We are all used to Linux. As a whitebox experiment, with open-source software and commercial off-the-shelf (COTS) hardware, Linux has become a mainstay of the IT industry. Even so, most deployers of Linux buy a brand name computer from the HPs and Dells of the world.
Storage has generally been more proprietary, except for the network-attached storage space, where add-ons to the Linux distributions provide CIFS and NFS support. Storage is a bit more like the mainframe industry of old, with expensive boxes leading the market in the enterprise.
One of the issues that is bearing strongly on the future of storage is the price of a terabyte on a RAID appliance versus a retail SATA drive. The large disparity is reaching a tipping point, at which IT professionals ask what they are paying for.
This instability in the value proposition is getting additional pushes from solid state drives (SSDs) that require fewer but faster arrays, from all-flash products that subordinate the arrays to second tier storage, and from open-source storage software that allows COTS-based storage appliances to be constructed. All of these are pressuring the Big Iron makers, who are becoming more aggressive on pricing, especially in the low end of the enterprise and the mid-range market.
But there's more pressure, and it's not just on storage. The success of cloud computing is creating new system manufacturers, such as Google and Amazon. They design their own hardware and software, using COTS to a great extent, and cherry-picking code to mix with their own developments.
These companies are huge users of gear, but they do what the other vendors like IBM and HP do. They have the hardware built in China. And that brings the next complication to the equation. Those Chinese ODMs have grown to a huge size, piggybacking off the traditional vendors and the new cloud service providers. Now, they are looking to enter the US end-user market with branded products.
So where does that leave the industry? Savvy CIOs are realizing that they can bypass the middleman and do as Amazon does. They can buy direct from the ODMs. Facebook has even placed its designs into the public domain to make this easier to do, and several vendors offer labelled versions of whiteboxes at very aggressive prices.
We are about to see storage hitting the same path, with appliances running OpenStack Swift. Beyond that, Ceph and Red Hat will bring unified storage boxes that will handle all the major storage protocols. These are all well-tested code sets, and deployment into enterprise environments is a reasonable risk. Gartner even had a session at this month's Data Center Conference entitled “Building Your Own Storage: Can You Mimic the Big Cloud Providers?”
Networking is not immune. The hardware complexity of switches has dropped to the level that they too can be “whiteboxed,” and we can expect inexpensive 10-gigabit Ethernet switches to hit the market in 2014. Again, YouTube is offering a design to the open platform community.
In fact, we will see a lot of inexpensive gear in 2014. Some will be less than stellar in quality, but, by the nature of COTS and its standardization -- and the easy availability of reference designs from Intel and others -- we can expect most of it to work well and perform to specification.
The risks in going down the whitebox path are less technical than logistical. Newbie companies won't have a well-organized support operation, for instance. Hotlines might end up in Cantonese, too. The more reputable companies (some of the multi-billion dollar ODMs) will get this sorted pretty quickly, and they will leverage their track records with cloud providers and OEMs, so likely will be good suppliers for the whitebox model.
Old loyalties die hard, but the traditional US system and storage vendors face tougher times as their own suppliers become their competitors. That's why Dell went private, and why it wants to convert to a software and services model. That was the path IBM successfully chose when the mainframe was “Unixed.”
There'll be a whole lot of strategies put together to use whiteboxes, ranging from specific projects to low-tier storage, and departmental systems to cloud rollouts. What fits will be a question of your organization's risk aversion, technical confidence, and cost pressures. The one thing that's certain is that you'll be looking at new platform suppliers next year, and in five years will have found a portfolio of suppliers that you trust.