HP/Violin World Record Dual Socket TPC-E Benchmark HP/Violin World Record Blade Server TPC-E Benchmark

Violin flash Memory Arrays provide a simple, reliable solution for scalable high performance storage.

Data intensive applications like Databases, Business Intelligence/Analytics, and Data Warehousing often suffer from storage I/O bottlenecks that cripple performance.  Flash memory Arrays ensure the lowest latency data access, provide high bandwidth connections, and scale to hundreds of TB’s so even the largest databases can be stored in memory. Benefits include:

• Faster transaction and query performance
• Higher CPU utilization
• Reduced power, cooling and space costs by 50-80%
• Fewer DB licenses required

Top Five Places to use Flash Memory Arrays with SQL Server

1. Tempdb Space: Each instance of SQL Server builds its own tempdb database for internal and application/users to work with application data to perform sorts, data aggregations, joins, grouping, and other ddl operations (table creates, index creates etc.).  This database is often too small or single striped with one data file and one log file, which can cause latency across the board.  This database is critical to SQL Server performance where low-latency response time is critical.
2. Recoverability: One of the major operational questions for any database is how quickly can one recover a database in the event of a failure, corruption or data loss.  Native SQL Server backup, SQL LiteSpeed, Hyperbak, IBM TSM all provide the capability of backing up your database and they all provide, depending on your database recovery mode, the ability to restore to point-in-time.  The key is how quickly the database can be backed up and restored, in other words the device media the backup files are stored on, at a possible 150 microsecond latency it is hard to beat Violin flash storage.
3. Data and Log Files: There are many factors in the speed of you database and queries, not the least is your design of file groups, data files and log files.   Spreading your data out with multiple file groups, data files or log files across multiple LUNs can increase performance, but the cost of the additional disk, physical space, and power can make it an expensive proposition.  Using Violin flash appliances, not large enclosures with disk, makes the possibility of spreading your data out and improving performance, much more cost effective because it reduces your physical footprint and power requirements.
4. Database Maintenance (rebuilding indexes and statistics): Regular database maintenance of rebuilding your fragmented indexes or stale statistics is a necessity for a well performing database and is  always a juggling game of process time vs. system availability vs. overall system performance.  With low-latency, high IOPS Violin Memory appliances the process time is reduced significantly thereby increasing system uptime as well as improving overall system performance during index rebuilds.
5. High Availability: Transactional Replication, Merge Replication, Peer-to-Peer Replication, Mirroring, Log-Shipping,  GoldenGate,  all of these tools can benefit from a fast low-latency storage solution.  For example, when using Mirroring for SharePoint databases requires Mirroring to be in a ‘High Safety with Automatic failover (synchronous)’ operating mode, requiring, among other things, a two-phase commit or a confirmed write on each mirrored pair.  Since the net effect of this is doubling your process time a low-latency write solution can only benefit performance.

• Processing time for a 10 million record job was reduced from 6 hrs. to 1.5 hrs
• A record processing job was reduced from 8 hours to 10 minutes – nearly a 50x improvement
• A weekly database refresh that used to take 12 hrs. was reduced to just over 2 hrs.

Introducing Violin Memory Arrays also had a big impact on the overall technology architecture:

• 13 servers were consolidated down to 2, reducing licensing costs and lowering OPEX
• Avoided the cost of new server and SAN infrastructure; repurposed existing systems.

Now that they have significantly improved their data processing capabilities, this customer is rethinking how they run their business. Jobs that used to run on weekends or off-hours can be run during the day for greater agility. New services can be introduced that previously were not viable with the former technology. As the VP of Operations put it: “Since we are no longer I/O limited, we can now look to how we can improve our entire business model.”

Read the full case study