In a fairly common sense, hyperscale is the practice of making sure that too much of a good thing doesn’t become a bad thing. Recently, it’s become a bright-and-steady buzz word in the enterprise IT sphere because it’s an important way to deal with the ebb and flow of attention to app infrastructure.

In their reference design for hyperscale datacenters, Intel defines hyperscale as such: “Hyperscale computing is a distributed computing environment in which the volume of data and the demand for certain types of workloads can increase exponentially, yet still be accommodated quickly in a cost-effective manner.”

As with every innovation that walks in the door of the IT and even DevOps domain it’s always about making the best out of infrastructure and that means increasing efficiency, lowering costs, and lowering latency (for both developers and customers.)

Let’s look at a round-up of hyperscale news from the perch that DevOps and system operations in general gives us, looking out from the server racks and keyboards into the industry.

Chef 11 sneaks DevOps into hyperscale

Early this month, Opscode saw that hyperscale would become a big thing in the DevOps camp so Chef has been enhanced to deal with high scalability and therefore web hyperscale operations. The entire core API has been rewritten in Erlang but kept backwards compatible; although the web user interface will remain in Ruby.

On scalability in general, Timothy Prickett Morgan wrote about the current improvements, “Chef 11 has an order of magnitude reduction in memory footprint compared to the Ruby version in Chef 10. Chef 11 can manage up to 10,000 nodes from a single server, which is a factor of four more than the Chef 10 server could handle.”

• Additional improvements to the Chef 11 client tool suite include:
• Knife command line tool now includes knife-essentials.
• Knife ssh returns an exit code based on remote command execution.
• Shef has been renamed “chef-shell”.
• A new tool “chef-apply” is included for running single cookbooks.

Infrastructure remains extremely important for Fusion-io and HP

Fusion-io has announced Fusion ioScale for datacenters to help keep up with the hardware side of hyperscale rollouts when it comes to disk. John Furrier writes that each ioScale device “offers up to 3.2 terabytes of flash storage, with performance tuned to the needs of hyperscale environments.” As we’ve seen in the flash storage market, speeding up access and enabling more bandwidth and less latency allows for swifter “spin up” when traffic spikes or just grows naturally.

Further, HP Moonshot is reaching deeply into hyperscale with their early April roll out and announcements of mass produced, commodity servers with the 1500 series. These smaller servers appear to solidly provide 20% less space usage, 80% less power usage, and thus much less expense for the same power usage. Just like how blades initially changed data center environments, these new Moonshot servers provide a modular environment where infrastructure administration can slot more in quickly—or even remove and replace burnt out systems.

And then DevOps wanders into infrastructure—the domain of software-led-infrastructure

“Software is the gatekeeper to hyperscale,” says AMD’s Corporate Vice President and General Manager, Server Business Suresh Gopalakrishnan. With all this going on about increasing flash storage and smarter, cheaper servers, something still needs to coordinate them and that’s going to be where another vein of DevOps enters into hyperscale.

In an interview with John Furrier and Dave Vellante in theCube, Gopalakrishnan spoke to the rise of interest in the paradigm of software defined “X” (where X is a type of service, in this case infrastructure.) With the advent and rise of HP Moonshot, flash storage, tighter and faster networking, data centers need to not just provide a short cycle developing and deploying applications; but enterprise also needs to look at how operations work internally for provisioning and providing the infrastructure for those apps.

On the market now, companies such as Riverbed are delivering SLI concepts for Software-Defined-Networking thoughtfully to businesses seeking to take advantage of this. As well as CA Technologies is working in this space to bring their own clients not just DevOps, but abstract it away from the infrastructure.