The most common question that we are asked by IT planners considering an all-flash array is “can the data center really become all-flash?”. The answer is yes but that requires that the all-flash array achieve data center scale. This means it needs to meet the performance demands of a variety of workloads while meeting the overall capacity and availability needs of the enterprise. At EMC World 2015, XtremIO announced the 4.0 release of its software and hardware to meet that data center scale.
At the heart of any XtremIO system is the X-Brick, the nodes that make up the XtremIO scale-out architecture. A key part of the 4.0 announcement is a 40TB brick that, when combined with XtremIO data efficiency features, should deliver about 180TBs per brick. That means that a fully built-out, 8 node XtremIO cluster should deliver a couple of PBs of capacity and over 1 million mixed read/write performance. This new X-Brick also provides 40 CPU cores to support consistent performance across that cluster, regardless of workload.
A Safe Basket For All Your Eggs
There has always been this nirvana-like goal of a single storage system for the entire data center. The reality is that the data center simply has not had that option until high-capacity, data efficient, easily expandable, flash arrays came to market. Other than potentially archive and backup data, all-flash arrays can meet that demand. The primary concern with any consolidation project, but especially storage, is putting all your workloads in one basket. The key to addressing this concern is providing resiliency within the unit and protection if the data center that houses the unit fails.
To that end the XtremIO 4.0 improves the system’s ability to add more X-Bricks to the storage cluster without disruption and each X-Brick can sustain dual SSD drive failures. X-Bricks can be added without incurring downtime while automatically load balancing the workload to include the new node. Of course online cluster expansion can’t protect from a complete system failure caused by a data center-wide disaster. For this situation XtremIO, as part of the 4.0 release, has integrated its disaster recovery capabilities with EMC’s RecoverPoint.
There are two basic problems with All-Flash DR. First, many companies don’t have an integrated disaster recovery solution. Instead, they leverage a third party solution that the customer has to spend time installing and managing. The second is that for those all-flash vendors that do have integrated replication, they require that the customer replicate to another all-flash array. While the pricing of flash has come down, many customers cannot cost-justify an entire system essentially sitting idle at a DR site.
To address these problems XtremIO 4.0 tapped the EMC portfolio, providing tight integration with RecoverPoint. What makes this integration interesting is that it does NOT require the addition of a write splitter on the host(s). The use of XtremIO’s in-memory snapshots and this tight integration allows for full recovery in the secondary solution with a recovery point objective (RPO) of less than 1 minute. The solution also allows the replication to EMC’s non-flash products or even a primary storage that the XtremIO is replacing. The old primary can become the new DR system.
EMC claims that XtremIO is now on a run rate to be a billion dollar company, if it was still stand-alone. 4.0 advances its capabilities and should accelerate its pace to that goal. While the higher capacity X-Brick is something to be expected, the integration with RecoverPoint is more unexpected and certainly a smart move. These new capabilities make XtremIO 4.0 worth strong consideration for any IT planner looking at an all-flash future.