Growth in data, its transmission and desire for analytics continue to fuel demand for storage. Beyond sheer capacity, we see technologies and architectures rising to meet the challenge of keeping pace with demand through virtualisation. In this piece we evaluate EMC’s vision and portfolio.
EMC’s background lies in making virtualised disk arrays for high-end and midrange workloads. The company has also pursued multiple storage technologies through both internal research and ambitious acquisitions. EMC’s purchases were designed to drive ownership of more IP and to experiment with technologies outside its core portfolio. Examples include content addressed storage, network file virtualization and scale-out NAS Isilon.
A federated future
In May 2010, EMC announced VPLEX, an x86 appliance that oversees virtualisation of block based storage arrays. EMC VPLEX provides data mobility and migration, pooling and mirroring, embracing resources which can be geographically distributed. It addresses challenging issues such as global mapping of data and cache coherency for both EMC and third party arrays. This defines EMC’s vision for storage virtualisation on the layer above the disk arrays, which EMC calls storage federation. VPLEX is thus as a federated platform, with a focus on managing data across local and multiple remote data centers.
Designed for distance
VPLEX uses distributed cache coherency technology for a single common namespace across local and remote distance. It has three categories of cover: Local, Geo/Metro and Global. While Local cover can be established with one cluster, Geo/Metro and Global cover require two or more clusters. The VPLEX appliance presents itself as storage to hosts and vice versa. VPLEX GeoSynchrony software runs on the appliance which drives the main functions and is controlled by a management console on a separate server. This software is dedicated to maintaining the integrity of data federation as well as the movement of data. But it also provides software functions such as mirroring and thin provisioning. Given EMC’s ownership of VMware, it is no surprise that VMmotion is supported by VPLEX.
VPLEX introduces a virtualisation layer above storage and servers and it has been engineered to provide storage management across distributed IT platforms. It also controls storage assets to boost reliability and power rapid disaster recovery (DR) provision. While these are laudable objectives, we see other customer requirements that need response. Customer’s increasing frustration with orchestrating workloads between server hypervisors and storage will only escalate.
In the meantime, the EMC federated approach means that certain storage functions will not be provided in the VPLEX layer which will continue to reside in the underlying disk arrays. That could be a purely philosophical design decision, but it could also appear to be a calculated decision by EMC to protect its own array business.
In earlier analysis we reviewed how DataCore has consistently championed storage hypervisors, examined NetApp’s array based approach, reviewed FalconStor’s multi-pronged storage hypervisor approach, assessed the rich set of features that are in the HDS storage hypervisor, established IBM’s relevance to large customers and evaluated Virsto’s software only approach.
This research led to our earlier analysis on establishing the essentials of the storage hypervisor market. In parallel, we have feedback from customers who are actively considering their options and how to effectively orchestrate workloads across server and storage layers. Increased dependence on Server Virtualisation is forcing more acute storage pain as we painted in our analysis of bottlenecks.
EMC has laid out its storage federation vision, but there are other competing design philosophies to solve essential problems. We believe that as customers explore multi-tenancy and virtualised workloads they are likely to completely reassess their fundamental storage provision. We foresee fundamental change and disruption.
Image credit: Sabinche