANCE OPTIMIZATION WITH ADVANCED SNAPSHOTS In the StorTrends implementation of advanced snapshots, further potential for accelerated performance is realized by giving the administrator partial flexibility to configure data chunk sizes. A well-known fact of server management is that different applications create different kinds of I/O loads. For instance, database applications typically provide random 8kB read and write loads to a server, whereas file servers often receive 64kB sequential read and write requests due to OS-side caching. To accommodate all these applications on the same platform, it is possible to tweak the chunk size to be either 8kB or 64kB. The benefit of synchronizing the chunk size and the most frequently received I/O request size is that if ROW snapshots are being used, a large number of additional I/Os (in the form of read-modify-write cycles) can be avoided. A best case scenario makes it possible to perform penalty-less ROW snapshots in this way; at this time, no other implementation has been able to improve on such phenomenal snapshot performance. Conclusion WHITEPAPER 19 Snapshots are the way forward to ensure data consistency and protection in modern storage systems. The two methods of performing snapshots, copyon-write (COW) and redirect-on-write (ROW), each of which have distinct advantages and disadvantages. An administrator expects maximum flexibility in creating, deleting, and rolling back snapshots, and expects that each operation take a minimum amount of time. The most advanced snapshot implementations allow the administrator to mount snapshots as both readonly and writable volumes, for the purpose of experimentation. In addition to all the above features, an administrator expects stability and performance from any snapshot implementation. The manner in which snapshots can be used

20 to enhance data protection, especially in the creation of replication schemes, shows what an invaluable technology they truly are. The snapshot implementation offered in the StorTrends software by American Megatrends, featured in every StorTrends IP-SAN and NAS appliance, can be considered one of the premier snapshot implementations in the industry. StorTrends provides all of the above features, at a minimum performance premium, with minimal waste of space, and with almost no performance impact or data degradation. In addition, the snapshot implementation found in StorTrends is lightweight and flexible, making it easy to add new features and fine-tune existing ones. AMI uses its snapshot implementation to enable various other attractive features, primary among them being snapshotassisted replication and data tiering. StorTrends makes sophisticated snapshot technology that was once available only to enterprise-level users accessible to those in small and medium-sized business (SMB). Why StorTrends? Consider StorTrends for your advanced snapshots solution because StorTrends from American Megatrends, Inc. is Performance Storage with Proven Value. StorTrends SAN and NAS storage appliances are installed worldwide and trusted by companies and institutions in a wide range of industries including education, energy, finance, state & local government, healthcare, manufacturing, marketing, retail, R&D and many more. StorTrends enables users with the features and tools necessary to meet the challenges and demands of today s business environments by offering key network storage functionality such as unified storage, simplified management, business continuity, disaster recovery, high efficiency and virtualization support. Since 2001, StorTrends has built a community of over 1,000 satisfied customers thanks to award-winning products, integrated data protection and world-class support, while continuing to enhance its products with patented enterpriseclass features. For more information on StorTrends solutions from AMI, visit to sales@ami.com, or call U-BUY-AMI. WHITEPAPER 20