This article is more than 1 year old

EMC: Kerr-ching! $430m XtremIO gulp's paying off... Hello, $1bn a year

Record-breaker will get software upgrade soon

EMC president Chad Sakac says the XtremIO will be the fastest product in EMC’s history to reach a billion dollar run-rate. A new release is coming to keep the momentum up.

XtremIO was bought by EMC for $430m in May 2012, at which point it did not have a shipping product. The first product was announced a year later, May, 2013.

It is a scale-out design using InfiniBand clustered X-Brick nodes with 10 to 20TB of raw capacity. Deduplication and compression provide greater effective capacity.

EMC quarterly results come out on Wednesday and we can expect XtremIO all-flash array revenues to be at or approaching $250m. The reason is that all-flash array’s (AFA) price/performance means most transaction-based data is best stored on them and not on disk drive arrays like VMAX and VNX. The move is unstoppable and networked disk drive arrays are becoming the home for other data – the bulk stuff.

XtremIO front panel

Sakac, who is EMC’s SVP for global systems engineering, has a Virtual Geek blog that explains his thinking in more detail than the highlights listed here:

  • AFAs are the way to service transactional workloads.
  • AFAs are not right … for workloads that require mainframe support.
  • AFAs are not right… for workloads that require "classic enterprise array" data services like huge at scale replication … There are no AFAs that support this (yet).
  • AFAs are not right …for the "oceans of object storage" or for traditional (vs. new transactional and in-memory) batch HDFS use cases.

Sakac thinks that AFA use-cases will expand as flash costs come down and data management services available on them grow and mature.

Version 4 software for XtremIO will be a non-disruptive upgrade and willadd;

  • Non-disruptive X-brick node addition to XtremIO cluster making it simpler to add performance and capacity,
  • Native RecoverPoint for replication “with rich consistency groups, fan in/out, broad RPOs, rich data replication, and WAN optimization to XtremIO.”

Currently VPLEX is needed to provide replication for XtremIO arrays. That limitation goes away with v4.0. RecoverPoint on XtremIO integrates with the array’s snapshot engine and “RecoverPoint itself will soon have two modes of operating - continuous/journaled and snap/ship.”

Sakac said EMC is working on adding some Quality-of-Service models that would enable XtremIO products to compete better with Solidfire generally and in the service provider market particularly. El Reg reckons that we might see an all-flash EVO:RACK system from EMC to compete with the greater scalability of its SF-series arrays. The SF9010 grows to a 7.5 million IOPS, 100-node beast compared to XtremIO’s 1.5 million IOPS, 6-node cluster. EMC’s website does not provide details of XtremIO clusters with more than six nodes.

Sakac says he thinks more about Solidfire than other AFA competitors.

EMC needs to use its data management services strength as part of its competitive response to Pure Storage, the leading independent (and startup) all-flash array supplier. That company has its FlashRecover snapshot-based replication product. Solidfire also has a replication capability.

Sakac promises much more news about V4.0 XtremIO software shortly. ®

More about

More about

More about


Send us news

Other stories you might like