DellEMC XtremIO X2 Tech Preview #2 – Native Replication

How to achieve the desired RPO with replication with minimum cost

Replicating data has always been a copying challenge in terms of achieving the required protection for the application. This challenge becomes even more severe when using and AFA with high change rate workloads.

The issue is that when the application change rate is extremely high, the data that needs to be sent over the wire is in many cases above the infrastructure capabilities. It then becomes difficult to achieve the desired protection (RPO) without upgrading the bandwidth and increasing the infrastructure cost.

XtremIO Native Replication is designed for critical application with high utilization and bandwidth requirements.

By leveraging XtremIO inline always on data services, the XtremIO Native Replication is the most efficient replication product found today in the market.

The inline, always-on global deduplication of XtremIO ensures that the data blocks will always be replicated just once. Unlike XtremIO, other storage arrays deduplication implementations are based on a deduplication process that utilizes the source & target arrays resources and therefore has high impact on the application and replication performance. In many cases, this impact causes the administrator to compromise on the deduplication data service. In some of the products the deduplication process stops on high utilization, which causes on-unique data to be replicated. This is tremendously inefficient and highly costly.

XtremIO unique content-aware storage architecture deduplicates the data by design, as every non-unique block is automatically pointed to the block that is already stored on the SSDs. Thus there is no need for an additional process and therefore no additional penalty exists for deduplication.

With XtremIO Native Replication:

1. Only unique changes are replicated – on deduplication only the metadata is copied without copying the data itself.
2. Global deduplication – The deduplication is not only on the replicated volume itself, but also across volumes, thus when replicating, for instance, multiple data-stores even in different replication sessions the deduplicated blocks are replicated only once. The same works for repurposing environments that are protected to the same target cluster.
3. The deduplication is even across clusters – In a Fan-In scenarios the global deduplication improves the efficiency and sometimes even dramatically. For instance, when software is upgraded or hotfixes are implemented on all hosts/clusters, there is no need to replicate the data but just the metadata.
4. Compression – The data is sent compressed on the wire.

If the typical data reduction ratio is 4:1, which reduces the replicated data by 75%, the global deduplication improves the efficiency even higher.

The efficiency of XtremIO Native Replication reduces the utilization on the source and target arrays and on the wire. By reducing the utilization and improving the copy efficiency, with XtremIO Native Replication there is no need to compromise on the desired RPO nor on the cost.

By reducing significantly the capacity that is transferred on the wire, the replication process becomes much faster, which enables XtremIO Native Replication to support the AFA high workloads.

XtremIO Implementation

XtremIO Native Replication is an Asynchronous replication that supports all disaster recovery operations, including “test a copy” at the target host and “failover” to the target host.

It supports an RPO of 30 seconds or more and 1000 of recovery points per volume, with the ability to test and failover to any recovery point that exists in the target site and without the need to copy metadata or roll backwards, it’s just done instantaneously!

The retention of the native replication is policy driven. A policy needs to be set just once and can then be used then for multiple protection sessions.

The Retention Policy defines the protection window and the number of recovery-points that will be kept within the protection window.

There are different failure scenarios that requires different granularity for the recovery points. For instance for disaster-recovery or logical corruption scenarios the granularity should be as small as possible but is required for a short period of time such as 30 to 60 minutes. However, for other failure scenarios we might want to have a larger protection window but with a larger granularity.

The Retention Policy in XtremIO supports these different scenarios and allows to define up to 3 time periods with different granularities.

The retention is managed automatically by XtremIO according to the Retention Policy definition.

Thus by assigning the retention policy to the protection session, it’s easily to get the required protection:

When needed, any of the recovery points created by the protection session, can be then used for Test and Failover operations by a single command.

All in one place

One of the key benefits of XtremIO is the simplicity of the management. By using the protection wizard it’s easy to create both local and remote protections.

A high level summary is displayed at the top of the protection screen with the ability to drill-down to a more detailed information including a visual view that presents easily the configuration and status of the protection session.

The information is comprehensive and includes aggregated information to get a high level view of the protection status of a selected volume/consistency group. This information includes:

• Protection Sessions Status
• RPO compliance

In addition, detailed information, current and overtime, per protection session, is presented:

• Protection Session Status
• Session statistics
• Snapshot-sets (recovery-points)

Below you can see a tech preview demo, showing how our Content Addressable Storage (CAS) based Native Replication will look like:

Summary:

XtremIO is bringing to the market a very unique and efficient Native Replication solution. The Native Replication feature was designed with the same inherent architectural characteristics that define the XtremIO storage array:

• Content Addressable Storage (CAS) architecture as the underlying super-efficient meta data constructs, ensuring that all COPY operations are always performed in-memory
• Always on, always inline data services
• Deduplication as part of each IOs data path rather than a post processing process; this ensures that only unique data blocks are written and replicated to the target site

One of the key factors of the success of XtremIO, was that customers understood the unique architectural, performance and functional benefits of the XtremIO array. They simply got it. In the same manner, XtremIO customers should understand the unique technological advantages of XtremIO’s native replication solution – making this yet another winning feature in the XtremIO feature set.