By now, everybody knows that flash is the go-to technology for high performance workloads. The marketplace is being flooded with all-flash arrays from established storage companies and startups alike to capitalize on the demand. Each vendor is eager to say “Me, too!” and check off all of the features that everyone would expect from an enterprise storage solution. However, when evaluating the right solution for your SQL environment, it is important to know how effective and efficient these features have been implemented.
Consistent IOPS and low latency are the keys to predictability of performance in any SQL environment. Not all flash arrays deliver this consistency and in some cases they can perform wildly inconsistently depending on whether data services are turned on or off or if system-level garbage collection kicks on, these are not issues for the X-Brick configuration.
Performance is just the first step; flash-optimized workloads need to leverage this for the new application services for things like real-time analytics, database workload development, test instance consolidation, and in-memory cloning and provisioning at enterprise scale. This is the true goal for all-flash arrays in the simple, agile 3rd platform data center- workload acceleration, workload consolidation, and workload services.
As a scale-out design, XtremIO reaches higher performance and capacity levels by creating clusters of multiple X-Bricks. These XtremIO clusters outperform all other arrays without compromising availability, data services, or simplicity. Both the aggregate capacity and aggregate performance increase linearly with every additional X-Brick in the cluster. This is accomplished through XtremIO’s powerful combination of:
- X-Brick building blocks to scale processing and I/O with capacity
- RDMA Infiniband fabric combined with in-memory global metadata
- XIOS software that ensures inherent software balance and data protection across the cluster.
XtremIO leverages its multi-controller scale-out design and direct memory-to-memory RDMA fabric to maintain all metadata in memory. This makes it impervious to changes in workload – it doesn’t matter what LUN sizes are used, whether there are random or sequential access patterns, or if there is locality of reference or not. The performance is consistent and predictable all the time. And best of all, heavy metadata operations like inline deduplication, thin provisioning allocations and internal array copy operations are conducted entirely in memory, at instantaneous speed, without impacting I/O to the SSD portion of the array.
Whether you are leveraging 1,2 or 4 X-Bricks in your cluster, the management of your XtremIO array remains simple and straight forward. Every volume on XtremIO inherently gets the full performance of the array. You no longer have to think about how to map indexes, tables, logs, journals, temporary space, and archives to appropriate RAID levels and spindle counts in order to drive performance. Just create a volume with just a few clicks and put the entire database structure in it.
Another key to unlocking the remaining benefits of XtremIO for your SQL environment lies in the industry’s leading data reduction technology with a combination of deduplication enabled by powerful content addressing and metadata management. XtremIO calculates a unique fingerprint for every data block based on its contents and uniformly distributes data based on its fingerprint value. For duplicate fingerprints, it never writes the data, but replaces duplicate data with pointers to previously written data before it is written to disk. This content-based data placement not only delivers better performance, better flash endurance, and makes the array easier to use, it has the secondary benefit of telling the array what it needs to know to de-duplicate data inline in the data path at full speed.
With such effective single instancing in place, we can now make use of the in-memory metadata sharing to create zero-impact copies for use in test/dev or for reporting/analytics.
The majority of test/dev environments suffer from a limited number of copies due to a lack of capacity or performance. On top of this, the amount of time it takes to copy the data wastes time and causes a performance impact on the production environment. XtremIO changes the dynamic.
With XtremIO, even multiple high-performance databases can be consolidated onto a single array with plenty of performance headroom. And with XtremIO’s instantaneous zero-copy volume cloning feature, multiple copies of a database can be created for development and test purposes without consuming additional space in the array and with the full performance potential of the parent database--all without threatening the production database SLA in any way. This improves application roll-out times since every developer can get a high-performance, personal sandbox of the production database. You can even run analytics on the same storage, eliminating the need for expensive ETL exports. Re-imagine your database and analytics workloads without the old limitations.
With XtremIO’s consistent performance and data services, you can consolidate your analytics and do new things. React instantly to change by running interactive reports while OLTP is running. Adopt flexible business practices for reporting against OLTP schema. Now users can avoid extracting data in ETL operations, and instead query the live database.
XtremIO addresses much more than just performance for your SQL Server environment.
- Database consolidation- with unmatched performance, and free space-efficient cloning
- Storage capacity- with inline data reduction, thin provisioning, and the lowest data protection overhead in the industry
- OPEX- with the simplest management, deployment, and test/dev agile data management in the industry
It’s time to accelerate, simplify, consolidate, and virtualize your database and analytics workloads. It’s time to take control with XtremIO.