Analysis Remember Dot Hill, LSI/Engenio, and Xyratex when they were all independent? They made storage array hardware for the storage system suppliers such as Dell, EMC, HP, IBM, NetApp, Sun and others. Yes, it was that long ago.
The business was hard because storage system suppliers could use contract manufacturers to make their own gear. Moreover, they preferred to sell arrays that weren't too different, hardware-wise, from their competitors.
LSI got out of it by selling off Engenio to NetApp. Xyratex attracted the unwelcome attentions of an activist investor and was sold to Seagate – of all firms.
Dot Hill lingers on doggedly, making around $200m a year in revenues, but suffered losses from 2009 to 2012 of $13 to $22m. It scraped a profit of $5m in 2013 and, hopefully, will make more in 2014.
NetApp is building E-Series arrays using Engenio technology, including its SANtricity OS, and selling these into Big Data/HPC-style markets which prioritise fast data access over Data ONTAP storage software richness and expense. That purchase demonstrated a storage system provider moving into an adjacent market.
Seagate's purchase of Xyratex was followed by it buying the LSI storage controller chip business from Avago in June 2014.
What Seagate is doing is vertically integrating upwards, adding a storage array hardware business layered on top of its disk drive and nascent (exaggerating a little) SSD and PCIe flash business. It's moving from components to systems. Why?
First off, it has been in storage systems for a while, as has fellow disk maker WD. Both make low-end filers: 2-, 4- and 8-bay NAS systems with basic software for small businesses. Such systems are sold cheap and don't compete with Seagate and WD's main system customers for disk drives – the mainstream enterprise storage system suppliers such as Dell, EMC, HDS, HP, IBM, NetApp, and Oracle/Sun (DEHHINO).
Secondly, Xyratex was moving away from selling storage array components to DEHHINO customers to selling complete ClusterStor high-capacity, high-performance arrays to the niche but growing HPC market, and a new set of OEMs such as Cray.
The storage array technology used here was being adopted for Big Data storage by enterprises who needed to analyse billions of unstructured data items to assess how their business was doing and improve its efficiency and focus. Data ingress/egress speed, cost/TB and power use were crucial parameters.
Thirdly, hyperscale storage needs existed in the monster cloud-based service companies such as Amazon, Facebook, Google, eBay, and others. They were into Big Data in effect before the term was even invented, and they each had their individual needs, but also a commonality of needing hundreds of thousands of disk drives a year connected into systems with architectures that were unique to them - but not satisfied by VMAX, ONTAP or DS8000 arrays. No sir, not at all.
They wanted commodity disk drives which they hooked up into systems themselves. They saw no reason to buy commodity drives from DEHHINO suppliers at their markups and started buying raw drives from component suppliers instead.
What Seagate saw, when it bought Xyratex, was that the confluence of Big Data with HPC-style storage could provide growth and higher margins if it sold raw drives to hyperscalers and Big Data arrays to business. And this would compensate for the loss of the premium performance disk drive business to SSDs and all-flash arrays. This strategy became detectable last year.