As company earnings continue to be announced in the 2nd half of 2007 more evidence is becoming available which points to a softening of storage demand, especially in the U.S.A. Opinions range from the credit crunch of rising interest rates, product mix issues and full channel inventories that need to be drawn down. While all of the above reasons may be a factor, I’d like to propose one of my own observations.
The storage industry is approaching an inflection point. Multiple public companies are all speaking about IT consolidation and virtualization. For newly formed companies and businesses their criteria requires designing around scalability, minimum solution costs and most importantly no vendor dependence.
I believe that companies that enjoy margins north of 50 points for storage solutions will be forced to be more competitive with storage solutions that go the way of commodity.
Let’s take as an example: block and file storage. Solutions are provided today via a variety of proprietary embedded methods that range from traditional server/storage configurations to specialized appliances. Storage services for all solutions typically provide support for industry standard protocols (iSCSI, NFS, CIFS, NDMP) and data services (snapshots, replication, RAID, compression, compliance, … etc.). Storage solutions today charge you licensing fees or RTUs (Right To Use) for each protocol and data service you need. This is a healthy revenue stream for certain storage vendors.
RAID performed by intelligent hardware controllers is a solid solution today. However businesses are moving toward solutions based on software RAID because of enhanced protection provided by the software RAID/file system combination. The design of the application is a factor too when multiple hardware failures can be tolerated because data copies are distributed and recoverable. A good example of using cheap, non redundant hardware with software capable of handling multiple hardware failures is the Google File System (GFS).
Another factor at work here is a trend that can almost be defined as a default standard today—that of basing your IT solutions off of a Linux distribution. There are multiple offerings of Linux today. One distribution that is gaining adoption is CentOS. Both public and private companies are deploying CentOS as an alternative to a popular Linux vendor distribution. CentOS is built from publicly available open source SRPMS provided by a prominent North American Enterprise Linux vendor. CentOS conforms fully with the upstream vendors redistribution policies and aims to be 100% binary compatible. (CentOS mainly changes packages to remove upstream vendor branding and artwork). All perfectly legal under the terms and conditions of open source licensing.
How would the competitive landscape change if storage protocol support and data services could be added to a unix distribution? In other words take a proven enterprise unix os that is open sourced and build in additional features specific to storage. In fact why not add other features that are becoming expected standards. A good example here is virtulization. If this could be done it would be an enabler for creating storage product offerings out of non proprietary software using commodity servers and storage. This would change the current landscape especially if cost savings and prevention of vendor lock in is achievable.
There is a growing customer base that is building their IT infrastructure from hybrid storage solutions. They don’t quite fit nicely into the standard file, block or even object industry segments today. In some ways these customers want to slightly tweak the storage, servers and software to create the hybrid. This hybrid also turns out to be their competitive weapon.
Helping customers achieve their own individual hybrid is certainly an opportunity worth pursuing.