No… I’m not here to announce EMC’s next great product but rather how two existing products can be leveraged together to provide performance and peace of mind. In today’s demanding business environments; high performance and continuous availability are always on the top of mind. This blog post is here to tell you that you don’t need to sacrifice one for the other.
One of the most interesting EMC product synergies is XtremIO and VPLEX. This allows customers to realize the performance and local high availability benefits of XtremIO combined with VPLEX’s virtualized storage which provides highly availability for a multisite cluster. System upgrades, software patching and hardware failures can cause host-level or even site-level failures. Leveraging this powerful combination for SQL Server allows for instant failover between datacenters in all of these scenarios.
XtremIO is able to handle extremely high I/O through use of its balanced architecture, in-line data reduction, and virtually provisioned storage… eliminating many traditional storage best practices and requirements for complex and time consuming fine tuning.
VPLEX Metro is a storage virtualization appliance that sits between the hosts and the back-end arrays. It delivers application and data availability within a data center and over distance with full infrastructure utilization and zero downtime.
In a recently posted solution, we look at how to leverage these XtremIO and VPLEX together to provide a high performance solution for SQL Server where a single copy of a database can be shared and accessed in multiple locations over distance. While the solution also shows the high performance achieved by XtremIO, the test objectives were the following:
- How VPLEX Metro, XtremIO and SQL Server cluster provide an efficient solution
- Demonstrate the minimal impact of any storage, host or site failure
- Show consistent performance or better in the event of a failover
So, what did we test?
We simulated an active/active OLTP workload on both SQL Server 2012 and 2014 and conducted failover testing with both host (planned and unplanned) and site failure with automatic failover. For this solution a 2 X-Brick cluster was used in conjunction with VPLEX Metro.
The configuration was pushed to close to 200,000 IOPS from the SQL Server side and still maintained low latency with multiple concurrent SQL Server database workloads in a VPLEX environment.
What did we find?
- Disk I/O performance was very similar for SQL Server 2012 and 2014.
- Fast and simple setup of the environment with little to no storage tuning.
- Management and monitoring of the environment is simple and straight forward.
- XtremIO, virtualized by VPLEX, works seamlessly for virtualized SQL Server environments.
- This solution provides support for even the most demanding transactional workloads.
- All client activities are kept at the same level during the instant failover process.
- Sustained performance with minimized disruption in host or site failovers.
Remember though, for SQL Server there are multiple benefits with XtremIO beyond just performance. Yes, there is a blog post on some of these additional capabilities. I’m so glad you asked.