ATTO DriveAssure
Drive Assessment and Latency Management for Enhanced Performance

DriveAssure Guaranteed Latency:

What is Latency?

Latency is defined as the time between a request and a response. In a storage system utilizing an ATTO Technology Inc. RAID product, it is the time from when a workstation or server application issues a request for data from storage to the time that the ATTO RAID product return a response from the drives.

Latency is not throughput. Throughput, typically measured in MB/second, is concerned with how quickly data can be moved to and from the drives. Latency is concerned with how smoothly data can be transferred.

For example: If you transfer 1000MB of data in 100 seconds, the throughput is 10MB/sec. But this tells you nothing about how long it takes to read or write 1MB of data. Latency is measured by how long it takes to complete reads and writes to disk storage. Some may take serveral seconds, an eternity to many applications, while some may execute much faster.

Latency Impact

In a storage system, latency is impacted by all aspects of the network topology, from host adapters to the drives themselves. When ATTO RAID products are used, they not only minimze latency through pre-fetching, caching and early response technologies, ATTO also has proprietary algorithms to minimize the impact of drive latency.

ATTO RAID Guarantees Latency

Throughput alone is not an accurate indicator of system performance for video applications. Latency introduced via a slow drive or intermittently slow drive in a storage array if left unmanaged will lead to reduced system performance. Figure 1 is an example of how reads and writes with larger than usual delays dramatically reduce the total available stream count even through there is ample throughput.

Figure 1: Latency Response without DriveAssure. Unmanaged drive latency results in lower realtime performance.

ATTO ExpressSAS RAID products with DriveAssure technology, for parity and redundant RAID levels compensate when one drive in a RAID group does not respond within a specified period of time. The data from a slow drive can be generated using the parity or redundancy information from the rest of the RAID group, with only a nominal change in throughput.

Figure 2 shows how ATTO RAID technology efficiently manages reads and writes, forgiving intermittent drive problems while keeping data moving smoothly.

Figure 2 Latency Response with DriveAssure. ATTO RAID products have the ability to cap latency at a user specified level resulting in up to an 80% improvenment in latency response whten compared with the competition.

ATTO DriveAssure with guaranteed latency enables the fastest low latency data transfers while preventing premature drive failures and slow downs, efficiently utilizing all avaiable drive bandwidth. Enlist a good data management plan which includes ATTO RAID products.

DriveAssure Data Recovery Mode:

Your drive has failed, now what?

The capacities available in dish drives are constantly on the rise, with greater capacity comes greater risk of a second or even third drive failure during a RAID group rebuild. If this is the case most RAID products will fail the RAID group and make it offline making your data inaccessible. There are companies that can help you to recover data if this happens to you, but what if your RAID product had this recovery capability built in?

Is it really a drive failure?

There are several failure scenarios which are relatively simple for a user to fix, if they have the know-how. In the event that a RAID group is offline there are three methods that should be tried before any attempt at data recovery.:

1. Power Cycle the drives and the drive chassis
2. Check all connectors and reseat cabling
3. Verify that the host application and operating system are not causing an issue, reboot your system and attempt to access data

There are other methods to determine if a drive has actually failed, all should be explored before using any data recovery program.

Why does a RAID group go offline?

1. When other drives fail while perform a drive rebuild
2. Several drive failures on a parity protected RAID group
3. A media error on a Raid 0 group, RAID 0 is striping only and offers no parity or redundancy protection
4. A cabling or internal connection issue could cause multiple drives to fail forcing a RAID group offline
5. If a user replaces the wrong drive in the event of a drive failure. (Note: using the identify drive feature before removing a failed drive will eliminate this common RAID group failure case)

DriveAssure can recover data

1) Rebuild Recovery Mode
Forces a rebuild to continue, even when errors are present. When complete, the Raid group is placed online and all data can be accessed; however, data needs to be verified to confirm that it is intact as a drive failure may have damaged some portion of data.

2) Basic Recover Mode
Attempts access to a RAID group using any drive that was the member of the RAID group - not all data may be recovered from a failed disk. This gives you the ability to attempt to copy a portion of your data to a backup in the event of multiple drive failures.

3) Extreme Recovery Mode
Is an advanced mode which allows RAID group access using any drive that may at some time been part of the RAID group, in the event of multiple drive failures. Data may be beyond recall depending on the failure mode of the drives in question but attempts will be made to give you access to all remaining RAID group data.

All recovery modes should be used with care as the are powerful tool which is misused could cause additional problems. Default access is read only, but write access is also available so the file system utilities can be executed.

ATTO DriveAssure with Data Recovery Mode attempts to keep your data protected even when catastrophic events occur with disk drives. Enlist a good data management plan which includes ATTO RAID products to ensure your data is safe.

Video:

The ATTO Difference - DriveAssure

Learn how you can increase performance, protect your data, and save money and time with ATTO's DriveAssure technology.